Merge pull request #21662 from michaelnebel/csharp/rc3.21/updateintegrationtests

C#: Update integration tests to use SDK 10.0.201.

.bazelrc

@@ -11,6 +11,8 @@ build --compilation_mode opt
 common --override_module=semmle_code=%workspace%/misc/bazel/semmle_code_stub
 
 build --repo_env=CC=clang --repo_env=CXX=clang++
+# Disable Android SDK auto-detection (we don't use it, and rules_android has Bazel 9 compatibility issues)
+build --repo_env=ANDROID_HOME=
 
 # print test output, like sembuild does.
 # Set to `errors` if this is too verbose.
@@ -34,7 +36,7 @@ common --@rules_dotnet//dotnet/settings:strict_deps=false
 common --@rules_rust//rust/toolchain/channel=nightly
 
 # Reduce this eventually to empty, once we've fixed all our usages of java, and https://github.com/bazel-contrib/rules_go/issues/4193 is fixed
-common --incompatible_autoload_externally="+@rules_java,+@rules_shell"
+common --incompatible_autoload_externally="+@rules_cc,+@rules_java,+@rules_shell"
 
 build --java_language_version=17
 build --tool_java_language_version=17

.bazelversion

@@ -1 +1 @@
-8.4.2
+9.0.0

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

@@ -27,6 +27,7 @@ jobs:
         uses: github/codeql-action/init@main
         with:
           languages: javascript # does not matter
+          tools: nightly
       - uses: ./.github/actions/os-version
         id: os_version
       ### Build the extractor ###

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

@@ -30,6 +30,7 @@ jobs:
         uses: github/codeql-action/init@main
         with:
           languages: javascript # does not matter
+          tools: nightly
       - uses: ./.github/actions/os-version
         id: os_version
       - uses: actions/cache@v3
@@ -75,6 +76,7 @@ jobs:
         uses: github/codeql-action/init@main
         with:
           languages: javascript # does not matter
+          tools: nightly
       - uses: ./.github/actions/os-version
         id: os_version
       - uses: actions/cache@v3

MODULE.bazel

@@ -15,20 +15,22 @@ local_path_override(
 # see https://registry.bazel.build/ for a list of available packages
 
 bazel_dep(name = "platforms", version = "1.0.0")
-bazel_dep(name = "rules_go", version = "0.56.1")
+bazel_dep(name = "rules_cc", version = "0.2.16")
+bazel_dep(name = "rules_go", version = "0.59.0")
+bazel_dep(name = "rules_java", version = "9.0.3")
 bazel_dep(name = "rules_pkg", version = "1.0.1")
-bazel_dep(name = "rules_nodejs", version = "6.2.0-codeql.1")
+bazel_dep(name = "rules_nodejs", version = "6.7.3")
 bazel_dep(name = "rules_python", version = "0.40.0")
 bazel_dep(name = "rules_shell", version = "0.5.0")
 bazel_dep(name = "bazel_skylib", version = "1.8.1")
 bazel_dep(name = "abseil-cpp", version = "20240116.1", repo_name = "absl")
 bazel_dep(name = "nlohmann_json", version = "3.11.3", repo_name = "json")
 bazel_dep(name = "fmt", version = "12.1.0-codeql.1")
-bazel_dep(name = "rules_kotlin", version = "2.1.3-codeql.1")
-bazel_dep(name = "gazelle", version = "0.40.0")
+bazel_dep(name = "rules_kotlin", version = "2.2.2-codeql.1")
+bazel_dep(name = "gazelle", version = "0.47.0")
 bazel_dep(name = "rules_dotnet", version = "0.21.5-codeql.1")
 bazel_dep(name = "googletest", version = "1.14.0.bcr.1")
-bazel_dep(name = "rules_rust", version = "0.66.0")
+bazel_dep(name = "rules_rust", version = "0.68.1.codeql.1")
 bazel_dep(name = "zstd", version = "1.5.5.bcr.1")
 
 bazel_dep(name = "buildifier_prebuilt", version = "6.4.0", dev_dependency = True)
@@ -41,7 +43,7 @@ RUST_EDITION = "2024"
 # a nightly toolchain is required to enable experimental_use_cc_common_link, which we require internally
 # we prefer to run the same version as internally, even if experimental_use_cc_common_link is not really
 # required in this repo
-RUST_VERSION = "nightly/2025-08-01"
+RUST_VERSION = "nightly/2026-01-22"
 
 rust = use_extension("@rules_rust//rust:extensions.bzl", "rust")
 rust.toolchain(
@@ -53,26 +55,26 @@ rust.toolchain(
     ],
     # generated by buildutils-internal/scripts/fill-rust-sha256s.py (internal repo)
     sha256s = {
-        "2025-08-01/rustc-nightly-x86_64-unknown-linux-gnu.tar.xz": "9bbeaf5d3fc7247d31463a9083aa251c995cc50662c8219e7a2254d76a72a9a4",
-        "2025-08-01/rustc-nightly-x86_64-apple-darwin.tar.xz": "c9ea539a8eff0d5d162701f99f9e1aabe14dd0dfb420d62362817a5d09219de7",
-        "2025-08-01/rustc-nightly-aarch64-apple-darwin.tar.xz": "ae83feebbc39cfd982e4ecc8297731fe79c185173aee138467b334c5404b3773",
-        "2025-08-01/rustc-nightly-x86_64-pc-windows-msvc.tar.xz": "9f170c30d802a349be60cf52ec46260802093cb1013ad667fc0d528b7b10152f",
-        "2025-08-01/clippy-nightly-x86_64-unknown-linux-gnu.tar.xz": "9ae5f3cd8f557c4f6df522597c69d14398cf604cfaed2b83e767c4b77a7eaaf6",
-        "2025-08-01/clippy-nightly-x86_64-apple-darwin.tar.xz": "983cb9ee0b6b968188e04ab2d33743d54764b2681ce565e1b3f2b9135c696a3e",
-        "2025-08-01/clippy-nightly-aarch64-apple-darwin.tar.xz": "ed2219dbc49d088225e1b7c5c4390fa295066e071fddaa2714018f6bb39ddbf0",
-        "2025-08-01/clippy-nightly-x86_64-pc-windows-msvc.tar.xz": "911f40ab5cbdd686f40e00965271fe47c4805513a308ed01f30eafb25b448a50",
-        "2025-08-01/cargo-nightly-x86_64-unknown-linux-gnu.tar.xz": "106463c284e48e4904c717471eeec2be5cc83a9d2cae8d6e948b52438cad2e69",
-        "2025-08-01/cargo-nightly-x86_64-apple-darwin.tar.xz": "6ad35c40efc41a8c531ea43235058347b6902d98a9693bf0aed7fc16d5590cef",
-        "2025-08-01/cargo-nightly-aarch64-apple-darwin.tar.xz": "dd28c365e9d298abc3154c797720ad36a0058f131265c9978b4c8e4e37012c8a",
-        "2025-08-01/cargo-nightly-x86_64-pc-windows-msvc.tar.xz": "7b431286e12d6b3834b038f078389a00cac73f351e8c3152b2504a3c06420b3b",
-        "2025-08-01/llvm-tools-nightly-x86_64-unknown-linux-gnu.tar.xz": "e342e305d7927cc288d386983b2bc253cfad3776b113386e903d0b302648ef47",
-        "2025-08-01/llvm-tools-nightly-x86_64-apple-darwin.tar.xz": "e44dd3506524d85c37b3a54bcc91d01378fd2c590b2db5c5974d12f05c1b84d1",
-        "2025-08-01/llvm-tools-nightly-aarch64-apple-darwin.tar.xz": "0c1b5f46dd81be4a9227b10283a0fcaa39c14fea7e81aea6fd6d9887ff6cdc41",
-        "2025-08-01/llvm-tools-nightly-x86_64-pc-windows-msvc.tar.xz": "423e5fd11406adccbc31b8456ceb7375ce055cdf45e90d2c3babeb2d7f58383f",
-        "2025-08-01/rust-std-nightly-x86_64-unknown-linux-gnu.tar.xz": "3c0ceb46a252647a1d4c7116d9ccae684fa5e42aaf3296419febd2c962c3b41d",
-        "2025-08-01/rust-std-nightly-x86_64-apple-darwin.tar.xz": "3be416003cab10f767390a753d1d16ae4d26c7421c03c98992cf1943e5b0efe8",
-        "2025-08-01/rust-std-nightly-aarch64-apple-darwin.tar.xz": "4046ac0ef951cb056b5028a399124f60999fa37792eab69d008d8d7965f389b4",
-        "2025-08-01/rust-std-nightly-x86_64-pc-windows-msvc.tar.xz": "191ed9d8603c3a4fe5a7bbbc2feb72049078dae2df3d3b7d5dedf3abbf823e6e",
+        "2026-01-22/rustc-nightly-x86_64-unknown-linux-gnu.tar.xz": "88db619323cc1321630d124efa51ed02fabc5e020f08cfa0eda2c0ac1afbe69a",
+        "2026-01-22/rustc-nightly-x86_64-apple-darwin.tar.xz": "08484da3fa38db56f93629aeabdc0ae9ff8ed9704c0792d35259cbc849b3f54c",
+        "2026-01-22/rustc-nightly-aarch64-apple-darwin.tar.xz": "a39c0b21b7058e364ea1bd43144e42e4bf1efade036b2e82455f2afce194ee81",
+        "2026-01-22/rustc-nightly-x86_64-pc-windows-msvc.tar.xz": "d00248ee9850dbb6932b2578e32ff74fc7c429854c1aa071066ca31b65385a3b",
+        "2026-01-22/clippy-nightly-x86_64-unknown-linux-gnu.tar.xz": "70656a0ce994ffff16d5a35a7b170a0acd41e9bb54a589c96ed45bf97b094a4d",
+        "2026-01-22/clippy-nightly-x86_64-apple-darwin.tar.xz": "fe242519fa961522734733009705aec3c2d9a20cc57291f2aa614e5e6262c88f",
+        "2026-01-22/clippy-nightly-aarch64-apple-darwin.tar.xz": "38bb226363ec97c9722edf966cd58774a683e19fd2ff2a6030094445d51e06f9",
+        "2026-01-22/clippy-nightly-x86_64-pc-windows-msvc.tar.xz": "6da9b4470beea67abfebf046f141eee0d2a8db7c7a9e4e2294478734fd477228",
+        "2026-01-22/cargo-nightly-x86_64-unknown-linux-gnu.tar.xz": "99004e9d10c43a01499642f53bb3184d41137a95d65bfb217098840a9e79e892",
+        "2026-01-22/cargo-nightly-x86_64-apple-darwin.tar.xz": "6e021394cf8d8400ac6cfdfcef24e4d74f988e91eb8028b36de3a64ce3502990",
+        "2026-01-22/cargo-nightly-aarch64-apple-darwin.tar.xz": "4b2494cb69ab64132cddbc411a38ea9f1105e54d6f986e43168d54f79510c673",
+        "2026-01-22/cargo-nightly-x86_64-pc-windows-msvc.tar.xz": "c36613cf57407212d10d37b76e49a60ff42336e953cdff9e177283f530a83fc1",
+        "2026-01-22/llvm-tools-nightly-x86_64-unknown-linux-gnu.tar.xz": "0b123c5027dbd833aae6845ffe9bd07d309bf798746a7176aadaea68fbcbd05d",
+        "2026-01-22/llvm-tools-nightly-x86_64-apple-darwin.tar.xz": "a47864491ad5619158c950ab7570fb6e487d5117338585c27334d45824b406d8",
+        "2026-01-22/llvm-tools-nightly-aarch64-apple-darwin.tar.xz": "db9bc826d6e2e7e914505d50157682e516ceb90357e83d77abddc32c2d962f41",
+        "2026-01-22/llvm-tools-nightly-x86_64-pc-windows-msvc.tar.xz": "ffaa406932b2fe62e01dad61cf4ed34860a5d2a6f9306ca340d79e630d930039",
+        "2026-01-22/rust-std-nightly-x86_64-unknown-linux-gnu.tar.xz": "e9c0d5e06e18a4b509391b3088f29293e310cdc8ccc865be8fa3f09733326925",
+        "2026-01-22/rust-std-nightly-x86_64-apple-darwin.tar.xz": "25d75995cee679a4828ca9fe48c5a31a67c3b0846018440ef912e5a6208f53f6",
+        "2026-01-22/rust-std-nightly-aarch64-apple-darwin.tar.xz": "e4132bf3f2eed4684c86756a02315bcf481c23e675e3e25630fc604c9cb4594c",
+        "2026-01-22/rust-std-nightly-x86_64-pc-windows-msvc.tar.xz": "961bb535ef95ae8a5fa4e224cb94aff190f155c45a9bcf7a53e184b024aa41b1",
     },
     versions = [RUST_VERSION],
 )
@@ -188,6 +190,15 @@ pip.parse(
 )
 use_repo(pip, "codegen_deps")
 
+python = use_extension("@rules_python//python/extensions:python.bzl", "python")
+python.toolchain(
+    is_default = True,
+    python_version = "3.12",
+)
+use_repo(python, "python_3_12", "python_versions")
+
+register_toolchains("@python_versions//3.12:all")
+
 swift_deps = use_extension("//swift/third_party:load.bzl", "swift_deps")
 
 # following list can be kept in sync with `bazel mod tidy`
@@ -221,10 +232,6 @@ use_repo(
     kotlin_extractor_deps,
     "codeql_kotlin_defaults",
     "codeql_kotlin_embeddable",
-    "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",
@@ -234,10 +241,7 @@ use_repo(
     "kotlin-compiler-2.1.20-Beta1",
     "kotlin-compiler-2.2.0-Beta1",
     "kotlin-compiler-2.2.20-Beta2",
-    "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-2.3.0",
     "kotlin-compiler-embeddable-1.8.0",
     "kotlin-compiler-embeddable-1.9.0-Beta",
     "kotlin-compiler-embeddable-1.9.20-Beta",
@@ -247,10 +251,7 @@ use_repo(
     "kotlin-compiler-embeddable-2.1.20-Beta1",
     "kotlin-compiler-embeddable-2.2.0-Beta1",
     "kotlin-compiler-embeddable-2.2.20-Beta2",
-    "kotlin-stdlib-1.6.0",
-    "kotlin-stdlib-1.6.20",
-    "kotlin-stdlib-1.7.0",
-    "kotlin-stdlib-1.7.20",
+    "kotlin-compiler-embeddable-2.3.0",
     "kotlin-stdlib-1.8.0",
     "kotlin-stdlib-1.9.0-Beta",
     "kotlin-stdlib-1.9.20-Beta",
@@ -260,14 +261,15 @@ use_repo(
     "kotlin-stdlib-2.1.20-Beta1",
     "kotlin-stdlib-2.2.0-Beta1",
     "kotlin-stdlib-2.2.20-Beta2",
+    "kotlin-stdlib-2.3.0",
 )
 
 go_sdk = use_extension("@rules_go//go:extensions.bzl", "go_sdk")
-go_sdk.download(version = "1.25.0")
+go_sdk.download(version = "1.26.0")
 
 go_deps = use_extension("@gazelle//:extensions.bzl", "go_deps")
 go_deps.from_file(go_mod = "//go/extractor:go.mod")
-use_repo(go_deps, "org_golang_x_mod", "org_golang_x_tools")
+use_repo(go_deps, "com_github_stretchr_testify", "org_golang_x_mod", "org_golang_x_tools")
 
 ripunzip_archive = use_repo_rule("//misc/ripunzip:ripunzip.bzl", "ripunzip_archive")
 

actions/ql/lib/CHANGELOG.md

@@ -1,3 +1,17 @@
+## 0.4.29
+
+No user-facing changes.
+
+## 0.4.28
+
+No user-facing changes.
+
+## 0.4.27
+
+### Bug Fixes
+
+* Fixed a crash when analysing a `${{ ... }}` expression over around 300 characters in length.
+
 ## 0.4.26
 
 ### Major Analysis Improvements

actions/ql/lib/change-notes/released/0.4.27.md

@@ -0,0 +1,5 @@
+## 0.4.27
+
+### Bug Fixes
+
+* Fixed a crash when analysing a `${{ ... }}` expression over around 300 characters in length.

actions/ql/lib/change-notes/released/0.4.28.md

@@ -0,0 +1,3 @@
+## 0.4.28
+
+No user-facing changes.

actions/ql/lib/change-notes/released/0.4.29.md

@@ -0,0 +1,3 @@
+## 0.4.29
+
+No user-facing changes.

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

@@ -1,2 +1,2 @@
 ---
-lastReleaseVersion: 0.4.26
+lastReleaseVersion: 0.4.29

actions/ql/lib/codeql/actions/ast/internal/Ast.qll

@@ -27,8 +27,8 @@ string getADelimitedExpression(YamlString s, int offset) {
   // not just the last (greedy match) or first (reluctant match).
   result =
     s.getValue()
-        .regexpFind("\\$\\{\\{(?:[^}]|}(?!}))*\\}\\}", _, offset)
-        .regexpCapture("(\\$\\{\\{(?:[^}]|}(?!}))*\\}\\})", 1)
+        .regexpFind("\\$\\{\\{(?:[^}]|}(?!}))*+\\}\\}", _, offset)
+        .regexpCapture("(\\$\\{\\{(?:[^}]|}(?!}))*+\\}\\})", 1)
         .trim()
 }
 

actions/ql/lib/qlpack.yml

@@ -1,5 +1,5 @@
 name: codeql/actions-all
-version: 0.4.26
+version: 0.4.30-dev
 library: true
 warnOnImplicitThis: true
 dependencies:

actions/ql/src/CHANGELOG.md

@@ -1,3 +1,15 @@
+## 0.6.21
+
+No user-facing changes.
+
+## 0.6.20
+
+No user-facing changes.
+
+## 0.6.19
+
+No user-facing changes.
+
 ## 0.6.18
 
 No user-facing changes.

actions/ql/src/change-notes/released/0.6.19.md

@@ -0,0 +1,3 @@
+## 0.6.19
+
+No user-facing changes.

actions/ql/src/change-notes/released/0.6.20.md

@@ -0,0 +1,3 @@
+## 0.6.20
+
+No user-facing changes.

actions/ql/src/change-notes/released/0.6.21.md

@@ -0,0 +1,3 @@
+## 0.6.21
+
+No user-facing changes.

actions/ql/src/codeql-pack.release.yml

@@ -1,2 +1,2 @@
 ---
-lastReleaseVersion: 0.6.18
+lastReleaseVersion: 0.6.21

actions/ql/src/qlpack.yml

@@ -1,5 +1,5 @@
 name: codeql/actions-queries
-version: 0.6.18
+version: 0.6.22-dev
 library: false
 warnOnImplicitThis: true
 groups: [actions, queries]

actions/ql/test/library-tests/very-long-expression/very_long_expression_node.expected

@@ -0,0 +1 @@
+| 97418 |

actions/ql/test/library-tests/very-long-expression/very_long_expression_node.ql

@@ -0,0 +1,5 @@
+import codeql.actions.ast.internal.Ast
+
+int getAnExpressionLength() { result = any(ExpressionImpl e).toString().length() }
+
+select max(getAnExpressionLength())

cpp/downgrades/7e7c2f55670f8123d514cf542ccb1938118ac561/upgrade.properties

@@ -0,0 +1,5 @@
+description: Add trap_filename, source_file_uses_trap and in_trap relations
+compatibility: full
+trap_filename.rel: delete
+source_file_uses_trap.rel: delete
+in_trap.rel: delete

cpp/downgrades/83100310bf73eefc37c1d8d0ac98b2ca3019c7b6/upgrade.properties

@@ -0,0 +1,2 @@
+description: Sections for databaseMetadata and overlayChangedFiles
+compatibility: full

cpp/downgrades/9439176c1d1312787926458dd54d65a849069118/preprocdirects.ql

@@ -0,0 +1,13 @@
+class PreprocessorDirective extends @preprocdirect {
+  string toString() { none() }
+}
+
+class Location extends @location_default {
+  string toString() { none() }
+}
+
+from PreprocessorDirective ppd, int kind, int kind_new, Location l
+where
+  preprocdirects(ppd, kind, l) and
+  if kind = 17 then kind_new = /* ppd_warning */ 18 else kind_new = kind
+select ppd, kind_new, l

cpp/downgrades/9439176c1d1312787926458dd54d65a849069118/upgrade.properties

@@ -0,0 +1,4 @@
+description: Support embed preprocessor directive
+compatibility: partial
+embeds.rel: delete
+preprocdirects.rel: run preprocdirects.qlo

cpp/ql/lib/CHANGELOG.md

@@ -1,3 +1,42 @@
+## 8.0.0
+
+### Breaking Changes
+
+* CodeQL version 2.24.2 accidentally introduced a syntactical breaking change to `BarrierGuard<...>::getAnIndirectBarrierNode` and `InstructionBarrierGuard<...>::getAnIndirectBarrierNode`. These breaking changes have now been reverted so that the original code compiles again.
+* `MustFlow`, the inter-procedural must-flow data flow analysis library, has been re-worked to use parameterized modules. Like in the case of data flow and taint tracking, instead of extending the `MustFlowConfiguration` class, the user should now implement a module with the `MustFlow::ConfigSig` signature, and instantiate the `MustFlow::Global` parameterized module with the implemented module.
+
+### Minor Analysis Improvements
+
+* Refactored the "Year field changed using an arithmetic operation without checking for leap year" query (`cpp/leap-year/unchecked-after-arithmetic-year-modification`) to address large numbers of false positive results.
+
+### Bug Fixes
+
+* The `allowInterproceduralFlow` predicate of must-flow data flow configurations now correctly handles direct recursion.
+
+## 7.1.1
+
+### Minor Analysis Improvements
+
+* Added remote flow source models for the `winhttp.h` windows header and the Azure SDK core library for C/C++.
+
+## 7.1.0
+
+### New Features
+
+* Added a subclass `Embed` of `PreprocessorDirective` for C23 and C++26 `#embed` preprocessor directives.
+* Added modules `DataFlow::ParameterizedBarrierGuard` and `DataFlow::ParameterizedInstructionBarrierGuard`. These modules provide the same features as `DataFlow::BarrierGuard` and `DataFlow::InstructionBarrierGuard`, but allow for an additional parameter to support properly using them in dataflow configurations that uses flow states.
+
+### Minor Analysis Improvements
+
+* The `Buffer.qll` library will no longer report incorrect buffer sizes on certain malformed databases. As a result, the queries `cpp/static-buffer-overflow`, `cpp/overflow-buffer`, `cpp/badly-bounded-write`, `cpp/overrunning-write`, `cpp/overrunning-write-with-float`, and `cpp/very-likely-overrunning-write` will report fewer false positives on such databases.
+* Added `taint` summary models and `sql-injection` barrier models for the MySQL `mysql_real_escape_string` and `mysql_real_escape_string_quote` escaping functions.
+* The predicate `SummarizedCallable.propagatesFlow` has been extended with the columns `Provenance p` and `boolean isExact`, and as a consequence the predicates `SummarizedCallable.hasProvenance` and `SummarizedCallable.hasExactModel` have been removed.
+
+### Bug Fixes
+
+* Fixed a bug in the `GuardCondition` library which sometimes prevented binary logical operators from being recognized as guard conditions. As a result, queries using `GuardCondition` may see improved results.
+* Fixed a bug which caused `Node.asDefinition()` to not have a result for certain assignments.
+
 ## 7.0.0
 
 ### Breaking Changes

cpp/ql/lib/change-notes/released/7.1.0.md

@@ -0,0 +1,17 @@
+## 7.1.0
+
+### New Features
+
+* Added a subclass `Embed` of `PreprocessorDirective` for C23 and C++26 `#embed` preprocessor directives.
+* Added modules `DataFlow::ParameterizedBarrierGuard` and `DataFlow::ParameterizedInstructionBarrierGuard`. These modules provide the same features as `DataFlow::BarrierGuard` and `DataFlow::InstructionBarrierGuard`, but allow for an additional parameter to support properly using them in dataflow configurations that uses flow states.
+
+### Minor Analysis Improvements
+
+* The `Buffer.qll` library will no longer report incorrect buffer sizes on certain malformed databases. As a result, the queries `cpp/static-buffer-overflow`, `cpp/overflow-buffer`, `cpp/badly-bounded-write`, `cpp/overrunning-write`, `cpp/overrunning-write-with-float`, and `cpp/very-likely-overrunning-write` will report fewer false positives on such databases.
+* Added `taint` summary models and `sql-injection` barrier models for the MySQL `mysql_real_escape_string` and `mysql_real_escape_string_quote` escaping functions.
+* The predicate `SummarizedCallable.propagatesFlow` has been extended with the columns `Provenance p` and `boolean isExact`, and as a consequence the predicates `SummarizedCallable.hasProvenance` and `SummarizedCallable.hasExactModel` have been removed.
+
+### Bug Fixes
+
+* Fixed a bug in the `GuardCondition` library which sometimes prevented binary logical operators from being recognized as guard conditions. As a result, queries using `GuardCondition` may see improved results.
+* Fixed a bug which caused `Node.asDefinition()` to not have a result for certain assignments.

cpp/ql/lib/change-notes/released/7.1.1.md

@@ -0,0 +1,5 @@
+## 7.1.1
+
+### Minor Analysis Improvements
+
+* Added remote flow source models for the `winhttp.h` windows header and the Azure SDK core library for C/C++.

cpp/ql/lib/change-notes/released/8.0.0.md

@@ -0,0 +1,14 @@
+## 8.0.0
+
+### Breaking Changes
+
+* CodeQL version 2.24.2 accidentally introduced a syntactical breaking change to `BarrierGuard<...>::getAnIndirectBarrierNode` and `InstructionBarrierGuard<...>::getAnIndirectBarrierNode`. These breaking changes have now been reverted so that the original code compiles again.
+* `MustFlow`, the inter-procedural must-flow data flow analysis library, has been re-worked to use parameterized modules. Like in the case of data flow and taint tracking, instead of extending the `MustFlowConfiguration` class, the user should now implement a module with the `MustFlow::ConfigSig` signature, and instantiate the `MustFlow::Global` parameterized module with the implemented module.
+
+### Minor Analysis Improvements
+
+* Refactored the "Year field changed using an arithmetic operation without checking for leap year" query (`cpp/leap-year/unchecked-after-arithmetic-year-modification`) to address large numbers of false positive results.
+
+### Bug Fixes
+
+* The `allowInterproceduralFlow` predicate of must-flow data flow configurations now correctly handles direct recursion.

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

@@ -1,2 +1,2 @@
 ---
-lastReleaseVersion: 7.0.0
+lastReleaseVersion: 8.0.0

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

@@ -0,0 +1,14 @@
+# partial model of the MySQL api
+extensions:
+  - addsTo:
+      pack: codeql/cpp-all
+      extensible: summaryModel
+    data: # namespace, type, subtypes, name, signature, ext, input, output, kind, provenance
+      - ["", "", False, "mysql_real_escape_string", "", "", "Argument[*2]", "Argument[*1]", "taint", "manual"]
+      - ["", "", False, "mysql_real_escape_string_quote", "", "", "Argument[*2]", "Argument[*1]", "taint", "manual"]
+  - addsTo:
+      pack: codeql/cpp-all
+      extensible: barrierModel
+    data: # namespace, type, subtypes, name, signature, ext, output, kind, provenance
+      - ["", "", False, "mysql_real_escape_string", "", "", "Argument[*1]", "sql-injection", "manual"]
+      - ["", "", False, "mysql_real_escape_string_quote", "", "", "Argument[*1]", "sql-injection", "manual"]

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

@@ -24,6 +24,13 @@ extensions:
       - ["", "", False, "MapViewOfFileNuma2", "", "", "ReturnValue[*]", "local", "manual"]
       # ntifs.h
       - ["", "", False, "NtReadFile", "", "", "Argument[*5]", "local", "manual"]
+      # winhttp.h
+      - ["", "", False, "WinHttpReadData", "", "", "Argument[*1]", "remote", "manual"]
+      - ["", "", False, "WinHttpReadDataEx", "", "", "Argument[*1]", "remote", "manual"]
+      - ["", "", False, "WinHttpQueryHeaders", "", "", "Argument[*3]", "remote", "manual"]
+      - ["", "", False, "WinHttpQueryHeadersEx", "", "", "Argument[*5]", "remote", "manual"]
+      - ["", "", False, "WinHttpQueryHeadersEx", "", "", "Argument[*6]", "remote", "manual"]
+      - ["", "", False, "WinHttpQueryHeadersEx", "", "", "Argument[**8]", "remote", "manual"]
   - addsTo:
       pack: codeql/cpp-all
       extensible: summaryModel
@@ -46,4 +53,6 @@ extensions:
       - ["", "", False, "RtlMoveMemory", "", "", "Argument[*@1]", "Argument[*@0]", "value", "manual"]
       - ["", "", False, "RtlMoveVolatileMemory", "", "", "Argument[*@1]", "Argument[*@0]", "value", "manual"]
       # winternl.h
-      - ["", "", False, "RtlInitUnicodeString", "", "", "Argument[*1]", "Argument[*0].Field[*Buffer]", "value", "manual"]
+      - ["", "", False, "RtlInitUnicodeString", "", "", "Argument[*1]", "Argument[*0].Field[*Buffer]", "value", "manual"]
+      # winhttp.h
+      - ["", "", False, "WinHttpCrackUrl", "", "", "Argument[*0]", "Argument[*3]", "taint", "manual"]

cpp/ql/lib/ext/azure.core.model.yml

@@ -0,0 +1,41 @@
+extensions:
+  - addsTo:
+      pack: codeql/cpp-all
+      extensible: sourceModel
+    data: # namespace, type, subtypes, name, signature, ext, output, kind, provenance
+      - ["Azure::Core::Http", "RawResponse", True, "GetHeaders", "", "", "ReturnValue[*]", "remote", "manual"]
+      - ["Azure::Core::Http", "RawResponse", True, "GetBody", "", "", "ReturnValue[*]", "remote", "manual"]
+      - ["Azure::Core::Http", "RawResponse", True, "ExtractBodyStream", "", "", "ReturnValue[*]", "remote", "manual"]
+      - ["Azure::Core::Http", "Request", True, "GetHeaders", "", "", "ReturnValue", "remote", "manual"]
+      - ["Azure::Core::Http", "Request", True, "GetHeader", "", "", "ReturnValue", "remote", "manual"]
+      - ["Azure::Core::Http", "Request", True, "GetBodyStream", "", "", "ReturnValue[*]", "remote", "manual"]
+
+  - addsTo:
+      pack: codeql/cpp-all
+      extensible: summaryModel
+    data: # namespace, type, subtypes, name, signature, ext, input, output, kind, provenance
+      - ["Azure::Core", "Url", True, "Url", "", "", "Argument[*0]", "Argument[-1]", "taint", "manual"]
+      - ["Azure::Core", "Url", True, "SetScheme", "", "", "Argument[*0]", "Argument[-1]", "taint", "manual"]
+      - ["Azure::Core", "Url", True, "SetHost", "", "", "Argument[*0]", "Argument[-1]", "taint", "manual"]
+      - ["Azure::Core", "Url", True, "SetPort", "", "", "Argument[0]", "Argument[-1]", "taint", "manual"]
+      - ["Azure::Core", "Url", True, "SetPath", "", "", "Argument[*0]", "Argument[-1]", "taint", "manual"]
+      - ["Azure::Core", "Url", True, "SetQueryParameters", "", "", "Argument[0]", "Argument[-1]", "taint", "manual"]
+      - ["Azure::Core", "Url", True, "AppendPath", "", "", "Argument[*0]", "Argument[-1]", "taint", "manual"]
+      - ["Azure::Core", "Url", True, "AppendQueryParameter", "", "", "Argument[*1]", "Argument[-1]", "taint", "manual"]
+      - ["Azure::Core", "Url", True, "GetHost", "", "", "Argument[-1]", "ReturnValue[*]", "taint", "manual"]
+      - ["Azure::Core", "Url", True, "GetPath", "", "", "Argument[-1]", "ReturnValue[*]", "taint", "manual"]
+      - ["Azure::Core", "Url", True, "GetPort", "", "", "Argument[-1]", "ReturnValue", "taint", "manual"]
+      - ["Azure::Core", "Url", True, "GetQueryParameters", "", "", "Argument[-1]", "ReturnValue", "taint", "manual"]
+      - ["Azure::Core", "Url", True, "GetScheme", "", "", "Argument[-1]", "ReturnValue[*]", "taint", "manual"]
+      - ["Azure::Core", "Url", True, "GetRelativeUrl", "", "", "Argument[-1]", "ReturnValue", "taint", "manual"]
+      - ["Azure::Core", "Url", True, "GetAbsoluteUrl", "", "", "Argument[-1]", "ReturnValue", "taint", "manual"]
+      - ["Azure::Core", "Url", True, "Decode", "", "", "Argument[*0]", "ReturnValue", "taint", "manual"]
+      - ["Azure::Core", "Url", True, "Encode", "", "", "Argument[*0]", "ReturnValue", "taint", "manual"]
+      - ["Azure::Core::IO", "BodyStream", True, "Read", "", "", "Argument[-1]", "Argument[*0]", "taint", "manual"]
+      - ["Azure::Core::IO", "BodyStream", True, "ReadToCount", "", "", "Argument[-1]", "Argument[*0]", "taint", "manual"]
+      - ["Azure::Core::IO", "BodyStream", True, "ReadToEnd", "", "", "Argument[-1]", "ReturnValue.Element", "taint", "manual"]
+      - ["Azure", "Nullable", True, "Nullable", "", "", "Argument[0]", "Argument[-1]", "taint", "manual"]
+      - ["Azure", "Nullable", True, "operator=", "", "", "Argument[*0]", "Argument[-1]", "value", "manual"]
+      - ["Azure", "Nullable", True, "Value", "", "", "Argument[-1]", "ReturnValue[*]", "taint", "manual"]
+      - ["Azure", "Nullable", True, "operator->", "", "", "Argument[-1]", "ReturnValue[*]", "taint", "manual"]
+      - ["Azure", "Nullable", True, "operator*", "", "", "Argument[-1]", "ReturnValue[*]", "taint", "manual"]

cpp/ql/lib/qlpack.yml

@@ -1,5 +1,5 @@
 name: codeql/cpp-all
-version: 7.0.0
+version: 8.0.1-dev
 groups: cpp
 dbscheme: semmlecode.cpp.dbscheme
 extractor: cpp

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

@@ -192,6 +192,15 @@ class Element extends ElementBase {
    */
   predicate isAffectedByMacro() { affectedByMacro(this) }
 
+  /**
+   * INTERNAL: Do not use.
+   *
+   * Holds if this element is affected by the expansion of `mi`.
+   */
+  predicate isAffectedByMacro(MacroInvocation mi) {
+    affectedbymacroexpansion(underlyingElement(this), unresolveElement(mi))
+  }
+
   private Element getEnclosingElementPref() {
     enclosingfunction(underlyingElement(this), unresolveElement(result)) or
     result.(Function) = stmtEnclosingElement(this) or

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

@@ -239,6 +239,9 @@ class MacroInvocation extends MacroAccess {
     macro_argument_unexpanded(underlyingElement(this), i, result)
   }
 
+  /** Gets the number of arguments for this macro invocation. */
+  int getNumberOfArguments() { result = count(int i | exists(this.getUnexpandedArgument(i)) | i) }
+
   /**
    * Gets the `i`th _expanded_ argument of this macro invocation, where the
    * first argument has `i = 0`. The result has been expanded for macros _and_

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

@@ -328,3 +328,27 @@ class PreprocessorPragma extends PreprocessorDirective, @ppd_pragma {
 class PreprocessorLine extends PreprocessorDirective, @ppd_line {
   override string toString() { result = "#line " + this.getHead() }
 }
+
+/**
+ * A C23 or C++26 `#embed` preprocessor directive. For example, the following code
+ * contains one `Embed` directive:
+ * ```cpp
+ * char arr[] = {
+ * #embed "bin"
+ * };
+ * ```
+ */
+class Embed extends PreprocessorDirective, @ppd_embed {
+  override string toString() { result = "#embed " + this.getIncludeText() }
+
+  /**
+   * Gets the token which occurs after `#embed`, for example `"filename"`
+   * or `<filename>`.
+   */
+  string getIncludeText() { result = this.getHead() }
+
+  /**
+   * Gets the file directly embedded by this `#embed`.
+   */
+  File getEmbeddedFile() { embeds(underlyingElement(this), unresolveElement(result)) }
+}

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

@@ -62,11 +62,13 @@ private Class getRootType(FieldAccess fa) {
  * unspecified type of `v` is a `ReferenceType`.
  */
 private int getVariableSize(Variable v) {
-  exists(Type t |
-    t = v.getUnspecifiedType() and
-    not t instanceof ReferenceType and
-    result = t.getSize()
-  )
+  result =
+    unique(Type t |
+      t = v.getUnspecifiedType() and
+      not t instanceof ReferenceType
+    |
+      t.getSize()
+    )
 }
 
 /**
@@ -79,30 +81,32 @@ private int getSize(VariableAccess va) {
     not v instanceof Field and
     result = getVariableSize(v)
     or
-    exists(Class c, int trueSize |
-      // Otherwise, we find the "outermost" object and compute the size
-      // as the difference between the size of the type of the "outermost
-      // object" and the offset of the field relative to that type.
-      // For example, consider the following structs:
-      // ```
-      // struct S {
-      //   uint32_t x;
-      //   uint32_t y;
-      // };
-      // struct S2 {
-      //   S s;
-      //   uint32_t z;
-      // };
-      // ```
-      // Given an object `S2 s2` the size of the buffer `&s2.s.y`
-      // is the size of the base object type (i.e., `S2`) minutes the offset
-      // of `y` relative to the type `S2` (i.e., `4`). So the size of the
-      // buffer is `12 - 4 = 8`.
-      c = getRootType(va) and
-      // we calculate the size based on the last field, to avoid including any padding after it
-      trueSize = max(Field f | | f.getOffsetInClass(c) + getVariableSize(f)) and
-      result = trueSize - v.(Field).getOffsetInClass(c)
-    )
+    result =
+      unique(Class c, int trueSize |
+        // Otherwise, we find the "outermost" object and compute the size
+        // as the difference between the size of the type of the "outermost
+        // object" and the offset of the field relative to that type.
+        // For example, consider the following structs:
+        // ```
+        // struct S {
+        //   uint32_t x;
+        //   uint32_t y;
+        // };
+        // struct S2 {
+        //   S s;
+        //   uint32_t z;
+        // };
+        // ```
+        // Given an object `S2 s2` the size of the buffer `&s2.s.y`
+        // is the size of the base object type (i.e., `S2`) minus the offset
+        // of `y` relative to the type `S2` (i.e., `4`). So the size of the
+        // buffer is `12 - 4 = 8`.
+        c = getRootType(va) and
+        // we calculate the size based on the last field, to avoid including any padding after it
+        trueSize = max(Field f | | f.getOffsetInClass(c) + getVariableSize(f))
+      |
+        trueSize - v.(Field).getOffsetInClass(c)
+      )
   )
 }
 
@@ -116,12 +120,8 @@ private int isSource(Expr bufferExpr, Element why) {
   exists(Variable bufferVar | bufferVar = bufferExpr.(VariableAccess).getTarget() |
     // buffer is a fixed size array
     exists(bufferVar.getUnspecifiedType().(ArrayType).getSize()) and
-    result =
-      unique(int size | // more generous than .getSize() itself, when the array is a class field or similar.
-        size = getSize(bufferExpr)
-      |
-        size
-      ) and
+    // more generous than .getSize() itself, when the array is a class field or similar.
+    result = getSize(bufferExpr) and
     why = bufferVar and
     not memberMayBeVarSize(_, bufferVar) and
     not exists(BuiltInOperationBuiltInOffsetOf offsetof | offsetof.getAChild*() = bufferExpr) and

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

@@ -14,7 +14,9 @@ class PackedTimeType extends Type {
   }
 }
 
-private predicate timeType(string typeName) { typeName = ["_SYSTEMTIME", "SYSTEMTIME", "tm"] }
+private predicate timeType(string typeName) {
+  typeName = ["_SYSTEMTIME", "SYSTEMTIME", "tm", "TIME_FIELDS", "_TIME_FIELDS", "PTIME_FIELDS"]
+}
 
 /**
  * A type that is used to represent times and dates in an 'unpacked' form, that is,
@@ -95,3 +97,24 @@ class StructTmMonthFieldAccess extends MonthFieldAccess {
 class StructTmYearFieldAccess extends YearFieldAccess {
   StructTmYearFieldAccess() { this.getTarget().getName() = "tm_year" }
 }
+
+/**
+ * A `DayFieldAccess` for the `TIME_FIELDS` struct.
+ */
+class TimeFieldsDayFieldAccess extends DayFieldAccess {
+  TimeFieldsDayFieldAccess() { this.getTarget().getName() = "Day" }
+}
+
+/**
+ * A `MonthFieldAccess` for the `TIME_FIELDS` struct.
+ */
+class TimeFieldsMonthFieldAccess extends MonthFieldAccess {
+  TimeFieldsMonthFieldAccess() { this.getTarget().getName() = "Month" }
+}
+
+/**
+ * A `YearFieldAccess` for the `TIME_FIELDS` struct.
+ */
+class TimeFieldsYearFieldAccess extends YearFieldAccess {
+  TimeFieldsYearFieldAccess() { this.getTarget().getName() = "Year" }
+}

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

@@ -8,7 +8,8 @@ import semmle.code.cpp.ir.IR
 private import codeql.util.Void
 private import codeql.controlflow.Guards as SharedGuards
 private import semmle.code.cpp.ir.ValueNumbering
-private import semmle.code.cpp.ir.implementation.raw.internal.TranslatedExpr
+private import semmle.code.cpp.ir.implementation.raw.internal.TranslatedExpr as TE
+private import semmle.code.cpp.ir.implementation.raw.internal.TranslatedFunction as TF
 private import semmle.code.cpp.ir.implementation.raw.internal.InstructionTag
 
 private class BasicBlock = IRCfg::BasicBlock;
@@ -683,24 +684,26 @@ final class GuardCondition = GuardConditionImpl;
  */
 private class GuardConditionFromBinaryLogicalOperator extends GuardConditionImpl instanceof Cpp::BinaryLogicalOperation
 {
+  GuardConditionImpl l;
+  GuardConditionImpl r;
+
+  GuardConditionFromBinaryLogicalOperator() {
+    super.getLeftOperand() = l and
+    super.getRightOperand() = r
+  }
+
   override predicate valueControls(Cpp::BasicBlock controlled, GuardValue v) {
-    exists(Cpp::BinaryLogicalOperation binop, GuardCondition lhs, GuardCondition rhs |
-      this = binop and
-      lhs = binop.getLeftOperand() and
-      rhs = binop.getRightOperand() and
-      lhs.valueControls(controlled, v) and
-      rhs.valueControls(controlled, v)
-    )
+    // `l || r` does not control `r` even though `l` does.
+    not r.(Cpp::Expr).getBasicBlock() = controlled and
+    l.valueControls(controlled, v)
+    or
+    r.valueControls(controlled, v)
   }
 
   override predicate valueControlsEdge(Cpp::BasicBlock pred, Cpp::BasicBlock succ, GuardValue v) {
-    exists(Cpp::BinaryLogicalOperation binop, GuardCondition lhs, GuardCondition rhs |
-      this = binop and
-      lhs = binop.getLeftOperand() and
-      rhs = binop.getRightOperand() and
-      lhs.valueControlsEdge(pred, succ, v) and
-      rhs.valueControlsEdge(pred, succ, v)
-    )
+    l.valueControlsEdge(pred, succ, v)
+    or
+    r.valueControlsEdge(pred, succ, v)
   }
 
   pragma[nomagic]
@@ -1026,7 +1029,7 @@ private class GuardConditionFromIR extends GuardConditionImpl {
 
 private predicate excludeAsControlledInstruction(Instruction instr) {
   // Exclude the temporaries generated by a ternary expression.
-  exists(TranslatedConditionalExpr tce |
+  exists(TE::TranslatedConditionalExpr tce |
     instr = tce.getInstruction(ConditionValueFalseStoreTag())
     or
     instr = tce.getInstruction(ConditionValueTrueStoreTag())
@@ -1038,6 +1041,14 @@ private predicate excludeAsControlledInstruction(Instruction instr) {
   or
   // Exclude unreached instructions, as their AST is the whole function and not a block.
   instr instanceof UnreachedInstruction
+  or
+  // Exclude instructions generated by a translated function as they map to the function itself
+  // and the function is considered the last basic block of a function body.
+  any(TF::TranslatedFunction tf).getInstruction(_) = instr
+  or
+  // `ChiInstruction`s generated by instructions in the above case don't come from `getInstruction` (since they are generated by AliasedSSA)
+  // so we need to special case them.
+  excludeAsControlledInstruction(instr.(ChiInstruction).getPartial())
 }
 
 /**

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

@@ -95,6 +95,7 @@
 
 import cpp
 private import new.DataFlow
+private import semmle.code.cpp.controlflow.IRGuards
 private import semmle.code.cpp.ir.dataflow.internal.DataFlowPrivate as Private
 private import semmle.code.cpp.ir.dataflow.internal.DataFlowUtil
 private import internal.FlowSummaryImpl
@@ -367,6 +368,8 @@ private predicate elementSpec(
 ) {
   sourceModel(namespace, type, subtypes, name, signature, ext, _, _, _, _) or
   sinkModel(namespace, type, subtypes, name, signature, ext, _, _, _, _) or
+  barrierModel(namespace, type, subtypes, name, signature, ext, _, _, _, _) or
+  barrierGuardModel(namespace, type, subtypes, name, signature, ext, _, _, _, _, _) or
   summaryModel(namespace, type, subtypes, name, signature, ext, _, _, _, _, _)
 }
 
@@ -1028,6 +1031,84 @@ private module Cached {
       isSinkNode(n, kind, model) and n.asNode() = node
     )
   }
+
+  private newtype TKindModelPair =
+    TMkPair(string kind, string model) { isBarrierGuardNode(_, _, kind, model) }
+
+  private GuardValue convertAcceptingValue(Public::AcceptingValue av) {
+    av.isTrue() and result.asBooleanValue() = true
+    or
+    av.isFalse() and result.asBooleanValue() = false
+    or
+    // NOTE: The below cases don't contribute anything currently since the
+    // callers immediately use `.asBooleanValue()` to convert the `GuardValue`
+    // to a boolean. Once we're willing to accept the breaking change of
+    // converting the barrier guard API to use `GuardValue`s instead `Boolean`s
+    // we can remove this restriction.
+    av.isNoException() and result.getDualValue().isThrowsException()
+    or
+    av.isZero() and result.asIntValue() = 0
+    or
+    av.isNotZero() and result.getDualValue().asIntValue() = 0
+    or
+    av.isNull() and result.isNullValue()
+    or
+    av.isNotNull() and result.isNonNullValue()
+  }
+
+  private predicate barrierGuardChecks(IRGuardCondition g, Expr e, boolean gv, TKindModelPair kmp) {
+    exists(
+      SourceSinkInterpretationInput::InterpretNode n, Public::AcceptingValue acceptingvalue,
+      string kind, string model
+    |
+      isBarrierGuardNode(n, acceptingvalue, kind, model) and
+      n.asNode().asExpr() = e and
+      kmp = TMkPair(kind, model) and
+      gv = convertAcceptingValue(acceptingvalue).asBooleanValue() and
+      n.asNode().(Private::ArgumentNode).getCall().asCallInstruction() = g
+    )
+  }
+
+  private newtype TKindModelPairIntPair =
+    MkKindModelPairIntPair(TKindModelPair pair, int indirectionIndex) {
+      indirectionIndex > 0 and
+      Private::nodeHasInstruction(_, _, indirectionIndex) and
+      exists(pair)
+    }
+
+  private predicate indirectBarrierGuardChecks(
+    IRGuardCondition g, Expr e, boolean gv, TKindModelPairIntPair kmp
+  ) {
+    exists(
+      SourceSinkInterpretationInput::InterpretNode interpretNode,
+      Public::AcceptingValue acceptingvalue, string kind, string model, int indirectionIndex,
+      Private::ArgumentNode arg
+    |
+      isBarrierGuardNode(interpretNode, acceptingvalue, kind, model) and
+      arg = interpretNode.asNode() and
+      arg.asIndirectExpr(indirectionIndex) = e and
+      kmp = MkKindModelPairIntPair(TMkPair(kind, model), indirectionIndex) and
+      gv = convertAcceptingValue(acceptingvalue).asBooleanValue() and
+      arg.getCall().asCallInstruction() = g
+    )
+  }
+
+  /**
+   * Holds if `node` is specified as a barrier with the given kind in a MaD flow
+   * model.
+   */
+  cached
+  predicate barrierNode(DataFlow::Node node, string kind, string model) {
+    exists(SourceSinkInterpretationInput::InterpretNode n |
+      isBarrierNode(n, kind, model) and n.asNode() = node
+    )
+    or
+    DataFlow::ParameterizedBarrierGuard<TKindModelPair, barrierGuardChecks/4>::getABarrierNode(TMkPair(kind,
+        model)) = node
+    or
+    DataFlow::ParameterizedBarrierGuard<TKindModelPairIntPair, indirectBarrierGuardChecks/4>::getAnIndirectBarrierNode(MkKindModelPairIntPair(TMkPair(kind,
+          model), _)) = node
+  }
 }
 
 import Cached
@@ -1044,6 +1125,12 @@ predicate sourceNode(DataFlow::Node node, string kind) { sourceNode(node, kind,
  */
 predicate sinkNode(DataFlow::Node node, string kind) { sinkNode(node, kind, _) }
 
+/**
+ * Holds if `node` is specified as a barrier with the given kind in a MaD flow
+ * model.
+ */
+predicate barrierNode(DataFlow::Node node, string kind) { barrierNode(node, kind, _) }
+
 private predicate interpretSummary(
   Function f, string input, string output, string kind, string provenance, string model
 ) {
@@ -1058,40 +1145,22 @@ private predicate interpretSummary(
 
 // adapter class for converting Mad summaries to `SummarizedCallable`s
 private class SummarizedCallableAdapter extends SummarizedCallable {
-  SummarizedCallableAdapter() { interpretSummary(this, _, _, _, _, _) }
+  string input_;
+  string output_;
+  string kind;
+  Provenance p_;
+  string model_;
 
-  private predicate relevantSummaryElementManual(
-    string input, string output, string kind, string model
-  ) {
-    exists(Provenance provenance |
-      interpretSummary(this, input, output, kind, provenance, model) and
-      provenance.isManual()
-    )
-  }
-
-  private predicate relevantSummaryElementGenerated(
-    string input, string output, string kind, string model
-  ) {
-    exists(Provenance provenance |
-      interpretSummary(this, input, output, kind, provenance, model) and
-      provenance.isGenerated()
-    )
-  }
+  SummarizedCallableAdapter() { interpretSummary(this, input_, output_, kind, p_, model_) }
 
   override predicate propagatesFlow(
-    string input, string output, boolean preservesValue, string model
+    string input, string output, boolean preservesValue, Provenance p, boolean isExact, string model
   ) {
-    exists(string kind |
-      this.relevantSummaryElementManual(input, output, kind, model)
-      or
-      not this.relevantSummaryElementManual(_, _, _, _) and
-      this.relevantSummaryElementGenerated(input, output, kind, model)
-    |
-      if kind = "value" then preservesValue = true else preservesValue = false
-    )
-  }
-
-  override predicate hasProvenance(Provenance provenance) {
-    interpretSummary(this, _, _, _, provenance, _)
+    input = input_ and
+    output = output_ and
+    (if kind = "value" then preservesValue = true else preservesValue = false) and
+    p = p_ and
+    isExact = true and
+    model = model_
   }
 }

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

@@ -20,6 +20,8 @@ module Input implements InputSig<Location, DataFlowImplSpecific::CppDataFlow> {
 
   class SinkBase = Void;
 
+  predicate callableFromSource(SummarizedCallableBase c) { exists(c.getBlock()) }
+
   ArgumentPosition callbackSelfParameterPosition() { result = TDirectPosition(-1) }
 
   ReturnKind getStandardReturnValueKind() { result = getReturnValueKind("") }
@@ -149,16 +151,27 @@ module SourceSinkInterpretationInput implements
   }
 
   predicate barrierElement(
-    Element n, string output, string kind, Public::Provenance provenance, string model
+    Element e, string output, string kind, Public::Provenance provenance, string model
   ) {
-    none()
+    exists(
+      string namespace, string type, boolean subtypes, string name, string signature, string ext
+    |
+      barrierModel(namespace, type, subtypes, name, signature, ext, output, kind, provenance, model) and
+      e = interpretElement(namespace, type, subtypes, name, signature, ext)
+    )
   }
 
   predicate barrierGuardElement(
-    Element n, string input, Public::AcceptingValue acceptingvalue, string kind,
+    Element e, string input, Public::AcceptingValue acceptingvalue, string kind,
     Public::Provenance provenance, string model
   ) {
-    none()
+    exists(
+      string package, string type, boolean subtypes, string name, string signature, string ext
+    |
+      barrierGuardModel(package, type, subtypes, name, signature, ext, input, acceptingvalue, kind,
+        provenance, model) and
+      e = interpretElement(package, type, subtypes, name, signature, ext)
+    )
   }
 
   private newtype TInterpretNode =

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

@@ -23,7 +23,7 @@ class Expr extends StmtParent, @expr {
   predicate hasChild(Expr e, int n) { e = this.getChild(n) }
 
   /** Gets the enclosing function of this expression, if any. */
-  Function getEnclosingFunction() { result = exprEnclosingElement(this) }
+  override Function getEnclosingFunction() { result = exprEnclosingElement(this) }
 
   /** Gets the nearest enclosing set of curly braces around this expression in the source, if any. */
   BlockStmt getEnclosingBlock() { result = this.getEnclosingStmt().getEnclosingBlock() }

cpp/ql/lib/semmle/code/cpp/internal/Overlay.qll

@@ -34,6 +34,38 @@ private string getSingleLocationFilePath(@element e) {
     macroinvocations(e, _, loc, _)
     or
     preprocdirects(e, _, loc)
+    or
+    diagnostics(e, _, _, _, _, loc)
+    or
+    usings(e, _, loc, _)
+    or
+    static_asserts(e, _, _, loc, _)
+    or
+    derivations(e, _, _, _, loc)
+    or
+    frienddecls(e, _, _, loc)
+    or
+    comments(e, _, loc)
+    or
+    exprs(e, _, loc)
+    or
+    stmts(e, _, loc)
+    or
+    initialisers(e, _, _, loc)
+    or
+    attributes(e, _, _, _, loc)
+    or
+    attribute_args(e, _, _, _, loc)
+    or
+    namequalifiers(e, _, _, loc)
+    or
+    enumconstants(e, _, _, _, _, loc)
+    or
+    type_mentions(e, _, loc, _)
+    or
+    lambda_capture(e, _, _, _, _, _, loc)
+    or
+    concept_templates(e, _, loc)
   |
     result = getLocationFilePath(loc)
   )
@@ -45,13 +77,13 @@ private string getSingleLocationFilePath(@element e) {
 overlay[local]
 private string getMultiLocationFilePath(@element e) {
   exists(@location_default loc |
-    exists(@var_decl vd | var_decls(vd, e, _, _, loc))
+    var_decls(_, e, _, _, loc)
     or
-    exists(@fun_decl fd | fun_decls(fd, e, _, _, loc))
+    fun_decls(_, e, _, _, loc)
     or
-    exists(@type_decl td | type_decls(td, e, loc))
+    type_decls(_, e, loc)
     or
-    exists(@namespace_decl nd | namespace_decls(nd, e, loc, _))
+    namespace_decls(_, e, loc, _)
   |
     result = getLocationFilePath(loc)
   )
@@ -62,19 +94,29 @@ private string getMultiLocationFilePath(@element e) {
  * overlay variant.
  */
 overlay[local]
-private predicate holdsInBase() { not isOverlay() }
+private predicate isBase() { not isOverlay() }
+
+/**
+ * Holds if `path` was extracted in the overlay database.
+ */
+overlay[local]
+private predicate overlayHasFile(string path) {
+  isOverlay() and
+  files(_, path) and
+  path != ""
+}
 
 /**
  * Discards an element from the base variant if:
- * - It has a single location in a changed file, or
- * - All of its locations are in changed files.
+ * - It has a single location in a file extracted in the overlay, or
+ * - All of its locations are in files extracted in the overlay.
  */
 overlay[discard_entity]
 private predicate discardElement(@element e) {
-  holdsInBase() and
+  isBase() and
   (
-    overlayChangedFiles(getSingleLocationFilePath(e))
+    overlayHasFile(getSingleLocationFilePath(e))
     or
-    forex(string path | path = getMultiLocationFilePath(e) | overlayChangedFiles(path))
+    forex(string path | path = getMultiLocationFilePath(e) | overlayHasFile(path))
   )
 }

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

@@ -8,81 +8,143 @@ private import cpp
 private import semmle.code.cpp.ir.IR
 
 /**
- * A configuration of a data flow analysis that performs must-flow analysis. This is different
- * from `DataFlow.qll` which performs may-flow analysis (i.e., it finds paths where the source _may_
- * flow to the sink).
- *
- * Like in `DataFlow.qll`, each use of the `MustFlow.qll` library must define its own unique extension
- * of this abstract class. To create a configuration, extend this class with a subclass whose
- * characteristic predicate is a unique singleton string and override `isSource`, `isSink` (and
- * `isAdditionalFlowStep` if additional steps are required).
+ * Provides an inter-procedural must-flow data flow analysis.
  */
-abstract class MustFlowConfiguration extends string {
-  bindingset[this]
-  MustFlowConfiguration() { any() }
-
+module MustFlow {
   /**
-   * Holds if `source` is a relevant data flow source.
+   * An input configuration of a data flow analysis that performs must-flow analysis. This is different
+   * from `DataFlow.qll` which performs may-flow analysis (i.e., it finds paths where the source _may_
+   * flow to the sink).
    */
-  abstract predicate isSource(Instruction source);
+  signature module ConfigSig {
+    /**
+     * Holds if `source` is a relevant data flow source.
+     */
+    predicate isSource(Instruction source);
 
-  /**
-   * Holds if `sink` is a relevant data flow sink.
-   */
-  abstract predicate isSink(Operand sink);
+    /**
+     * Holds if `sink` is a relevant data flow sink.
+     */
+    predicate isSink(Operand sink);
 
-  /**
-   * Holds if data flow through `instr` is prohibited.
-   */
-  predicate isBarrier(Instruction instr) { none() }
+    /**
+     * Holds if data flow through `instr` is prohibited.
+     */
+    default predicate isBarrier(Instruction instr) { none() }
 
-  /**
-   * Holds if the additional flow step from `node1` to `node2` must be taken
-   * into account in the analysis.
-   */
-  predicate isAdditionalFlowStep(Operand node1, Instruction node2) { none() }
+    /**
+     * Holds if the additional flow step from `node1` to `node2` must be taken
+     * into account in the analysis.
+     */
+    default predicate isAdditionalFlowStep(Operand node1, Instruction node2) { none() }
 
-  /** Holds if this configuration allows flow from arguments to parameters. */
-  predicate allowInterproceduralFlow() { any() }
-
-  /**
-   * Holds if data must flow from `source` to `sink` for this configuration.
-   *
-   * The corresponding paths are generated from the end-points and the graph
-   * included in the module `PathGraph`.
-   */
-  final predicate hasFlowPath(MustFlowPathNode source, MustFlowPathSink sink) {
-    this.isSource(source.getInstruction()) and
-    source.getASuccessor*() = sink
+    /** Holds if this configuration allows flow from arguments to parameters. */
+    default predicate allowInterproceduralFlow() { any() }
   }
-}
 
-/** Holds if `node` flows from a source. */
-pragma[nomagic]
-private predicate flowsFromSource(Instruction node, MustFlowConfiguration config) {
-  not config.isBarrier(node) and
-  (
-    config.isSource(node)
-    or
-    exists(Instruction mid |
-      step(mid, node, config) and
-      flowsFromSource(mid, pragma[only_bind_into](config))
-    )
-  )
-}
+  /**
+   * Constructs a global must-flow computation.
+   */
+  module Global<ConfigSig Config> {
+    import Config
 
-/** Holds if `node` flows to a sink. */
-pragma[nomagic]
-private predicate flowsToSink(Instruction node, MustFlowConfiguration config) {
-  flowsFromSource(node, pragma[only_bind_into](config)) and
-  (
-    config.isSink(node.getAUse())
-    or
-    exists(Instruction mid |
-      step(node, mid, config) and
-      flowsToSink(mid, pragma[only_bind_into](config))
-    )
-  )
+    /**
+     * Holds if data must flow from `source` to `sink`.
+     *
+     * The corresponding paths are generated from the end-points and the graph
+     * included in the module `PathGraph`.
+     */
+    predicate flowPath(PathNode source, PathSink sink) {
+      isSource(source.getInstruction()) and
+      source.getASuccessor*() = sink
+    }
+
+    /** Holds if `node` flows from a source. */
+    pragma[nomagic]
+    private predicate flowsFromSource(Instruction node) {
+      not isBarrier(node) and
+      (
+        isSource(node)
+        or
+        exists(Instruction mid |
+          step(mid, node) and
+          flowsFromSource(mid)
+        )
+      )
+    }
+
+    /** Holds if `node` flows to a sink. */
+    pragma[nomagic]
+    private predicate flowsToSink(Instruction node) {
+      flowsFromSource(node) and
+      (
+        isSink(node.getAUse())
+        or
+        exists(Instruction mid |
+          step(node, mid) and
+          flowsToSink(mid)
+        )
+      )
+    }
+
+    /** Holds if `nodeFrom` flows to `nodeTo`. */
+    private predicate step(Instruction nodeFrom, Instruction nodeTo) {
+      Cached::localStep(nodeFrom, nodeTo)
+      or
+      allowInterproceduralFlow() and
+      Cached::flowThroughCallable(nodeFrom, nodeTo)
+      or
+      isAdditionalFlowStep(nodeFrom.getAUse(), nodeTo)
+    }
+
+    private newtype TLocalPathNode =
+      MkLocalPathNode(Instruction n) {
+        flowsToSink(n) and
+        (
+          isSource(n)
+          or
+          exists(PathNode mid | step(mid.getInstruction(), n))
+        )
+      }
+
+    /** A `Node` that is in a path from a source to a sink. */
+    class PathNode extends TLocalPathNode {
+      Instruction n;
+
+      PathNode() { this = MkLocalPathNode(n) }
+
+      /** Gets the underlying node. */
+      Instruction getInstruction() { result = n }
+
+      /** Gets a textual representation of this node. */
+      string toString() { result = n.getAst().toString() }
+
+      /** Gets the location of this element. */
+      Location getLocation() { result = n.getLocation() }
+
+      /** Gets a successor node, if any. */
+      PathNode getASuccessor() { step(this.getInstruction(), result.getInstruction()) }
+    }
+
+    private class PathSink extends PathNode {
+      PathSink() { isSink(this.getInstruction().getAUse()) }
+    }
+
+    /**
+     * Provides the query predicates needed to include a graph in a path-problem query.
+     */
+    module PathGraph {
+      private predicate reach(PathNode n) { n instanceof PathSink or reach(n.getASuccessor()) }
+
+      /** Holds if `(a,b)` is an edge in the graph of data flow path explanations. */
+      query predicate edges(PathNode a, PathNode b) { a.getASuccessor() = b and reach(b) }
+
+      /** Holds if `n` is a node in the graph of data flow path explanations. */
+      query predicate nodes(PathNode n, string key, string val) {
+        reach(n) and key = "semmle.label" and val = n.toString()
+      }
+    }
+  }
 }
 
 cached
@@ -102,7 +164,7 @@ private module Cached {
     not f.isVirtual() and
     call.getPositionalArgument(n) = instr and
     f = call.getStaticCallTarget() and
-    getEnclosingNonVirtualFunctionInitializeParameter(init, f) and
+    isEnclosingNonVirtualFunctionInitializeParameter(init, f) and
     init.getParameter().getIndex() = pragma[only_bind_into](pragma[only_bind_out](n))
   }
 
@@ -111,7 +173,7 @@ private module Cached {
    * corresponding initialization instruction that receives the value of `instr` in `f`.
    */
   pragma[noinline]
-  private predicate getPositionalArgumentInitParam(
+  private predicate isPositionalArgumentInitParam(
     CallInstruction call, Instruction instr, InitializeParameterInstruction init, Function f
   ) {
     exists(int n |
@@ -126,18 +188,18 @@ private module Cached {
    * `instr` in `f`.
    */
   pragma[noinline]
-  private predicate getThisArgumentInitParam(
+  private predicate isThisArgumentInitParam(
     CallInstruction call, Instruction instr, InitializeParameterInstruction init, Function f
   ) {
     not f.isVirtual() and
     call.getStaticCallTarget() = f and
-    getEnclosingNonVirtualFunctionInitializeParameter(init, f) and
+    isEnclosingNonVirtualFunctionInitializeParameter(init, f) and
     call.getThisArgument() = instr and
     init.getIRVariable() instanceof IRThisVariable
   }
 
   /** Holds if `f` is the enclosing non-virtual function of `init`. */
-  private predicate getEnclosingNonVirtualFunctionInitializeParameter(
+  private predicate isEnclosingNonVirtualFunctionInitializeParameter(
     InitializeParameterInstruction init, Function f
   ) {
     not f.isVirtual() and
@@ -145,7 +207,7 @@ private module Cached {
   }
 
   /** Holds if `f` is the enclosing non-virtual function of `init`. */
-  private predicate getEnclosingNonVirtualFunctionInitializeIndirection(
+  private predicate isEnclosingNonVirtualFunctionInitializeIndirection(
     InitializeIndirectionInstruction init, Function f
   ) {
     not f.isVirtual() and
@@ -153,15 +215,16 @@ private module Cached {
   }
 
   /**
-   * Holds if `instr` is an argument (or argument indirection) to a call, and
-   * `succ` is the corresponding initialization instruction in the call target.
+   * Holds if `argument` is an argument (or argument indirection) to a call, and
+   * `parameter` is the corresponding initialization instruction in the call target.
    */
-  private predicate flowThroughCallable(Instruction argument, Instruction parameter) {
+  cached
+  predicate flowThroughCallable(Instruction argument, Instruction parameter) {
     // Flow from an argument to a parameter
     exists(CallInstruction call, InitializeParameterInstruction init | init = parameter |
-      getPositionalArgumentInitParam(call, argument, init, call.getStaticCallTarget())
+      isPositionalArgumentInitParam(call, argument, init, call.getStaticCallTarget())
       or
-      getThisArgumentInitParam(call, argument, init, call.getStaticCallTarget())
+      isThisArgumentInitParam(call, argument, init, call.getStaticCallTarget())
     )
     or
     // Flow from argument indirection to parameter indirection
@@ -170,7 +233,7 @@ private module Cached {
     |
       init = parameter and
       read.getPrimaryInstruction() = call and
-      getEnclosingNonVirtualFunctionInitializeIndirection(init, call.getStaticCallTarget())
+      isEnclosingNonVirtualFunctionInitializeIndirection(init, call.getStaticCallTarget())
     |
       exists(int n |
         read.getSideEffectOperand().getAnyDef() = argument and
@@ -205,92 +268,10 @@ private module Cached {
   }
 
   cached
-  predicate step(Instruction nodeFrom, Instruction nodeTo) {
+  predicate localStep(Instruction nodeFrom, Instruction nodeTo) {
     exists(Operand mid |
       instructionToOperandStep(nodeFrom, mid) and
       operandToInstructionStep(mid, nodeTo)
     )
-    or
-    flowThroughCallable(nodeFrom, nodeTo)
-  }
-}
-
-/**
- * Gets the enclosing callable of `n`. Unlike `n.getEnclosingCallable()`, this
- * predicate ensures that joins go from `n` to the result instead of the other
- * way around.
- */
-pragma[inline]
-private IRFunction getEnclosingCallable(Instruction n) {
-  pragma[only_bind_into](result) = pragma[only_bind_out](n).getEnclosingIRFunction()
-}
-
-/** Holds if `nodeFrom` flows to `nodeTo`. */
-private predicate step(Instruction nodeFrom, Instruction nodeTo, MustFlowConfiguration config) {
-  exists(config) and
-  Cached::step(pragma[only_bind_into](nodeFrom), pragma[only_bind_into](nodeTo)) and
-  (
-    config.allowInterproceduralFlow()
-    or
-    getEnclosingCallable(nodeFrom) = getEnclosingCallable(nodeTo)
-  )
-  or
-  config.isAdditionalFlowStep(nodeFrom.getAUse(), nodeTo)
-}
-
-private newtype TLocalPathNode =
-  MkLocalPathNode(Instruction n, MustFlowConfiguration config) {
-    flowsToSink(n, config) and
-    (
-      config.isSource(n)
-      or
-      exists(MustFlowPathNode mid | step(mid.getInstruction(), n, config))
-    )
-  }
-
-/** A `Node` that is in a path from a source to a sink. */
-class MustFlowPathNode extends TLocalPathNode {
-  Instruction n;
-
-  MustFlowPathNode() { this = MkLocalPathNode(n, _) }
-
-  /** Gets the underlying node. */
-  Instruction getInstruction() { result = n }
-
-  /** Gets a textual representation of this node. */
-  string toString() { result = n.getAst().toString() }
-
-  /** Gets the location of this element. */
-  Location getLocation() { result = n.getLocation() }
-
-  /** Gets a successor node, if any. */
-  MustFlowPathNode getASuccessor() {
-    step(this.getInstruction(), result.getInstruction(), this.getConfiguration())
-  }
-
-  /** Gets the associated configuration. */
-  MustFlowConfiguration getConfiguration() { this = MkLocalPathNode(_, result) }
-}
-
-private class MustFlowPathSink extends MustFlowPathNode {
-  MustFlowPathSink() { this.getConfiguration().isSink(this.getInstruction().getAUse()) }
-}
-
-/**
- * Provides the query predicates needed to include a graph in a path-problem query.
- */
-module PathGraph {
-  private predicate reach(MustFlowPathNode n) {
-    n instanceof MustFlowPathSink or reach(n.getASuccessor())
-  }
-
-  /** Holds if `(a,b)` is an edge in the graph of data flow path explanations. */
-  query predicate edges(MustFlowPathNode a, MustFlowPathNode b) {
-    a.getASuccessor() = b and reach(b)
-  }
-
-  /** Holds if `n` is a node in the graph of data flow path explanations. */
-  query predicate nodes(MustFlowPathNode n, string key, string val) {
-    reach(n) and key = "semmle.label" and val = n.toString()
   }
 }

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

@@ -1144,7 +1144,7 @@ private newtype TDataFlowCall =
   }
 
 private predicate summarizedCallableIsManual(SummarizedCallable sc) {
-  sc.asSummarizedCallable().applyManualModel()
+  sc.asSummarizedCallable().hasManualModel()
 }
 
 /**

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

@@ -312,6 +312,13 @@ class Node extends TIRDataFlowNode {
    */
   Expr asDefinition() { result = this.asDefinition(_) }
 
+  private predicate isCertainStore() {
+    exists(SsaImpl::Definition def |
+      SsaImpl::defToNode(this, def, _) and
+      def.isCertain()
+    )
+  }
+
   /**
    * Gets the definition associated with this node, if any.
    *
@@ -361,11 +368,10 @@ class Node extends TIRDataFlowNode {
    * pointed to by `p`.
    */
   Expr asDefinition(boolean uncertain) {
-    exists(StoreInstruction store, SsaImpl::Definition def |
+    exists(StoreInstruction store |
       store = this.asInstruction() and
       result = asDefinitionImpl(store) and
-      SsaImpl::defToNode(this, def, _) and
-      if def.isCertain() then uncertain = false else uncertain = true
+      if this.isCertainStore() then uncertain = false else uncertain = true
     )
   }
 
@@ -1720,9 +1726,7 @@ private module Cached {
       SsaImpl::ssaFlow(n, succ) and
       bb1 = n.getBasicBlock() and
       bb2 = succ.getBasicBlock() and
-      bb1 != bb2 and
-      bb2.dominates(bb1) and
-      bb1.getASuccessor+() = bb2
+      bb2.strictlyDominates(bb1)
     )
   }
 
@@ -2417,6 +2421,19 @@ class ContentSet instanceof Content {
   }
 }
 
+private signature class ParamSig;
+
+private module WithParam<ParamSig P> {
+  /**
+   * Holds if the guard `g` validates the expression `e` upon evaluating to `branch`.
+   *
+   * The expression `e` is expected to be a syntactic part of the guard `g`.
+   * For example, the guard `g` might be a call `isSafe(x)` and the expression `e`
+   * the argument `x`.
+   */
+  signature predicate guardChecksSig(IRGuardCondition g, Expr e, boolean branch, P param);
+}
+
 /**
  * Holds if the guard `g` validates the expression `e` upon evaluating to `branch`.
  *
@@ -2438,7 +2455,7 @@ private predicate controls(IRGuardCondition g, Node n, boolean edge) {
  * This is expected to be used in `isBarrier`/`isSanitizer` definitions
  * in data flow and taint tracking.
  */
-module BarrierGuard<guardChecksSig/3 guardChecks> {
+module ParameterizedBarrierGuard<ParamSig P, WithParam<P>::guardChecksSig/4 guardChecks> {
   bindingset[value, n]
   pragma[inline_late]
   private predicate convertedExprHasValueNumber(ValueNumber value, Node n) {
@@ -2448,12 +2465,13 @@ module BarrierGuard<guardChecksSig/3 guardChecks> {
     )
   }
 
-  private predicate guardChecksNode(IRGuardCondition g, Node n, boolean branch) {
-    guardChecks(g, n.asOperand().getDef().getConvertedResultExpression(), branch)
+  private predicate guardChecksNode(IRGuardCondition g, Node n, boolean branch, P p) {
+    guardChecks(g, n.asOperand().getDef().getConvertedResultExpression(), branch, p)
   }
 
   /**
-   * Gets an expression node that is safely guarded by the given guard check.
+   * Gets an expression node that is safely guarded by the given guard check
+   * when the parameter is `p`.
    *
    * For example, given the following code:
    * ```cpp
@@ -2484,19 +2502,27 @@ module BarrierGuard<guardChecksSig/3 guardChecks> {
    *
    * NOTE: If an indirect expression is tracked, use `getAnIndirectBarrierNode` instead.
    */
-  Node getABarrierNode() {
+  Node getABarrierNode(P p) {
     exists(IRGuardCondition g, ValueNumber value, boolean edge |
       convertedExprHasValueNumber(value, result) and
       guardChecks(g,
-        pragma[only_bind_into](value.getAnInstruction().getConvertedResultExpression()), edge) and
+        pragma[only_bind_into](value.getAnInstruction().getConvertedResultExpression()), edge, p) and
       controls(g, result, edge)
     )
     or
-    result = SsaImpl::BarrierGuard<guardChecksNode/3>::getABarrierNode()
+    result = SsaImpl::BarrierGuard<P, guardChecksNode/4>::getABarrierNode(p)
   }
 
   /**
-   * Gets an indirect expression node that is safely guarded by the given guard check.
+   * Gets an expression node that is safely guarded by the given guard check.
+   *
+   * See `getABarrierNode/1` for examples.
+   */
+  Node getABarrierNode() { result = getABarrierNode(_) }
+
+  /**
+   * Gets an indirect expression node that is safely guarded by the given
+   * guard check with parameter `p`.
    *
    * For example, given the following code:
    * ```cpp
@@ -2528,6 +2554,13 @@ module BarrierGuard<guardChecksSig/3 guardChecks> {
    *
    * NOTE: If a non-indirect expression is tracked, use `getABarrierNode` instead.
    */
+  Node getAnIndirectBarrierNode(P p) { result = getAnIndirectBarrierNode(_, p) }
+
+  /**
+   * Gets an indirect expression node that is safely guarded by the given guard check.
+   *
+   * See `getAnIndirectBarrierNode/1` for examples.
+   */
   Node getAnIndirectBarrierNode() { result = getAnIndirectBarrierNode(_) }
 
   bindingset[value, n]
@@ -2542,10 +2575,10 @@ module BarrierGuard<guardChecksSig/3 guardChecks> {
   }
 
   private predicate guardChecksIndirectNode(
-    IRGuardCondition g, Node n, boolean branch, int indirectionIndex
+    IRGuardCondition g, Node n, boolean branch, int indirectionIndex, P p
   ) {
     guardChecks(g, n.asIndirectOperand(indirectionIndex).getDef().getConvertedResultExpression(),
-      branch)
+      branch, p)
   }
 
   /**
@@ -2582,19 +2615,91 @@ module BarrierGuard<guardChecksSig/3 guardChecks> {
    *
    * NOTE: If a non-indirect expression is tracked, use `getABarrierNode` instead.
    */
-  Node getAnIndirectBarrierNode(int indirectionIndex) {
+  Node getAnIndirectBarrierNode(int indirectionIndex, P p) {
     exists(IRGuardCondition g, ValueNumber value, boolean edge |
       indirectConvertedExprHasValueNumber(indirectionIndex, value, result) and
       guardChecks(g,
-        pragma[only_bind_into](value.getAnInstruction().getConvertedResultExpression()), edge) and
+        pragma[only_bind_into](value.getAnInstruction().getConvertedResultExpression()), edge, p) and
       controls(g, result, edge)
     )
     or
     result =
-      SsaImpl::BarrierGuardWithIntParam<guardChecksIndirectNode/4>::getABarrierNode(indirectionIndex)
+      SsaImpl::BarrierGuardWithIntParam<P, guardChecksIndirectNode/5>::getABarrierNode(indirectionIndex,
+        p)
   }
 }
 
+/**
+ * Provides a set of barrier nodes for a guard that validates an expression.
+ *
+ * This is expected to be used in `isBarrier`/`isSanitizer` definitions
+ * in data flow and taint tracking.
+ */
+module BarrierGuard<guardChecksSig/3 guardChecks> {
+  private predicate guardChecks(IRGuardCondition g, Expr e, boolean branch, Unit unit) {
+    guardChecks(g, e, branch) and
+    exists(unit)
+  }
+
+  private module P = ParameterizedBarrierGuard<Unit, guardChecks/4>;
+
+  predicate getABarrierNode = P::getABarrierNode/0;
+
+  /**
+   * Gets an indirect expression node with indirection index `indirectionIndex` that is
+   * safely guarded by the given guard check.
+   *
+   * For example, given the following code:
+   * ```cpp
+   * int* p;
+   * // ...
+   * *p = source();
+   * if(is_safe_pointer(p)) {
+   *   sink(*p);
+   * }
+   * ```
+   * and the following barrier guard check:
+   * ```ql
+   * predicate myGuardChecks(IRGuardCondition g, Expr e, boolean branch) {
+   *   exists(Call call |
+   *     g.getUnconvertedResultExpression() = call and
+   *     call.getTarget().hasName("is_safe_pointer") and
+   *     e = call.getAnArgument() and
+   *     branch = true
+   *   )
+   * }
+   * ```
+   * implementing `isBarrier` as:
+   * ```ql
+   * predicate isBarrier(DataFlow::Node barrier) {
+   *   barrier = DataFlow::BarrierGuard<myGuardChecks/3>::getAnIndirectBarrierNode(1)
+   * }
+   * ```
+   * will block flow from `x = source()` to `sink(x)`.
+   *
+   * NOTE: If a non-indirect expression is tracked, use `getABarrierNode` instead.
+   */
+  Node getAnIndirectBarrierNode(int indirectionIndex) {
+    result = P::getAnIndirectBarrierNode(indirectionIndex, _)
+  }
+
+  /**
+   * Gets an indirect expression node that is safely guarded by the given guard check.
+   *
+   * See `getAnIndirectBarrierNode/1` for examples.
+   */
+  Node getAnIndirectBarrierNode() { result = getAnIndirectBarrierNode(_) }
+}
+
+private module InstrWithParam<ParamSig P> {
+  /**
+   * Holds if the guard `g` validates the instruction `instr` upon evaluating to `branch`.
+   */
+  signature predicate instructionGuardChecksSig(
+    IRGuardCondition g, Instruction instr, boolean branch, P p
+  );
+}
+
 /**
  * Holds if the guard `g` validates the instruction `instr` upon evaluating to `branch`.
  */
@@ -2606,7 +2711,9 @@ signature predicate instructionGuardChecksSig(IRGuardCondition g, Instruction in
  * This is expected to be used in `isBarrier`/`isSanitizer` definitions
  * in data flow and taint tracking.
  */
-module InstructionBarrierGuard<instructionGuardChecksSig/3 instructionGuardChecks> {
+module ParameterizedInstructionBarrierGuard<
+  ParamSig P, InstrWithParam<P>::instructionGuardChecksSig/4 instructionGuardChecks>
+{
   bindingset[value, n]
   pragma[inline_late]
   private predicate operandHasValueNumber(ValueNumber value, Node n) {
@@ -2616,21 +2723,27 @@ module InstructionBarrierGuard<instructionGuardChecksSig/3 instructionGuardCheck
     )
   }
 
-  private predicate guardChecksNode(IRGuardCondition g, Node n, boolean branch) {
-    instructionGuardChecks(g, n.asOperand().getDef(), branch)
+  private predicate guardChecksNode(IRGuardCondition g, Node n, boolean branch, P p) {
+    instructionGuardChecks(g, n.asOperand().getDef(), branch, p)
   }
 
-  /** Gets a node that is safely guarded by the given guard check. */
-  Node getABarrierNode() {
+  /**
+   * Gets a node that is safely guarded by the given guard check with
+   * parameter `p`.
+   */
+  Node getABarrierNode(P p) {
     exists(IRGuardCondition g, ValueNumber value, boolean edge |
-      instructionGuardChecks(g, pragma[only_bind_into](value.getAnInstruction()), edge) and
+      instructionGuardChecks(g, pragma[only_bind_into](value.getAnInstruction()), edge, p) and
       operandHasValueNumber(value, result) and
       controls(g, result, edge)
     )
     or
-    result = SsaImpl::BarrierGuard<guardChecksNode/3>::getABarrierNode()
+    result = SsaImpl::BarrierGuard<P, guardChecksNode/4>::getABarrierNode(p)
   }
 
+  /** Gets a node that is safely guarded by the given guard check. */
+  Node getABarrierNode() { result = getABarrierNode(_) }
+
   bindingset[value, n]
   pragma[inline_late]
   private predicate indirectOperandHasValueNumber(ValueNumber value, int indirectionIndex, Node n) {
@@ -2641,25 +2754,65 @@ module InstructionBarrierGuard<instructionGuardChecksSig/3 instructionGuardCheck
   }
 
   private predicate guardChecksIndirectNode(
-    IRGuardCondition g, Node n, boolean branch, int indirectionIndex
+    IRGuardCondition g, Node n, boolean branch, int indirectionIndex, P p
   ) {
-    instructionGuardChecks(g, n.asIndirectOperand(indirectionIndex).getDef(), branch)
+    instructionGuardChecks(g, n.asIndirectOperand(indirectionIndex).getDef(), branch, p)
   }
 
+  /**
+   * Gets an indirect node with indirection index `indirectionIndex` that is
+   * safely guarded by the given guard check with parameter `p`.
+   */
+  Node getAnIndirectBarrierNode(int indirectionIndex, P p) {
+    exists(IRGuardCondition g, ValueNumber value, boolean edge |
+      instructionGuardChecks(g, pragma[only_bind_into](value.getAnInstruction()), edge, p) and
+      indirectOperandHasValueNumber(value, indirectionIndex, result) and
+      controls(g, result, edge)
+    )
+    or
+    result =
+      SsaImpl::BarrierGuardWithIntParam<P, guardChecksIndirectNode/5>::getABarrierNode(indirectionIndex,
+        p)
+  }
+
+  /**
+   * Gets an indirect node that is safely guarded by the given guard check
+   * with parameter `p`.
+   */
+  Node getAnIndirectBarrierNode(P p) { result = getAnIndirectBarrierNode(_, p) }
+
+  /** Gets an indirect node that is safely guarded by the given guard check. */
+  Node getAnIndirectBarrierNode() { result = getAnIndirectBarrierNode(_) }
+}
+
+/**
+ * Provides a set of barrier nodes for a guard that validates an instruction.
+ *
+ * This is expected to be used in `isBarrier`/`isSanitizer` definitions
+ * in data flow and taint tracking.
+ */
+module InstructionBarrierGuard<instructionGuardChecksSig/3 instructionGuardChecks> {
+  private predicate instructionGuardChecks(
+    IRGuardCondition g, Instruction i, boolean branch, Unit unit
+  ) {
+    instructionGuardChecks(g, i, branch) and
+    exists(unit)
+  }
+
+  private module P = ParameterizedInstructionBarrierGuard<Unit, instructionGuardChecks/4>;
+
+  predicate getABarrierNode = P::getABarrierNode/0;
+
   /**
    * Gets an indirect node with indirection index `indirectionIndex` that is
    * safely guarded by the given guard check.
    */
   Node getAnIndirectBarrierNode(int indirectionIndex) {
-    exists(IRGuardCondition g, ValueNumber value, boolean edge |
-      instructionGuardChecks(g, pragma[only_bind_into](value.getAnInstruction()), edge) and
-      indirectOperandHasValueNumber(value, indirectionIndex, result) and
-      controls(g, result, edge)
-    )
-    or
-    result =
-      SsaImpl::BarrierGuardWithIntParam<guardChecksIndirectNode/4>::getABarrierNode(indirectionIndex)
+    result = P::getAnIndirectBarrierNode(indirectionIndex, _)
   }
+
+  /** Gets an indirect node that is safely guarded by the given guard check. */
+  Node getAnIndirectBarrierNode() { result = getAnIndirectBarrierNode(_) }
 }
 
 /**

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

@@ -15,17 +15,79 @@ private import DataFlowPrivate
 import SsaImplCommon
 
 private module SourceVariables {
+  /**
+   * Holds if `store` is the `StoreInstruction` generated by a postfix
+   * increment or decrement operation `e`, and `postCrement` is the operand
+   * that represents the use of the evaluated value of `e`.
+   */
+  private predicate isUseAfterPostfixCrement0(StoreInstruction store, Operand postCrement) {
+    exists(
+      BinaryInstruction binary, IRBlock b, int iPre, int iPost, int iStore, Operand preCrement,
+      Instruction left
+    |
+      binary instanceof AddInstruction
+      or
+      binary instanceof PointerAddInstruction
+      or
+      binary instanceof SubInstruction
+      or
+      binary instanceof PointerSubInstruction
+    |
+      store.getSourceValue() = binary and
+      left = binary.getLeft() and
+      strictcount(left.getAUse()) = 2 and
+      left.getAUse() = preCrement and
+      left.getAUse() = postCrement and
+      b.getInstruction(iPre) = preCrement.getUse() and
+      b.getInstruction(iPost) = postCrement.getUse() and
+      b.getInstruction(iStore) = store and
+      iPre < iStore and
+      iStore < iPost
+    )
+  }
+
+  /**
+   * Holds if `store` is the `StoreInstruction` generated by an postfix
+   * increment or decrement operation `e`, and `postCrement` is the fully
+   * converted operand that represents the use of the evaluated value of `e`.
+   */
+  private predicate isUseAfterPostfixCrement(StoreInstruction store, Operand postCrement) {
+    isUseAfterPostfixCrement0(store, postCrement) and
+    conversionFlow(postCrement, _, false, _)
+    or
+    exists(Instruction instr, Operand postCrement0 |
+      isUseAfterPostfixCrement(store, postCrement0) and
+      conversionFlow(postCrement0, instr, false, _) and
+      instr = postCrement.getDef()
+    )
+  }
+
+  private predicate hasSavedPostfixCrementSourceVariable(
+    BaseSourceVariable base, StoreInstruction store, int ind
+  ) {
+    exists(BaseSourceVariableInstruction inst, int ind0 |
+      isUseAfterPostfixCrement(store, _) and
+      inst.getBaseSourceVariable() = base and
+      isDef(_, _, store.getDestinationAddressOperand(), inst, ind0, 0) and
+      ind = [ind0 .. countIndirectionsForCppType(base.getLanguageType()) + 1]
+    )
+  }
+
   cached
   private newtype TSourceVariable =
-    TMkSourceVariable(BaseSourceVariable base, int ind) {
+    TNormalSourceVariable(BaseSourceVariable base, int ind) {
       ind = [0 .. countIndirectionsForCppType(base.getLanguageType()) + 1]
+    } or
+    TSavedPostfixCrementSourceVariable(StoreInstruction store, int ind) {
+      hasSavedPostfixCrementSourceVariable(_, store, ind)
     }
 
-  class SourceVariable extends TSourceVariable {
+  abstract private class AbstractSourceVariable extends TSourceVariable {
     BaseSourceVariable base;
     int ind;
 
-    SourceVariable() { this = TMkSourceVariable(base, ind) }
+    bindingset[ind]
+    AbstractSourceVariable() { any() }
 
     /** Gets the IR variable associated with this `SourceVariable`, if any. */
     IRVariable getIRVariable() { result = base.(BaseIRVariable).getIRVariable() }
@@ -37,7 +99,7 @@ private module SourceVariables {
     BaseSourceVariable getBaseVariable() { result = base }
 
     /** Gets a textual representation of this element. */
-    string toString() { result = repeatStars(this.getIndirection()) + base.toString() }
+    abstract string toString();
 
     /**
      * Gets the number of loads performed on the base source variable
@@ -62,6 +124,53 @@ private module SourceVariables {
     /** Gets the location of this variable. */
     Location getLocation() { result = this.getBaseVariable().getLocation() }
   }
+
+  final class SourceVariable = AbstractSourceVariable;
+
+  /**
+   * A regular source variable. Most source variables are instances of this
+   * class.
+   */
+  class NormalSourceVariable extends AbstractSourceVariable, TNormalSourceVariable {
+    NormalSourceVariable() { this = TNormalSourceVariable(base, ind) }
+
+    final override string toString() {
+      result = repeatStars(this.getIndirection()) + base.toString()
+    }
+  }
+
+  /**
+   * Before a value is postfix incremented (or decremented) we "save" its
+   * current value so that the pre-incremented value can be returned to the
+   * enclosing expression. We use the source variables represented by this
+   * class to represent the "saved value".
+   */
+  class SavedPostfixCrementSourceVariable extends AbstractSourceVariable,
+    TSavedPostfixCrementSourceVariable
+  {
+    StoreInstruction store;
+
+    SavedPostfixCrementSourceVariable() {
+      this = TSavedPostfixCrementSourceVariable(store, ind) and
+      hasSavedPostfixCrementSourceVariable(base, store, ind)
+    }
+
+    final override string toString() {
+      result = repeatStars(this.getIndirection()) + base.toString() + " [before crement]"
+    }
+
+    /**
+     * Gets the `StoreInstruction` that writes the incremented (or decremented)
+     * value.
+     */
+    StoreInstruction getStoreInstruction() { result = store }
+
+    /**
+     * Gets the fully converted `Operand` that represents the use of the
+     * value before the increment.
+     */
+    Operand getOperand() { isUseAfterPostfixCrement(store, result) }
+  }
 }
 
 import SourceVariables
@@ -109,17 +218,43 @@ private newtype TDefImpl =
   TDirectDefImpl(Operand address, int indirectionIndex) {
     isDef(_, _, address, _, _, indirectionIndex)
   } or
+  TSavedPostfixCrementDefImpl(SavedPostfixCrementSourceVariable sv, int indirectionIndex) {
+    isDef(_, _, sv.getStoreInstruction().getDestinationAddressOperand(), _, sv.getIndirection(),
+      indirectionIndex)
+  } or
   TGlobalDefImpl(GlobalLikeVariable v, IRFunction f, int indirectionIndex) {
     // Represents the initial "definition" of a global variable when entering
     // a function body.
     isGlobalDefImpl(v, f, _, indirectionIndex)
   }
 
+pragma[nomagic]
+private predicate hasOperandAndIndirection(
+  SavedPostfixCrementSourceVariable sv, Operand operand, int indirection
+) {
+  sv.getOperand() = operand and
+  sv.getIndirection() = indirection
+}
+
+private predicate hasBeforePostCrementUseImpl(
+  SavedPostfixCrementSourceVariable sv, Operand operand, int indirectionIndex
+) {
+  not isDef(true, _, operand, _, _, _) and
+  exists(int indirection |
+    hasOperandAndIndirection(sv, operand, indirection) and
+    isUse(_, operand, _, indirection, indirectionIndex)
+  )
+}
+
 cached
 private newtype TUseImpl =
   TDirectUseImpl(Operand operand, int indirectionIndex) {
     isUse(_, operand, _, _, indirectionIndex) and
-    not isDef(true, _, operand, _, _, _)
+    not isDef(true, _, operand, _, _, _) and
+    not hasBeforePostCrementUseImpl(_, operand, indirectionIndex)
+  } or
+  TSavedPostfixCrementUseImpl(SavedPostfixCrementSourceVariable sv, int indirectionIndex) {
+    hasBeforePostCrementUseImpl(sv, _, indirectionIndex)
   } or
   TGlobalUse(GlobalLikeVariable v, IRFunction f, int indirectionIndex) {
     // Represents a final "use" of a global variable to ensure that
@@ -223,19 +358,8 @@ abstract class DefImpl extends TDefImpl {
    */
   abstract int getIndirection();
 
-  /**
-   * Gets the base source variable (i.e., the variable without
-   * any indirection) of this definition or use.
-   */
-  abstract BaseSourceVariable getBaseSourceVariable();
-
   /** Gets the variable that is defined or used. */
-  SourceVariable getSourceVariable() {
-    exists(BaseSourceVariable v, int indirection |
-      sourceVariableHasBaseAndIndex(result, v, indirection) and
-      defHasSourceVariable(this, v, indirection)
-    )
-  }
+  abstract SourceVariable getSourceVariable();
 
   /**
    * Holds if this definition is guaranteed to totally overwrite the
@@ -243,8 +367,8 @@ abstract class DefImpl extends TDefImpl {
    */
   abstract predicate isCertain();
 
-  /** Gets the value written to the destination variable by this definition. */
-  abstract Node0Impl getValue();
+  /** Gets the value written to the destination variable by this definition, if any. */
+  Node0Impl getValue() { none() }
 
   /** Gets the operand that represents the address of this definition, if any. */
   Operand getAddressOperand() { none() }
@@ -293,19 +417,8 @@ abstract class UseImpl extends TUseImpl {
   /** Gets the indirection index of this use. */
   final int getIndirectionIndex() { result = indirectionIndex }
 
-  /**
-   * Gets the base source variable (i.e., the variable without
-   * any indirection) of this definition or use.
-   */
-  abstract BaseSourceVariable getBaseSourceVariable();
-
   /** Gets the variable that is defined or used. */
-  SourceVariable getSourceVariable() {
-    exists(BaseSourceVariable v, int indirection |
-      sourceVariableHasBaseAndIndex(result, v, indirection) and
-      useHasSourceVariable(this, v, indirection)
-    )
-  }
+  abstract SourceVariable getSourceVariable();
 
   /**
    * Holds if this use is guaranteed to read the
@@ -314,18 +427,6 @@ abstract class UseImpl extends TUseImpl {
   abstract predicate isCertain();
 }
 
-pragma[noinline]
-private predicate defHasSourceVariable(DefImpl def, BaseSourceVariable bv, int ind) {
-  bv = def.getBaseSourceVariable() and
-  ind = def.getIndirection()
-}
-
-pragma[noinline]
-private predicate useHasSourceVariable(UseImpl use, BaseSourceVariable bv, int ind) {
-  bv = use.getBaseSourceVariable() and
-  ind = use.getIndirection()
-}
-
 pragma[noinline]
 private predicate sourceVariableHasBaseAndIndex(SourceVariable v, BaseSourceVariable bv, int ind) {
   v.getBaseVariable() = bv and
@@ -358,16 +459,12 @@ abstract private class DefAddressImpl extends DefImpl, TDefAddressImpl {
 
   final override predicate isCertain() { any() }
 
-  final override Node0Impl getValue() { none() }
-
   override Cpp::Location getLocation() { result = v.getLocation() }
 
-  final override SourceVariable getSourceVariable() {
+  final override NormalSourceVariable getSourceVariable() {
     result.getBaseVariable() = v and
     result.getIndirection() = 0
   }
-
-  final override BaseSourceVariable getBaseSourceVariable() { result = v }
 }
 
 private class DefVariableAddressImpl extends DefAddressImpl {
@@ -413,8 +510,17 @@ private class DirectDef extends DefImpl, TDirectDefImpl {
     isDef(_, _, address, result, _, indirectionIndex)
   }
 
-  override BaseSourceVariable getBaseSourceVariable() {
-    result = this.getBase().getBaseSourceVariable()
+  pragma[nomagic]
+  private predicate hasBaseSourceVariableAndIndirection(BaseSourceVariable v, int indirection) {
+    v = this.getBase().getBaseSourceVariable() and
+    indirection = this.getIndirection()
+  }
+
+  final override NormalSourceVariable getSourceVariable() {
+    exists(BaseSourceVariable v, int indirection |
+      sourceVariableHasBaseAndIndex(result, v, indirection) and
+      this.hasBaseSourceVariableAndIndirection(v, indirection)
+    )
   }
 
   override int getIndirection() { isDef(_, _, address, _, result, indirectionIndex) }
@@ -424,6 +530,32 @@ private class DirectDef extends DefImpl, TDirectDefImpl {
   override predicate isCertain() { isDef(true, _, address, _, _, indirectionIndex) }
 }
 
+/**
+ * A definition that "saves" the value of a variable before it is incremented
+ * or decremented.
+ */
+private class SavedPostfixCrementDefImpl extends DefImpl, TSavedPostfixCrementDefImpl {
+  SavedPostfixCrementSourceVariable sv;
+
+  SavedPostfixCrementDefImpl() { this = TSavedPostfixCrementDefImpl(sv, indirectionIndex) }
+
+  override Cpp::Location getLocation() { result = sv.getStoreInstruction().getLocation() }
+
+  final override predicate hasIndexInBlock(IRBlock block, int index) {
+    sv.getStoreInstruction() = block.getInstruction(index)
+  }
+
+  override string toString() { result = "Def of " + this.getSourceVariable() }
+
+  override SourceVariable getSourceVariable() { result = sv }
+
+  override int getIndirection() { result = sv.getIndirection() }
+
+  override predicate isCertain() {
+    isDef(true, _, sv.getStoreInstruction().getDestinationAddressOperand(), _, _, indirectionIndex)
+  }
+}
+
 private class DirectUseImpl extends UseImpl, TDirectUseImpl {
   Operand operand;
 
@@ -432,29 +564,22 @@ private class DirectUseImpl extends UseImpl, TDirectUseImpl {
   override string toString() { result = "Use of " + this.getSourceVariable() }
 
   final override predicate hasIndexInBlock(IRBlock block, int index) {
-    // See the comment in `ssa0`'s `OperandBasedUse` for an explanation of this
-    // predicate's implementation.
-    if this.getBase().getAst() = any(Cpp::PostfixCrementOperation c).getOperand()
-    then
-      exists(Operand op, int indirection, Instruction base |
-        indirection = this.getIndirection() and
-        base = this.getBase() and
-        op =
-          min(Operand cand, int i |
-            isUse(_, cand, base, indirection, indirectionIndex) and
-            block.getInstruction(i) = cand.getUse()
-          |
-            cand order by i
-          ) and
-        block.getInstruction(index) = op.getUse()
-      )
-    else operand.getUse() = block.getInstruction(index)
+    operand.getUse() = block.getInstruction(index)
   }
 
   private BaseSourceVariableInstruction getBase() { isUse(_, operand, result, _, indirectionIndex) }
 
-  override BaseSourceVariable getBaseSourceVariable() {
-    result = this.getBase().getBaseSourceVariable()
+  pragma[nomagic]
+  private predicate hasBaseSourceVariableAndIndirection(BaseSourceVariable bv, int indirection) {
+    this.getBase().getBaseSourceVariable() = bv and
+    this.getIndirection() = indirection
+  }
+
+  override NormalSourceVariable getSourceVariable() {
+    exists(BaseSourceVariable v, int indirection |
+      sourceVariableHasBaseAndIndex(result, v, indirection) and
+      this.hasBaseSourceVariableAndIndirection(v, indirection)
+    )
   }
 
   final Operand getOperand() { result = operand }
@@ -468,6 +593,34 @@ private class DirectUseImpl extends UseImpl, TDirectUseImpl {
   override Node getNode() { nodeHasOperand(result, operand, indirectionIndex) }
 }
 
+/**
+ * The use of the original "saved" variable after the variable has been incremented
+ * or decremented.
+ */
+private class SavedPostfixCrementUseImpl extends UseImpl, TSavedPostfixCrementUseImpl {
+  SavedPostfixCrementSourceVariable sv;
+
+  SavedPostfixCrementUseImpl() { this = TSavedPostfixCrementUseImpl(sv, indirectionIndex) }
+
+  override string toString() { result = "Use of " + this.getSourceVariable() }
+
+  final override predicate hasIndexInBlock(IRBlock block, int index) {
+    this.getOperand().getUse() = block.getInstruction(index)
+  }
+
+  override SourceVariable getSourceVariable() { result = sv }
+
+  final Operand getOperand() { result = sv.getOperand() }
+
+  final override Cpp::Location getLocation() { result = this.getOperand().getLocation() }
+
+  override int getIndirection() { result = sv.getIndirection() }
+
+  override predicate isCertain() { isUse(true, this.getOperand(), _, _, indirectionIndex) }
+
+  override Node getNode() { nodeHasOperand(result, this.getOperand(), indirectionIndex) }
+}
+
 pragma[nomagic]
 private predicate finalParameterNodeHasParameterAndIndex(
   FinalParameterNode n, Parameter p, int indirectionIndex
@@ -532,7 +685,18 @@ class FinalParameterUse extends UseImpl, TFinalParameterUse {
     result instanceof UnknownLocation
   }
 
-  override BaseIRVariable getBaseSourceVariable() { result.getIRVariable().getAst() = p }
+  pragma[nomagic]
+  private predicate hasBaseSourceVariableAndIndirection(BaseIRVariable v, int indirection) {
+    v.getIRVariable().getAst() = p and
+    indirection = this.getIndirection()
+  }
+
+  override NormalSourceVariable getSourceVariable() {
+    exists(BaseIRVariable v, int indirection |
+      sourceVariableHasBaseAndIndex(result, v, indirection) and
+      this.hasBaseSourceVariableAndIndirection(v, indirection)
+    )
+  }
 }
 
 /**
@@ -612,8 +776,17 @@ class GlobalUse extends UseImpl, TGlobalUse {
     hasReturnPosition(f, block, index)
   }
 
-  override BaseSourceVariable getBaseSourceVariable() {
-    baseSourceVariableIsGlobal(result, global, f)
+  pragma[nomagic]
+  private predicate hasBaseSourceVariableAndIndirection(BaseIRVariable v, int indirection) {
+    baseSourceVariableIsGlobal(v, global, f) and
+    indirection = this.getIndirection()
+  }
+
+  override NormalSourceVariable getSourceVariable() {
+    exists(BaseIRVariable v, int indirection |
+      sourceVariableHasBaseAndIndex(result, v, indirection) and
+      this.hasBaseSourceVariableAndIndirection(v, indirection)
+    )
   }
 
   final override Cpp::Location getLocation() { result = f.getLocation() }
@@ -658,15 +831,15 @@ class GlobalDefImpl extends DefImpl, TGlobalDefImpl {
     )
   }
 
-  /** Gets the global variable associated with this definition. */
-  override BaseSourceVariable getBaseSourceVariable() {
-    baseSourceVariableIsGlobal(result, global, f)
+  final override NormalSourceVariable getSourceVariable() {
+    exists(BaseSourceVariable v |
+      sourceVariableHasBaseAndIndex(result, v, indirectionIndex) and
+      baseSourceVariableIsGlobal(v, global, f)
+    )
   }
 
   override int getIndirection() { result = indirectionIndex }
 
-  override Node0Impl getValue() { none() }
-
   override predicate isCertain() { any() }
 
   /**
@@ -704,9 +877,15 @@ predicate defToNode(Node node, Definition def, SourceVariable sv) {
 }
 
 private predicate defToNode(Node node, Definition def) {
-  nodeHasOperand(node, def.getValue().asOperand(), def.getIndirectionIndex())
-  or
-  nodeHasInstruction(node, def.getValue().asInstruction(), def.getIndirectionIndex())
+  // Only definitions of `NormalSourceVariable` need to be converted into
+  // dataflow nodes. The other case, `SavedPostfixCrementSourceVariable`,
+  // are internal definitions that don't have a dataflow node representation.
+  def.getSourceVariable() instanceof NormalSourceVariable and
+  (
+    nodeHasOperand(node, def.getValue().asOperand(), def.getIndirectionIndex())
+    or
+    nodeHasInstruction(node, def.getValue().asInstruction(), def.getIndirectionIndex())
+  )
   or
   node.(InitialGlobalValue).getGlobalDef() = def
 }
@@ -940,6 +1119,16 @@ module SsaCached {
     SsaImpl::phiHasInputFromBlock(phi, inp, bb)
   }
 
+  cached
+  predicate uncertainWriteDefinitionInput(Definition uncertain, Definition inp) {
+    SsaImpl::uncertainWriteDefinitionInput(uncertain, inp)
+  }
+
+  cached
+  predicate ssaDefReachesEndOfBlock(IRBlock bb, Definition def) {
+    SsaImpl::ssaDefReachesEndOfBlock(bb, def, _)
+  }
+
   predicate variableRead = SsaInput::variableRead/4;
 
   predicate variableWrite = SsaInput::variableWrite/4;
@@ -1035,13 +1224,23 @@ class SynthNode extends DataFlowIntegrationImpl::SsaNode {
   SynthNode() { not this.asDefinition() instanceof SsaImpl::WriteDefinition }
 }
 
-signature predicate guardChecksNodeSig(IRGuards::IRGuardCondition g, Node e, boolean branch);
+private signature class ParamSig;
 
-signature predicate guardChecksNodeSig(
-  IRGuards::IRGuardCondition g, Node e, boolean branch, int indirectionIndex
-);
+private module ParamIntPair<ParamSig P> {
+  newtype TPair = MkPair(P p, int indirectionIndex) { nodeHasInstruction(_, _, indirectionIndex) }
+}
 
-module BarrierGuardWithIntParam<guardChecksNodeSig/4 guardChecksNode> {
+private module WithParam<ParamSig P> {
+  signature predicate guardChecksNodeSig(IRGuards::IRGuardCondition g, Node e, boolean gv, P param);
+}
+
+private module IntWithParam<ParamSig P> {
+  signature predicate guardChecksNodeSig(
+    IRGuards::IRGuardCondition g, Node e, boolean gv, int indirectionIndex, P param
+  );
+}
+
+module BarrierGuardWithIntParam<ParamSig P, IntWithParam<P>::guardChecksNodeSig/5 guardChecksNode> {
   private predicate ssaDefReachesCertainUse(Definition def, UseImpl use) {
     exists(SourceVariable v, IRBlock bb, int i |
       use.hasIndexInBlock(bb, i, v) and
@@ -1052,21 +1251,23 @@ module BarrierGuardWithIntParam<guardChecksNodeSig/4 guardChecksNode> {
 
   private predicate guardChecksInstr(
     IRGuards::Guards_v1::Guard g, IRGuards::GuardsInput::Expr instr, IRGuards::GuardValue gv,
-    int indirectionIndex
+    ParamIntPair<P>::TPair pair
   ) {
-    exists(Node node |
+    exists(Node node, int indirectionIndex, P p |
+      pair = ParamIntPair<P>::MkPair(p, indirectionIndex) and
       nodeHasInstruction(node, instr, indirectionIndex) and
-      guardChecksNode(g, node, gv.asBooleanValue(), indirectionIndex)
+      guardChecksNode(g, node, gv.asBooleanValue(), indirectionIndex, p)
     )
   }
 
   private predicate guardChecksWithWrappers(
     DataFlowIntegrationInput::Guard g, SsaImpl::Definition def, IRGuards::GuardValue val,
-    int indirectionIndex
+    ParamIntPair<P>::MkPair pair
   ) {
-    exists(Instruction e |
-      IRGuards::Guards_v1::ParameterizedValidationWrapper<int, guardChecksInstr/4>::guardChecks(g,
-        e, val, indirectionIndex)
+    exists(Instruction e, int indirectionIndex |
+      IRGuards::Guards_v1::ParameterizedValidationWrapper<ParamIntPair<P>::TPair, guardChecksInstr/4>::guardChecks(g,
+        e, val, pair) and
+      pair = ParamIntPair<P>::MkPair(_, indirectionIndex)
     |
       indirectionIndex = 0 and
       def.(Definition).getAUse().getDef() = e
@@ -1075,18 +1276,19 @@ module BarrierGuardWithIntParam<guardChecksNodeSig/4 guardChecksNode> {
     )
   }
 
-  Node getABarrierNode(int indirectionIndex) {
+  Node getABarrierNode(int indirectionIndex, P p) {
     // Only get the SynthNodes from the shared implementation, as the ExprNodes cannot
     // be matched on SourceVariable.
     result.(SsaSynthNode).getSynthNode() =
-      DataFlowIntegrationImpl::BarrierGuardDefWithState<int, guardChecksWithWrappers/4>::getABarrierNode(indirectionIndex)
+      DataFlowIntegrationImpl::BarrierGuardDefWithState<ParamIntPair<P>::MkPair, guardChecksWithWrappers/4>::getABarrierNode(ParamIntPair<P>::MkPair(p,
+          indirectionIndex))
     or
     // Calculate the guarded UseImpls corresponding to ExprNodes directly.
     exists(
       DataFlowIntegrationInput::Guard g, IRGuards::GuardValue branch, Definition def, IRBlock bb
     |
-      guardChecksWithWrappers(g, def, branch, indirectionIndex) and
       exists(UseImpl use |
+        guardChecksWithWrappers(g, def, branch, ParamIntPair<P>::MkPair(p, indirectionIndex)) and
         ssaDefReachesCertainUse(def, use) and
         use.getBlock() = bb and
         DataFlowIntegrationInput::guardControlsBlock(g, bb, branch) and
@@ -1096,15 +1298,16 @@ module BarrierGuardWithIntParam<guardChecksNodeSig/4 guardChecksNode> {
   }
 }
 
-module BarrierGuard<guardChecksNodeSig/3 guardChecksNode> {
+module BarrierGuard<ParamSig P, WithParam<P>::guardChecksNodeSig/4 guardChecksNode> {
   private predicate guardChecksNode(
-    IRGuards::IRGuardCondition g, Node e, boolean branch, int indirectionIndex
+    IRGuards::IRGuardCondition g, Node e, boolean gv, int indirectionIndex, P p
   ) {
-    guardChecksNode(g, e, branch) and indirectionIndex = 0
+    indirectionIndex = 0 and
+    guardChecksNode(g, e, gv, p)
   }
 
-  Node getABarrierNode() {
-    result = BarrierGuardWithIntParam<guardChecksNode/4>::getABarrierNode(0)
+  Node getABarrierNode(P p) {
+    result = BarrierGuardWithIntParam<P, guardChecksNode/5>::getABarrierNode(0, p)
   }
 }
 
@@ -1159,9 +1362,17 @@ class Definition extends SsaImpl::Definition {
   private Definition getAPhiInputOrPriorDefinition() {
     result = this.(PhiNode).getAnInput()
     or
-    SsaImpl::uncertainWriteDefinitionInput(this, result)
+    uncertainWriteDefinitionInput(this, result)
   }
 
+  /**
+   * Holds if this SSA definition is live at the end of basic block `bb`.
+   * That is, this definition reaches the end of basic block `bb`, at which
+   * point it is still live, without crossing another SSA definition of the
+   * same source variable.
+   */
+  predicate isLiveAtEndOfBlock(IRBlock bb) { ssaDefReachesEndOfBlock(bb, this) }
+
   /**
    * Gets a definition that ultimately defines this SSA definition and is
    * not itself a phi node.

cpp/ql/lib/semmle/code/cpp/ir/implementation/raw/internal/InstructionTag.qll

@@ -104,7 +104,11 @@ newtype TInstructionTag =
   } or
   SizeofVlaDimensionTag(int index) {
     exists(VlaDeclStmt v | exists(v.getTransitiveVlaDimensionStmt(index)))
-  }
+  } or
+  AssertionVarAddressTag() or
+  AssertionVarLoadTag() or
+  AssertionOpTag() or
+  AssertionBranchTag()
 
 class InstructionTag extends TInstructionTag {
   final string toString() { result = getInstructionTagId(this) }
@@ -296,4 +300,12 @@ string getInstructionTagId(TInstructionTag tag) {
   tag = CoAwaitBranchTag() and result = "CoAwaitBranch"
   or
   tag = BoolToIntConversionTag() and result = "BoolToIntConversion"
+  or
+  tag = AssertionVarAddressTag() and result = "AssertionVarAddress"
+  or
+  tag = AssertionVarLoadTag() and result = "AssertionVarLoad"
+  or
+  tag = AssertionOpTag() and result = "AssertionOp"
+  or
+  tag = AssertionBranchTag() and result = "AssertionBranch"
 }

cpp/ql/lib/semmle/code/cpp/ir/implementation/raw/internal/TranslatedAssertion.qll

@@ -0,0 +1,387 @@
+private import cpp
+private import semmle.code.cpp.ir.internal.IRUtilities
+private import semmle.code.cpp.ir.implementation.internal.OperandTag
+private import semmle.code.cpp.ir.internal.CppType
+private import semmle.code.cpp.ir.internal.TempVariableTag
+private import InstructionTag
+private import TranslatedElement
+private import TranslatedStmt
+private import TranslatedFunction
+
+/**
+ * Holds if `s` is a statement that may be an expanded assertion in a
+ * release build.
+ */
+pragma[nomagic]
+private predicate stmtCandidate(Stmt s) {
+  not s.isFromUninstantiatedTemplate(_) and
+  (
+    // The expansion of `__analysis_assume(x != 0);` when `__analysis_assume` is
+    // empty is the empty statement.
+    s instanceof EmptyStmt
+    or
+    // The expansion of `assert(x != 0)` when `assert` is `((void)0)` is a zero literal
+    // with a void type.
+    exists(Expr e |
+      e = s.(ExprStmt).getExpr() and
+      e.getValue() = "0" and
+      e.getActualType() instanceof VoidType
+    )
+  )
+}
+
+pragma[nomagic]
+private predicate macroInvocationLocation(int startline, Function f, MacroInvocation mi) {
+  mi.getMacroName() = ["assert", "__analysis_assume"] and
+  mi.getNumberOfArguments() = 1 and
+  mi.getLocation().hasLocationInfo(_, startline, _, _, _) and
+  f.getEntryPoint().isAffectedByMacro(mi)
+}
+
+pragma[nomagic]
+private predicate stmtParentLocation(int startline, Function f, StmtParent p) {
+  p.getEnclosingFunction() = f and
+  p.getLocation().hasLocationInfo(_, startline, _, _, _)
+}
+
+/**
+ * Holds if `mi` is a macro invocation with a name that is known
+ * to correspond to an assertion macro, and the macro invocation
+ * is the only thing on the line.
+ */
+pragma[nomagic]
+private predicate assertion0(MacroInvocation mi, Stmt s, string arg) {
+  stmtCandidate(s) and
+  s =
+    unique(StmtParent p, int startline, Function f |
+      macroInvocationLocation(startline, f, mi) and
+      stmtParentLocation(startline, f, p) and
+      // Also do not count the elements from the expanded macro, i.e., when checking
+      // if `assert(x)` is the only thing on the line we do not count the
+      // generated `((void)0)` expression.
+      not p = mi.getAnExpandedElement()
+    |
+      p
+    ) and
+  arg = mi.getUnexpandedArgument(0)
+}
+
+private Function getEnclosingFunctionForMacroInvocation(MacroInvocation mi) {
+  exists(Stmt s |
+    assertion0(mi, s, _) and
+    result = s.getEnclosingFunction()
+  )
+}
+
+/**
+ * Holds if `arg` has two components and the `i`'th component of the string
+ * `arg` is `s`, and the components are separated by an operation with
+ * opcode `opcode`.
+ */
+bindingset[arg]
+pragma[inline_late]
+private predicate parseArgument(string arg, string s, int i, Opcode opcode) {
+  s =
+    arg.regexpCapture("([a-zA-Z_][a-zA-Z_0-9]*|[0-9]+)\\s?<=\\s?([a-zA-Z_][a-zA-Z_0-9]*|[0-9]+)",
+      i + 1) and
+  opcode instanceof Opcode::CompareLE
+  or
+  s =
+    arg.regexpCapture("([a-zA-Z_][a-zA-Z_0-9]*|[0-9]+)\\s?>=\\s?([a-zA-Z_][a-zA-Z_0-9]*|[0-9]+)",
+      i + 1) and
+  opcode instanceof Opcode::CompareGE
+  or
+  s =
+    arg.regexpCapture("([a-zA-Z_][a-zA-Z_0-9]*|[0-9]+)\\s?<\\s?([a-zA-Z_][a-zA-Z_0-9]*|[0-9]+)",
+      i + 1) and
+  opcode instanceof Opcode::CompareLT
+  or
+  s =
+    arg.regexpCapture("([a-zA-Z_][a-zA-Z_0-9]*|[0-9]+)\\s?>\\s?([a-zA-Z_][a-zA-Z_0-9]*|[0-9]+)",
+      i + 1) and
+  opcode instanceof Opcode::CompareGT
+  or
+  s =
+    arg.regexpCapture("([a-zA-Z_][a-zA-Z_0-9]*|[0-9]+)\\s?!=\\s?([a-zA-Z_][a-zA-Z_0-9]*|[0-9]+)",
+      i + 1) and
+  opcode instanceof Opcode::CompareNE
+  or
+  s =
+    arg.regexpCapture("([a-zA-Z_][a-zA-Z_0-9]*|[0-9]+)\\s?==\\s?([a-zA-Z_][a-zA-Z_0-9]*|[0-9]+)",
+      i + 1) and
+  opcode instanceof Opcode::CompareEQ
+}
+
+private Element getAChildScope(Element scope) { result.getParentScope() = scope }
+
+private predicate hasAVariable(MacroInvocation mi, Stmt s, Element scope) {
+  assertion0(mi, s, _) and
+  s.getParent() = scope
+  or
+  hasAVariable(mi, s, getAChildScope(scope))
+}
+
+private LocalScopeVariable getVariable(MacroInvocation mi, int i) {
+  exists(string operand, string arg, Stmt s |
+    assertion0(mi, s, arg) and
+    parseArgument(arg, operand, i, _) and
+    result =
+      unique(Variable v |
+        v.getLocation().getStartLine() < s.getLocation().getStartLine() and
+        hasAVariable(mi, s, v.getParentScope()) and
+        v.hasName(operand)
+      |
+        v
+      )
+  )
+}
+
+/**
+ * Holds if the `i`'th component of the macro invocation `mi` with opcode
+ * `opcode` is a reference to `var`.
+ */
+private predicate hasVarAccessMacroArgument(MacroInvocation mi, Variable var, int i, Opcode opcode) {
+  exists(string arg, string s, Function f |
+    arg = mi.getUnexpandedArgument(0) and
+    f = getEnclosingFunctionForMacroInvocation(mi) and
+    parseArgument(arg, s, i, opcode) and
+    var = getVariable(mi, i)
+  )
+}
+
+/**
+ * Holds if the `i`'th component of the macro invocation `mi` with opcode
+ * `opcode` is a constant with the value `k`.
+ */
+private predicate hasConstMacroArgument(MacroInvocation mi, int k, int i, Opcode opcode) {
+  exists(string arg, string s |
+    assertion0(mi, _, arg) and
+    s.toInt() = k and
+    parseArgument(arg, s, i, opcode)
+  )
+}
+
+predicate hasAssertionOperand(MacroInvocation mi, int i) {
+  hasVarAccessMacroArgument(mi, _, i, _)
+  or
+  hasConstMacroArgument(mi, _, i, _)
+}
+
+private predicate hasAssertionOpcode(MacroInvocation mi, Opcode opcode) {
+  hasVarAccessMacroArgument(mi, _, _, opcode)
+  or
+  hasConstMacroArgument(mi, _, _, opcode)
+}
+
+/**
+ * Holds if `mi` is a macro invocation that is an assertion that should be generated
+ * in the control-flow graph at `s`.
+ */
+predicate assertion(MacroInvocation mi, Stmt s) {
+  assertion0(mi, s, _) and
+  hasAssertionOperand(mi, 0) and
+  hasAssertionOperand(mi, 1)
+}
+
+/** The translation of an operand of an assertion. */
+abstract private class TranslatedAssertionOperand extends TranslatedElement,
+  TTranslatedAssertionOperand
+{
+  MacroInvocation mi;
+  int index;
+
+  TranslatedAssertionOperand() { this = TTranslatedAssertionOperand(mi, index) }
+
+  MacroInvocation getMacroInvocation() { result = mi }
+
+  /**
+   * Gets the statement that is being replaced by the assertion that uses this
+   * operand.
+   */
+  Stmt getStmt() { assertion(mi, result) }
+
+  final override Locatable getAst() { result = this.getStmt() }
+
+  final override TranslatedElement getChild(int id) { none() }
+
+  final override Declaration getFunction() { result = this.getStmt().getEnclosingFunction() }
+
+  /** Gets the instruction which holds the result of this operand. */
+  abstract Instruction getResult();
+
+  final override string toString() { result = "Operand of assertion: " + mi }
+
+  /** Gets the index of this operand (i.e., `0` or `1`). */
+  final int getIndex() { result = index }
+}
+
+/** An operand of an assertion that is a variable access. */
+class TranslatedAssertionVarAccess extends TranslatedAssertionOperand {
+  TranslatedAssertionVarAccess() { hasVarAccessMacroArgument(mi, _, index, _) }
+
+  Variable getVariable() { hasVarAccessMacroArgument(mi, result, index, _) }
+
+  final override IRUserVariable getInstructionVariable(InstructionTag tag) {
+    tag = AssertionVarAddressTag() and
+    result.getVariable() = this.getVariable()
+  }
+
+  final override Instruction getFirstInstruction(EdgeKind kind) {
+    result = this.getInstruction(AssertionVarAddressTag()) and kind instanceof GotoEdge
+  }
+
+  final override Instruction getInstructionSuccessorInternal(InstructionTag tag, EdgeKind kind) {
+    tag = AssertionVarAddressTag() and
+    kind instanceof GotoEdge and
+    result = this.getInstruction(AssertionVarLoadTag())
+    or
+    tag = AssertionVarLoadTag() and
+    result = getTranslatedAssertionMacroInvocation(mi).getChildSuccessor(this, kind)
+  }
+
+  final override Instruction getALastInstructionInternal() {
+    result = this.getInstruction(AssertionVarLoadTag())
+  }
+
+  override predicate hasInstruction(Opcode opcode, InstructionTag tag, CppType resultType) {
+    exists(Variable v | v = this.getVariable() |
+      opcode instanceof Opcode::VariableAddress and
+      tag = AssertionVarAddressTag() and
+      resultType = getTypeForGLValue(v.getType())
+      or
+      opcode instanceof Opcode::Load and
+      tag = AssertionVarLoadTag() and
+      resultType = getTypeForPRValue(v.getType())
+    )
+  }
+
+  final override Instruction getInstructionRegisterOperand(InstructionTag tag, OperandTag operandTag) {
+    tag = AssertionVarLoadTag() and
+    operandTag instanceof AddressOperandTag and
+    result = this.getInstruction(AssertionVarAddressTag())
+  }
+
+  final override Instruction getResult() { result = this.getInstruction(AssertionVarLoadTag()) }
+}
+
+/** An operand of an assertion that is a constant access. */
+private class TranslatedAssertionConst extends TranslatedAssertionOperand {
+  TranslatedAssertionConst() { hasConstMacroArgument(mi, _, index, _) }
+
+  int getConst() { hasConstMacroArgument(mi, result, index, _) }
+
+  final override string getInstructionConstantValue(InstructionTag tag) {
+    tag = OnlyInstructionTag() and
+    result = this.getConst().toString()
+  }
+
+  final override Instruction getFirstInstruction(EdgeKind kind) {
+    result = this.getInstruction(OnlyInstructionTag()) and
+    kind instanceof GotoEdge
+  }
+
+  final override Instruction getInstructionSuccessorInternal(InstructionTag tag, EdgeKind kind) {
+    tag = OnlyInstructionTag() and
+    result = getTranslatedAssertionMacroInvocation(mi).getChildSuccessor(this, kind)
+  }
+
+  final override Instruction getALastInstructionInternal() {
+    result = this.getInstruction(OnlyInstructionTag())
+  }
+
+  override predicate hasInstruction(Opcode opcode, InstructionTag tag, CppType resultType) {
+    opcode instanceof Opcode::Constant and
+    tag = OnlyInstructionTag() and
+    resultType = getIntType()
+  }
+
+  final override Instruction getResult() { result = this.getInstruction(OnlyInstructionTag()) }
+}
+
+/**
+ * Gets the `TranslatedAssertionMacroInvocation` corresponding to the macro
+ * invocation `mi`.
+ */
+TranslatedAssertionMacroInvocation getTranslatedAssertionMacroInvocation(MacroInvocation mi) {
+  result.getMacroInvocation() = mi
+}
+
+/**
+ * A synthesized assertion which would have otherwise been invisible because the
+ * database represents a release build where assertions are disabled.
+ */
+private class TranslatedAssertionMacroInvocation extends TranslatedStmt {
+  MacroInvocation mi;
+
+  TranslatedAssertionMacroInvocation() { assertion(mi, stmt) }
+
+  final override Instruction getFirstInstruction(EdgeKind kind) {
+    result = this.getLeft().getFirstInstruction(kind)
+  }
+
+  TranslatedAssertionOperand getLeft() {
+    result.getMacroInvocation() = mi and
+    result.getIndex() = 0
+  }
+
+  TranslatedAssertionOperand getRight() {
+    result.getMacroInvocation() = mi and
+    result.getIndex() = 1
+  }
+
+  final override Instruction getInstructionSuccessorInternal(InstructionTag tag, EdgeKind kind) {
+    tag = AssertionOpTag() and
+    kind instanceof GotoEdge and
+    result = this.getInstruction(AssertionBranchTag())
+    or
+    tag = AssertionBranchTag() and
+    kind instanceof TrueEdge and
+    result = this.getParent().getChildSuccessor(this, _)
+  }
+
+  final override TranslatedElement getChildInternal(int id) {
+    id = 0 and result = this.getLeft()
+    or
+    id = 1 and result = this.getRight()
+  }
+
+  final override Instruction getALastInstructionInternal() {
+    result = this.getInstruction(AssertionBranchTag())
+  }
+
+  final override predicate hasInstruction(Opcode opcode, InstructionTag tag, CppType resultType) {
+    tag = AssertionOpTag() and
+    resultType = getBoolType() and
+    hasAssertionOpcode(mi, opcode)
+    or
+    tag = AssertionBranchTag() and
+    resultType = getVoidType() and
+    opcode instanceof Opcode::ConditionalBranch
+  }
+
+  final override Instruction getChildSuccessorInternal(TranslatedElement child, EdgeKind kind) {
+    child = this.getLeft() and
+    result = this.getRight().getFirstInstruction(kind)
+    or
+    child = this.getRight() and
+    kind instanceof GotoEdge and
+    result = this.getInstruction(AssertionOpTag())
+  }
+
+  final override Instruction getInstructionRegisterOperand(InstructionTag tag, OperandTag operandTag) {
+    tag = AssertionOpTag() and
+    (
+      operandTag instanceof LeftOperandTag and
+      result = this.getLeft().getResult()
+      or
+      operandTag instanceof RightOperandTag and
+      result = this.getRight().getResult()
+    )
+    or
+    tag = AssertionBranchTag() and
+    operandTag instanceof ConditionOperandTag and
+    result = this.getInstruction(AssertionOpTag())
+  }
+
+  MacroInvocation getMacroInvocation() { result = mi }
+}

cpp/ql/lib/semmle/code/cpp/ir/implementation/raw/internal/TranslatedElement.qll

@@ -12,6 +12,7 @@ private import TranslatedFunction
 private import TranslatedStmt
 private import TranslatedExpr
 private import IRConstruction
+private import TranslatedAssertion
 private import semmle.code.cpp.models.interfaces.SideEffect
 private import SideEffects
 
@@ -138,6 +139,14 @@ private predicate ignoreExprAndDescendants(Expr expr) {
   // conditionally constructed (until we have a mechanism for calling these only when the
   // temporary's constructor was run)
   isConditionalTemporaryDestructorCall(expr)
+  or
+  // An assertion in a release build is often defined as `#define assert(x) ((void)0)`.
+  // We generate a synthetic assertion in release builds, and when we do that the
+  // expression `((void)0)` should not be translated.
+  exists(MacroInvocation mi |
+    assertion(mi, _) and
+    expr = mi.getExpr().getFullyConverted()
+  )
 }
 
 /**
@@ -909,7 +918,8 @@ newtype TTranslatedElement =
   } or
   // The side effect that initializes newly-allocated memory.
   TTranslatedAllocationSideEffect(AllocationExpr expr) { not ignoreSideEffects(expr) } or
-  TTranslatedStaticStorageDurationVarInit(Variable var) { Raw::varHasIRFunc(var) }
+  TTranslatedStaticStorageDurationVarInit(Variable var) { Raw::varHasIRFunc(var) } or
+  TTranslatedAssertionOperand(MacroInvocation mi, int index) { hasAssertionOperand(mi, index) }
 
 /**
  * Gets the index of the first explicitly initialized element in `initList`

cpp/ql/lib/semmle/code/cpp/ir/implementation/raw/internal/TranslatedStmt.qll

@@ -10,6 +10,7 @@ private import TranslatedElement
 private import TranslatedExpr
 private import TranslatedFunction
 private import TranslatedInitialization
+private import TranslatedAssertion
 
 TranslatedStmt getTranslatedStmt(Stmt stmt) { result.getAst() = stmt }
 
@@ -324,8 +325,16 @@ abstract class TranslatedStmt extends TranslatedElement, TTranslatedStmt {
 
 class TranslatedEmptyStmt extends TranslatedStmt {
   TranslatedEmptyStmt() {
-    stmt instanceof EmptyStmt or
-    stmt instanceof LabelStmt or
+    // An assertion macro invocation can expand to
+    // an empty statement in release builds. In that case
+    // we synthesize the check that would have occurred.
+    // This is handled by `TranslatedAssertion.qll` and so
+    // we exclude these statements here.
+    not assertion(_, stmt) and
+    stmt instanceof EmptyStmt
+    or
+    stmt instanceof LabelStmt
+    or
     stmt instanceof SwitchCase
   }
 
@@ -381,7 +390,7 @@ class TranslatedDeclStmt extends TranslatedStmt {
 
   override TranslatedElement getLastChild() { result = this.getChild(this.getChildCount() - 1) }
 
-  private int getChildCount() { result = count(this.getDeclarationEntry(_)) }
+  private int getChildCount() { result = count(int i | exists(this.getDeclarationEntry(i))) }
 
   IRDeclarationEntry getIRDeclarationEntry(int index) {
     result.hasIndex(index) and
@@ -420,6 +429,15 @@ class TranslatedDeclStmt extends TranslatedStmt {
 class TranslatedExprStmt extends TranslatedStmt {
   override ExprStmt stmt;
 
+  TranslatedExprStmt() {
+    // An assertion macro invocation typically expand to the
+    // expression `((void)0)` in release builds. In that case
+    // we synthesize the check that would have occurred.
+    // This is handled by `TranslatedAssertion.qll` and so
+    // we exclude these statements here.
+    not assertion(_, stmt)
+  }
+
   TranslatedExpr getExpr() { result = getTranslatedExpr(stmt.getExpr().getFullyConverted()) }
 
   override TranslatedElement getChildInternal(int id) { id = 0 and result = this.getExpr() }

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

@@ -57,3 +57,4 @@ private import implementations.CAtlFile
 private import implementations.CAtlFileMapping
 private import implementations.CAtlTemporaryFile
 private import implementations.CRegKey
+private import implementations.WinHttp

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

@@ -16,17 +16,3 @@ private class MySqlExecutionFunction extends SqlExecutionFunction {
 
   override predicate hasSqlArgument(FunctionInput input) { input.isParameterDeref(1) }
 }
-
-/**
- * The `mysql_real_escape_string` family of functions from the MySQL C API.
- */
-private class MySqlBarrierFunction extends SqlBarrierFunction {
-  MySqlBarrierFunction() {
-    this.hasName(["mysql_real_escape_string", "mysql_real_escape_string_quote"])
-  }
-
-  override predicate barrierSqlArgument(FunctionInput input, FunctionOutput output) {
-    input.isParameterDeref(2) and
-    output.isParameterDeref(1)
-  }
-}

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

@@ -0,0 +1,50 @@
+private import cpp
+private import semmle.code.cpp.ir.dataflow.FlowSteps
+private import semmle.code.cpp.dataflow.new.DataFlow
+
+/** The `WINHTTP_HEADER_NAME` class from `winhttp.h`. */
+class WinHttpHeaderName extends Class {
+  WinHttpHeaderName() { this.hasGlobalName("_WINHTTP_HEADER_NAME") }
+}
+
+/** The `WINHTTP_EXTENDED_HEADER` class from `winhttp.h`. */
+class WinHttpExtendedHeader extends Class {
+  WinHttpExtendedHeader() { this.hasGlobalName("_WINHTTP_EXTENDED_HEADER") }
+}
+
+private class WinHttpHeaderNameInheritingContent extends TaintInheritingContent,
+  DataFlow::FieldContent
+{
+  WinHttpHeaderNameInheritingContent() {
+    this.getIndirectionIndex() = 2 and
+    (
+      this.getAField().getDeclaringType() instanceof WinHttpHeaderName
+      or
+      // The extended header looks like:
+      // struct WINHTTP_EXTENDED_HEADER {
+      //   union { [...] };
+      //   union { [...] };
+      // };
+      // So the first declaring type is the anonymous unions, and the declaring
+      // type of those anonymous unions is the `WINHTTP_EXTENDED_HEADER` struct.
+      this.getAField().getDeclaringType().getDeclaringType() instanceof WinHttpExtendedHeader
+    )
+  }
+}
+
+/** The `URL_COMPONENTS` class from `winhttp.h`. */
+class WinHttpUrlComponents extends Class {
+  WinHttpUrlComponents() { this.hasGlobalName("_WINHTTP_URL_COMPONENTS") }
+}
+
+private class WinHttpUrlComponentsInheritingContent extends TaintInheritingContent,
+  DataFlow::FieldContent
+{
+  WinHttpUrlComponentsInheritingContent() {
+    exists(Field f | f = this.getField() and f.getDeclaringType() instanceof WinHttpUrlComponents |
+      if f.getType().getUnspecifiedType() instanceof PointerType
+      then this.getIndirectionIndex() = 2
+      else this.getIndirectionIndex() = 1
+    )
+  }
+}

cpp/ql/lib/semmle/code/cpp/rangeanalysis/RangeAnalysisUtils.qll

@@ -404,7 +404,7 @@ predicate cmpWithLinearBound(
  * For example, if `t` is a signed 32-bit type then holds if `lb` is
  * `-2^31` and `ub` is `2^31 - 1`.
  */
-private predicate typeBounds(ArithmeticType t, float lb, float ub) {
+private predicate typeBounds0(ArithmeticType t, float lb, float ub) {
   exists(IntegralType integralType, float limit |
     integralType = t and limit = 2.pow(8 * integralType.getSize())
   |
@@ -423,6 +423,42 @@ private predicate typeBounds(ArithmeticType t, float lb, float ub) {
   t instanceof FloatingPointType and lb = -(1.0 / 0.0) and ub = 1.0 / 0.0
 }
 
+/**
+ * Gets the underlying type for an enumeration `e`.
+ *
+ * If the enumeration does not have an explicit type we approximate it using
+ * the following rules:
+ * - The result type is always `signed`, and
+ * - if the largest value fits in an `int` the result is `int`. Otherwise, the
+ * result is `long`.
+ */
+private IntegralType getUnderlyingTypeForEnum(Enum e) {
+  result = e.getExplicitUnderlyingType()
+  or
+  not e.hasExplicitUnderlyingType() and
+  result.isSigned() and
+  exists(IntType intType |
+    if max(e.getAnEnumConstant().getValue().toFloat()) >= 2.pow(8 * intType.getSize() - 1)
+    then result instanceof LongType
+    else result = intType
+  )
+}
+
+/**
+ * Holds if `lb` and `ub` are the lower and upper bounds of the unspecified
+ * type `t`.
+ *
+ * For example, if `t` is a signed 32-bit type then holds if `lb` is
+ * `-2^31` and `ub` is `2^31 - 1`.
+ *
+ * Unlike `typeBounds0`, this predicate also handles `Enum` types.
+ */
+private predicate typeBounds(Type t, float lb, float ub) {
+  typeBounds0(t, lb, ub)
+  or
+  typeBounds0(getUnderlyingTypeForEnum(t), lb, ub)
+}
+
 private Type stripReference(Type t) {
   if t instanceof ReferenceType then result = t.(ReferenceType).getBaseType() else result = t
 }

cpp/ql/lib/semmle/code/cpp/rangeanalysis/SimpleRangeAnalysis.qll

@@ -512,8 +512,8 @@ private module BoundsEstimate {
    */
   float getBoundsLimit() {
     // This limit is arbitrary, but low enough that it prevents timeouts on
-    // specific observed customer databases (and the in the tests).
-    result = 2.0.pow(40)
+    // specific observed customer databases (and in the tests).
+    result = 2.0.pow(29)
   }
 
   /** Gets the maximum number of bounds possible for `t` when widening is used. */
@@ -552,34 +552,47 @@ private module BoundsEstimate {
   private float nrOfBoundsPhiGuard(RangeSsaDefinition def, StackVariable v) {
     // If we have
     //
+    //   if (x < c) { e1 } else { e2 }
+    //   e3
+    //
+    // then `{ e1 }` and `{ e2 }` are both guard phi nodes guarded by `x < c`.
+    // The range analysis propagates bounds on `x` into both branches, filtered
+    // by the condition. In this case all lower bounds flow to `{ e1 }` and only
+    // lower bounds that are smaller than `c` flow to `{ e2 }`.
+    //
+    // The largest number of bounds possible for `e3` is the number of bounds on `x` plus
+    // one. This happens when all bounds flow from `x` to `e1` to `e3` and the
+    // bound `c` can flow to `e2` to `e3`.
+    //
+    // We want to optimize our bounds estimate for `e3`, as that is the estimate
+    // that can continue propagating forward. We don't know how the existing
+    // bounds will be split between the different branches. That depends on
+    // whether the range analysis is tracking lower bounds or upper bounds, and
+    // on the meaning of the condition.
+    //
+    // As a heuristic we divide the number of bounds on `x` by 2 to "average"
+    // the effect of the condition and add 1 to account for the bound from the
+    // condition itself. This will approximate estimates inside the branches,
+    // but will give a good estimate after the branches are merged.
+    //
+    // This also handles cases such as this one
+    //
     //   if (x < c) { e1 }
-    //   e2
+    //   e3
     //
-    // then `e2` is both a guard phi node (guarded by `x < c`) and a normal
-    // phi node (control is merged after the `if` statement).
-    //
-    // Assume `x` has `n` bounds. Then `n` bounds are propagated to the guard
-    // phi node `{ e1 }` and, since `{ e1 }` is input to `e2` as a normal phi
-    // node, `n` bounds are propagated to `e2`. If we also propagate the `n`
-    // bounds to `e2` as a guard phi node, then we square the number of
-    // bounds.
-    //
-    // However in practice `x < c` is going to cut down the number of bounds:
-    // The tracked bounds can't flow to both branches as that would require
-    // them to simultaneously be greater and smaller than `c`. To approximate
-    // this better, the contribution from a guard phi node that is also a
-    // normal phi node is 1.
-    exists(def.getAPhiInput(v)) and
-    isGuardPhiWithBound(def, v, _) and
-    result = 1
-    or
-    not exists(def.getAPhiInput(v)) and
-    // If there's different `access`es, then they refer to the same variable
-    // with the same lower bounds. Hence adding these guards make no sense (the
-    // implementation will take the union, but they'll be removed by
-    // deduplication). Hence we use `max` as an approximation.
-    result =
-      max(VariableAccess access | isGuardPhiWithBound(def, v, access) | nrOfBoundsExpr(access))
+    // where `e3` is both a guard phi node (guarded by `x < c`) and a normal
+    // phi node (control is merged after the `if` statement). Here half of the
+    // bounds flow into the branch and then to `e3` as a normal phi node and the
+    // "other" half flow from the condition to `e3` as a guard phi node.
+    exists(float varBounds |
+      // If there's different `access`es, then they refer to the same
+      // variable with the same lower bounds. Hence adding these guards makes no
+      // sense (the implementation will take the union, but they'll be removed by
+      // deduplication). Hence we use `max` as an approximation.
+      varBounds =
+        max(VariableAccess access | isGuardPhiWithBound(def, v, access) | nrOfBoundsExpr(access)) and
+      result = (varBounds + 1) / 2
+    )
     or
     def.isPhiNode(v) and
     not isGuardPhiWithBound(def, v, _) and
@@ -2180,6 +2193,16 @@ module SimpleRangeAnalysisInternal {
 
   /** Gets the estimate of the number of bounds for `e`. */
   float estimateNrOfBounds(Expr e) { result = BoundsEstimate::nrOfBoundsExpr(e) }
+
+  /** Counts the numbers of lower bounds that are computed internally for `e`. */
+  float countNrOfLowerBounds(Expr e) {
+    result = strictcount(float lb | lb = getLowerBoundsImpl(e) | lb)
+  }
+
+  /** Counts the numbers of upper bounds that are computed internally for `e`. */
+  float countNrOfUpperBounds(Expr e) {
+    result = strictcount(float ub | ub = getUpperBoundsImpl(e) | ub)
+  }
 }
 
 /** Provides predicates for debugging the simple range analysis library. */
@@ -2208,7 +2231,7 @@ private module Debug {
    */
   predicate countGetLowerBoundsImpl(Expr e, int n) {
     e = getRelevantLocatable() and
-    n = strictcount(float lb | lb = getLowerBoundsImpl(e) | lb)
+    n = SimpleRangeAnalysisInternal::countNrOfLowerBounds(e)
   }
 
   float debugNrOfBounds(Expr e) {

cpp/ql/lib/semmle/code/cpp/stmts/Stmt.qll

@@ -20,7 +20,7 @@ class Stmt extends StmtParent, @stmt {
   predicate hasChild(Element e, int n) { this.getChild(n) = e }
 
   /** Gets the enclosing function of this statement, if any. */
-  Function getEnclosingFunction() { result = stmtEnclosingElement(this) }
+  override Function getEnclosingFunction() { result = stmtEnclosingElement(this) }
 
   /**
    * Gets the nearest enclosing block of this statement in the source, if any.
@@ -159,7 +159,10 @@ private class TStmtParent = @stmt or @expr;
  *
  * This is normally a statement, but may be a `StmtExpr`.
  */
-class StmtParent extends ControlFlowNode, TStmtParent { }
+class StmtParent extends ControlFlowNode, TStmtParent {
+  /** Gets the enclosing function of this element, if any. */
+  Function getEnclosingFunction() { none() }
+}
 
 /**
  * A C/C++ 'expression' statement.

cpp/ql/lib/semmlecode.cpp.dbscheme

@@ -236,6 +236,34 @@ extractor_version(
     string frontend_version: string ref
 )
 
+/**
+ * Gives the TRAP filename that `trap` is associated with.
+ * For debugging only.
+ */
+trap_filename(
+    int trap: @trap,
+    string filename: string ref
+);
+
+/**
+ * In `build-mode: none` overlay mode, indicates that `source_file`
+ * (`/path/to/foo.c`) uses the TRAP file `trap_file`; i.e. it is the
+ * TRAP file corresponding to `foo.c`, something it transitively
+ * includes, or a template instantiation it transitively uses.
+ */
+source_file_uses_trap(
+    string source_file: string ref,
+    int trap_file: @trap ref
+);
+
+/**
+ * Holds if there is a definition of `element` in TRAP file `trap_file`.
+ */
+in_trap(
+    int element: @element ref,
+    int trap_file: @trap ref
+);
+
 pch_uses(
     int pch: @pch ref,
     int compilation: @compilation ref,
@@ -2353,6 +2381,7 @@ case @preprocdirect.kind of
 | 14 = @ppd_ms_import
 | 15 = @ppd_elifdef
 | 16 = @ppd_elifndef
+| 17 = @ppd_embed
 | 18 = @ppd_warning
 ;
 
@@ -2379,6 +2408,11 @@ includes(
     int included: @file ref
 );
 
+embeds(
+    unique int id: @ppd_embed ref,
+    int included: @file ref
+);
+
 link_targets(
     int id: @link_target,
     int binary: @file ref
@@ -2389,6 +2423,8 @@ link_parent(
     int link_target : @link_target ref
 );
 
+/*- Database metadata -*/
+
 /**
  * The CLI will automatically emit applicable tuples for this table,
  * such as `databaseMetadata("isOverlay", "true")` when building an
@@ -2399,6 +2435,8 @@ databaseMetadata(
   string value: string ref
 );
 
+/*- Overlay support -*/
+
 /**
  * The CLI will automatically emit tuples for each new/modified/deleted file
  * when building an overlay database.

cpp/ql/lib/upgrades/1402ab319d20cdc9289deb7bfc1c70f36be44d44/upgrade.properties

@@ -0,0 +1,2 @@
+description: Sections for databaseMetadata and overlayChangedFiles
+compatibility: full

cpp/ql/lib/upgrades/83100310bf73eefc37c1d8d0ac98b2ca3019c7b6/upgrade.properties

@@ -0,0 +1,2 @@
+description: Support embed preprocessor directive
+compatibility: partial

cpp/ql/lib/upgrades/9439176c1d1312787926458dd54d65a849069118/upgrade.properties

@@ -0,0 +1,2 @@
+description: Add trap_filename, source_file_uses_trap and in_trap relations
+compatibility: full

cpp/ql/src/CHANGELOG.md

@@ -1,3 +1,15 @@
+## 1.5.12
+
+No user-facing changes.
+
+## 1.5.11
+
+No user-facing changes.
+
+## 1.5.10
+
+No user-facing changes.
+
 ## 1.5.9
 
 ### Minor Analysis Improvements

cpp/ql/src/Likely Bugs/Leap Year/LeapYear.qll

@@ -308,3 +308,37 @@ private module PossibleYearArithmeticOperationCheckConfig implements DataFlow::C
 
 module PossibleYearArithmeticOperationCheckFlow =
   TaintTracking::Global<PossibleYearArithmeticOperationCheckConfig>;
+
+/**
+ * A time conversion function where either
+ * 1) an incorrect leap year date would result in an error that can be checked from the return value or
+ * 2) an incorrect leap year date is auto corrected (no checks required)
+ */
+class TimeConversionFunction extends Function {
+  boolean autoLeapYearCorrecting;
+
+  TimeConversionFunction() {
+    autoLeapYearCorrecting = false and
+    (
+      this.getName() =
+        [
+          "FileTimeToSystemTime", "SystemTimeToFileTime", "SystemTimeToTzSpecificLocalTime",
+          "SystemTimeToTzSpecificLocalTimeEx", "TzSpecificLocalTimeToSystemTime",
+          "TzSpecificLocalTimeToSystemTimeEx", "RtlLocalTimeToSystemTime",
+          "RtlTimeToSecondsSince1970", "_mkgmtime", "SetSystemTime", "VarUdateFromDate", "from_tm"
+        ]
+      or
+      // Matches all forms of GetDateFormat, e.g. GetDateFormatA/W/Ex
+      this.getName().matches("GetDateFormat%")
+    )
+    or
+    autoLeapYearCorrecting = true and
+    this.getName() =
+      ["mktime", "_mktime32", "_mktime64", "SystemTimeToVariantTime", "VariantTimeToSystemTime"]
+  }
+
+  /**
+   * Holds if the function is expected to auto convert a bad leap year date.
+   */
+  predicate isAutoLeapYearCorrecting() { autoLeapYearCorrecting = true }
+}

cpp/ql/src/Likely Bugs/Leap Year/UncheckedLeapYearAfterYearModification.ql

@@ -1,7 +1,7 @@
 /**
  * @name Year field changed using an arithmetic operation without checking for leap year
  * @description A field that represents a year is being modified by an arithmetic operation, but no proper check for leap years can be detected afterwards.
- * @kind problem
+ * @kind path-problem
  * @problem.severity warning
  * @id cpp/leap-year/unchecked-after-arithmetic-year-modification
  * @precision medium
@@ -11,49 +11,844 @@
 
 import cpp
 import LeapYear
+import semmle.code.cpp.controlflow.IRGuards
 
-from Variable var, LeapYearFieldAccess yfa
-where
-  exists(VariableAccess va |
-    yfa.getQualifier() = va and
-    var.getAnAccess() = va and
-    // The year is modified with an arithmetic operation. Avoid values that are likely false positives
-    yfa.isModifiedByArithmeticOperationNotForNormalization() and
-    // Avoid false positives
-    not (
-      // If there is a local check for leap year after the modification
-      exists(LeapYearFieldAccess yfacheck |
-        yfacheck.getQualifier() = var.getAnAccess() and
-        yfacheck.isUsedInCorrectLeapYearCheck() and
-        yfacheck.getBasicBlock() = yfa.getBasicBlock().getASuccessor*()
-      )
+/**
+ * Functions whose operations should never be considered a
+ * source or sink of a dangerous leap year operation.
+ * The general concept is to add conversion functions
+ * that convert one time type to another. Often
+ * other ignorable operation heuristics will filter these,
+ * but some cases, the simplest approach is to simply filter
+ * the function entirely.
+ * Note that flow through these functions should still be allowed
+ * we just cannot start or end flow from an operation to a
+ * year assignment in one of these functions.
+ */
+class IgnorableFunction extends Function {
+  IgnorableFunction() {
+    // arithmetic in known time conversion functions may look like dangerous operations
+    // we assume all known time conversion functions are safe.
+    this instanceof TimeConversionFunction
+    or
+    // Helper utility in postgres with string time conversions
+    this.getName() = "DecodeISO8601Interval"
+    or
+    // helper utility for date conversions in qtbase
+    this.getName() = "adjacentDay"
+    or
+    // Windows API function that does timezone conversions
+    this.getName().matches("%SystemTimeToTzSpecificLocalTime%")
+    or
+    // Windows APIs that do time conversions
+    this.getName().matches("%localtime%\\_s%")
+    or
+    // Windows APIs that do time conversions
+    this.getName().matches("%SpecificLocalTimeToSystemTime%")
+    or
+    // postgres function for diffing timestamps, date for leap year
+    // is not applicable.
+    this.getName().toLowerCase().matches("%timestamp%age%")
+    or
+    // Reading byte streams often involves operations of some base, but that's
+    // not a real source of leap year issues.
+    this.getName().toLowerCase().matches("%read%bytes%")
+    or
+    // A postgres function for local time conversions
+    // conversion operations (from one time structure to another) are generally ignorable
+    this.getName() = "localsub"
+    or
+    // Indication of a calendar not applicable to
+    // gregorian leap year, e.g., Hijri, Persian, Hebrew
+    this.getName().toLowerCase().matches("%hijri%")
+    or
+    this.getFile().getBaseName().toLowerCase().matches("%hijri%")
+    or
+    this.getName().toLowerCase().matches("%persian%")
+    or
+    this.getFile().getBaseName().toLowerCase().matches("%persian%")
+    or
+    this.getName().toLowerCase().matches("%hebrew%")
+    or
+    this.getFile().getBaseName().toLowerCase().matches("%hebrew%")
+    or
+    // misc. from string/char converters heuristic
+    this.getName()
+        .toLowerCase()
+        .matches(["%char%to%", "%string%to%", "%from%char%", "%from%string%"])
+    or
+    // boost's gregorian.cpp has year manipulations that are checked in complex ways.
+    // ignore the entire file as a source or sink.
+    this.getFile().getAbsolutePath().toLowerCase().matches("%boost%gregorian.cpp%")
+  }
+}
+
+/**
+ * The set of expressions which are ignorable; either because they seem to not be part of a year mutation,
+ * or because they seem to be a conversion pattern of mapping date scalars.
+ */
+abstract class IgnorableOperation extends Expr { }
+
+class IgnorableExprRem extends IgnorableOperation instanceof RemExpr { }
+
+/**
+ * An operation with 10, 100, 1000, 10000 as an operand is often a sign of conversion
+ * or atoi.
+ */
+class IgnorableExpr10MultipleComponent extends IgnorableOperation {
+  IgnorableExpr10MultipleComponent() {
+    this.(Operation).getAnOperand().getValue().toInt() in [10, 100, 1000, 10000]
+    or
+    exists(AssignOperation a | a.getRValue() = this |
+      a.getRValue().getValue().toInt() in [10, 100, 1000, 10000]
+    )
+  }
+}
+
+/**
+ * An operation involving a sub expression with char literal `48`, ignore as a likely string conversion. For example: `X - '0'`
+ */
+class IgnorableExpr48Mapping extends IgnorableOperation {
+  IgnorableExpr48Mapping() {
+    this.(SubExpr).getRightOperand().getValue().toInt() = 48
+    or
+    exists(AssignSubExpr e | e.getRValue() = this | e.getRValue().getValue().toInt() = 48)
+  }
+}
+
+/**
+ * A binary or arithmetic operation whereby one of the components is textual or a string.
+ */
+class IgnorableCharLiteralArithmetic extends IgnorableOperation {
+  IgnorableCharLiteralArithmetic() {
+    this.(BinaryArithmeticOperation).getAnOperand() instanceof TextLiteral
+    or
+    this instanceof TextLiteral and
+    any(AssignArithmeticOperation arith).getRValue() = this
+  }
+}
+
+/**
+ * Constants often used in date conversions (from one date data type to another)
+ * Numerous examples exist, like 1900 or 2000 that convert years from one
+ * representation to another.
+ * Also '0' is sometimes observed as an atoi style conversion.
+ */
+bindingset[c]
+predicate isLikelyConversionConstant(int c) {
+  exists(int i | i = c.abs() |
+    i =
+      [
+        146097, // days in 400-year Gregorian cycle
+        36524, // days in 100-year Gregorian subcycle
+        1461, // days in 4-year cycle (incl. 1 leap)
+        32044, // Fliegel-van Flandern JDN epoch shift
+        1721425, // JDN of 0001-01-01 (Gregorian)
+        1721119, // alt epoch offset
+        2400000, // MJD -> JDN conversion
+        2400001, // alt MJD -> JDN conversion
+        2141, // fixed-point month/day extraction
+        65536, // observed in some conversions
+        7834, // observed in some conversions
+        256, // observed in some conversions
+        292275056, // qdatetime.h Qt Core year range first year constant
+        292278994, // qdatetime.h Qt Core year range last year constant
+        1601, // Windows FILETIME epoch start year
+        1970, // Unix epoch start year
+        70, // Unix epoch start year short form
+        1899, // Observed in uses with 1900 to address off by one scenarios
+        1900, // Used when converting a 2 digit year
+        2000, // Used when converting a 2 digit year
+        1400, // Hijri base year, used when converting a 2 digit year
+        1980, // FAT filesystem epoch start year
+        227013, // constant observed for Hirji year conversion, and Hirji years are not applicable for gregorian leap year
+        10631, // constant observed for Hirji year conversion, and Hirji years are not applicable for gregorian leap year,
+        80, // 1980/01/01 is the start of the epoch on DOS
+        0
+      ]
+  )
+}
+
+/**
+ * An `isLikelyConversionConstant` constant indicates conversion that is ignorable, e.g.,
+ * julian to gregorian conversion or conversions from linux time structs
+ * that start at 1900, etc.
+ */
+class IgnorableConstantArithmetic extends IgnorableOperation {
+  IgnorableConstantArithmetic() {
+    exists(int i | isLikelyConversionConstant(i) |
+      this.(Operation).getAnOperand().getValue().toInt() = i
       or
-      // If there is a data flow from the variable that was modified to a function that seems to check for leap year
-      exists(VariableAccess source, ChecksForLeapYearFunctionCall fc |
-        source = var.getAnAccess() and
-        LeapYearCheckFlow::flow(DataFlow::exprNode(source), DataFlow::exprNode(fc.getAnArgument()))
-      )
-      or
-      // If there is a data flow from the field that was modified to a function that seems to check for leap year
-      exists(VariableAccess vacheck, YearFieldAccess yfacheck, ChecksForLeapYearFunctionCall fc |
-        vacheck = var.getAnAccess() and
-        yfacheck.getQualifier() = vacheck and
-        LeapYearCheckFlow::flow(DataFlow::exprNode(yfacheck), DataFlow::exprNode(fc.getAnArgument()))
-      )
-      or
-      // If there is a successor or predecessor that sets the month = 1
-      exists(MonthFieldAccess mfa, AssignExpr ae |
-        mfa.getQualifier() = var.getAnAccess() and
-        mfa.isModified() and
-        (
-          mfa.getBasicBlock() = yfa.getBasicBlock().getASuccessor*() or
-          yfa.getBasicBlock() = mfa.getBasicBlock().getASuccessor+()
-        ) and
-        ae = mfa.getEnclosingElement() and
-        ae.getAnOperand().getValue().toInt() = 1
+      exists(AssignArithmeticOperation a | this = a.getRValue() |
+        a.getRValue().getValue().toInt() = i
       )
     )
+  }
+}
+
+// If a unary minus assume it is some sort of conversion
+class IgnorableUnaryMinus extends IgnorableOperation {
+  IgnorableUnaryMinus() {
+    this instanceof UnaryMinusExpr
+    or
+    this.(Operation).getAnOperand() instanceof UnaryMinusExpr
+  }
+}
+
+/**
+ * An argument to a function is ignorable if the function that is called is an ignored function
+ */
+class OperationAsArgToIgnorableFunction extends IgnorableOperation {
+  OperationAsArgToIgnorableFunction() {
+    exists(Call c |
+      c.getAnArgument().getAChild*() = this and
+      c.getTarget() instanceof IgnorableFunction
+    )
+  }
+}
+
+/**
+ * A binary operation on two literals means the result is constant/known
+ * and the operation is basically ignorable (it's not a real operation but
+ * probably one visual simplicity what it means).
+ */
+class ConstantBinaryArithmeticOperation extends IgnorableOperation, BinaryArithmeticOperation {
+  ConstantBinaryArithmeticOperation() {
+    this.getLeftOperand() instanceof Literal and
+    this.getRightOperand() instanceof Literal
+  }
+}
+
+class IgnorableBinaryBitwiseOperation extends IgnorableOperation instanceof BinaryBitwiseOperation {
+}
+
+class IgnorableUnaryBitwiseOperation extends IgnorableOperation instanceof UnaryBitwiseOperation { }
+
+class IgnorableAssignmentBitwiseOperation extends IgnorableOperation instanceof AssignBitwiseOperation
+{ }
+
+/**
+ * An arithmetic operation where one of the operands is a pointer or char type, ignore it
+ */
+class IgnorablePointerOrCharArithmetic extends IgnorableOperation {
+  IgnorablePointerOrCharArithmetic() {
+    this instanceof BinaryArithmeticOperation and
+    exists(Expr op | op = this.(BinaryArithmeticOperation).getAnOperand() |
+      op.getUnspecifiedType() instanceof PointerType
+      or
+      op.getUnspecifiedType() instanceof CharType
+      or
+      // Operations on calls to functions that accept char or char*
+      op.(Call).getAnArgument().getUnspecifiedType().stripType() instanceof CharType
+      or
+      // Operations on calls to functions named like "strlen", "wcslen", etc
+      // NOTE: workaround for cases where the wchar_t type is not a char, but an unsigned short
+      // unclear if there is a best way to filter cases like these out based on type info.
+      op.(Call).getTarget().getName().matches("%len%")
+    )
+    or
+    exists(AssignArithmeticOperation a | a.getRValue() = this |
+      exists(Expr op | op = a.getAnOperand() |
+        op.getUnspecifiedType() instanceof PointerType
+        or
+        op.getUnspecifiedType() instanceof CharType
+        or
+        // Operations on calls to functions that accept char or char*
+        op.(Call).getAnArgument().getUnspecifiedType().stripType() instanceof CharType
+      )
+      or
+      // Operations on calls to functions named like "strlen", "wcslen", etc
+      // for example `strlen(foo) + bar`
+      this.(BinaryArithmeticOperation).getAnOperand().(Call).getTarget().getName().matches("%len%")
+    )
+  }
+}
+
+/**
+ * Holds for an expression that is an add or similar operation that could flow to a Year field.
+ */
+predicate isOperationSourceCandidate(Expr e) {
+  not e instanceof IgnorableOperation and
+  exists(Function f |
+    f = e.getEnclosingFunction() and
+    not f instanceof IgnorableFunction
+  ) and
+  (
+    e instanceof SubExpr
+    or
+    e instanceof AddExpr
+    or
+    e instanceof CrementOperation
+    or
+    e instanceof AssignSubExpr
+    or
+    e instanceof AssignAddExpr
   )
-select yfa,
-  "Field $@ on variable $@ has been modified, but no appropriate check for LeapYear was found.",
-  yfa.getTarget(), yfa.getTarget().toString(), var, var.toString()
+}
+
+/**
+ * A data flow that tracks an ignorable operation (such as a bitwise operation) to an operation source, so we may disqualify it.
+ */
+module IgnorableOperationToOperationSourceCandidateConfig implements DataFlow::ConfigSig {
+  predicate isSource(DataFlow::Node n) { n.asExpr() instanceof IgnorableOperation }
+
+  predicate isSink(DataFlow::Node n) { isOperationSourceCandidate(n.asExpr()) }
+
+  // looking for sources and sinks in the same function
+  DataFlow::FlowFeature getAFeature() {
+    result instanceof DataFlow::FeatureEqualSourceSinkCallContext
+  }
+}
+
+module IgnorableOperationToOperationSourceCandidateFlow =
+  TaintTracking::Global<IgnorableOperationToOperationSourceCandidateConfig>;
+
+/**
+ * The set of all expressions which is a candidate expression and also does not flow from to to some ignorable expression (eg. bitwise op)
+ * ```
+ * a = something <<< 2;
+ * myDate.year = a + 1;        // invalid
+ * ...
+ * a = someDate.year + 1;
+ * myDate.year = a;            // valid
+ * ```
+ */
+class OperationSource extends Expr {
+  OperationSource() {
+    isOperationSourceCandidate(this) and
+    // If the candidate came from an ignorable operation, ignore the candidate
+    // NOTE: we cannot easily flow the candidate to an ignorable operation as that can
+    // be tricky in practice, e.g., a mod operation on a year would be part of a leap year check
+    // but a mod operation ending in a year is more indicative of something to ignore (a conversion)
+    not exists(IgnorableOperationToOperationSourceCandidateFlow::PathNode sink |
+      sink.getNode().asExpr() = this and
+      sink.isSink()
+    )
+  }
+}
+
+class YearFieldAssignmentNode extends DataFlow::Node {
+  YearFieldAccess access;
+
+  YearFieldAssignmentNode() {
+    exists(Function f |
+      f = this.getEnclosingCallable().getUnderlyingCallable() and not f instanceof IgnorableFunction
+    ) and
+    (
+      this.asDefinition().(Assignment).getLValue() = access
+      or
+      this.asDefinition().(CrementOperation).getOperand() = access
+      or
+      exists(Call c | c.getAnArgument() = access and this.asDefiningArgument() = access)
+      or
+      exists(Call c, AddressOfExpr aoe |
+        c.getAnArgument() = aoe and
+        aoe.getOperand() = access and
+        this.asDefiningArgument() = aoe
+      )
+    )
+  }
+
+  YearFieldAccess getYearFieldAccess() { result = access }
+}
+
+/**
+ * A DataFlow configuration for identifying flows from an identified source
+ * to the Year field of a date object.
+ */
+module OperationToYearAssignmentConfig implements DataFlow::ConfigSig {
+  predicate isSource(DataFlow::Node n) { n.asExpr() instanceof OperationSource }
+
+  predicate isSink(DataFlow::Node n) {
+    n instanceof YearFieldAssignmentNode and
+    not isYearModifiedWithCheck(n) and
+    not isControlledByMonthEqualityCheckNonFebruary(n.asExpr())
+  }
+
+  predicate isBarrier(DataFlow::Node n) {
+    exists(ArrayExpr arr | arr.getArrayOffset() = n.asExpr())
+    or
+    n.getType().getUnspecifiedType() instanceof PointerType
+    or
+    n.getType().getUnspecifiedType() instanceof CharType
+    or
+    // If a type resembles "string" ignore flow (likely string conversion, currently ignored)
+    n.getType().getUnspecifiedType().stripType().getName().toLowerCase().matches("%string%")
+    or
+    n.asExpr() instanceof IgnorableOperation
+    or
+    // Flowing into variables that indicate likely non-gregorian years are barriers
+    // e.g., names similar to hijri, persian, lunar, chinese, hebrew, etc.
+    exists(Variable v |
+      v.getName()
+          .toLowerCase()
+          .matches(["%hijri%", "%persian%", "%lunar%", "%chinese%", "%hebrew%"]) and
+      v.getAnAccess() = [n.asIndirectExpr(), n.asExpr()]
+    )
+    or
+    isLeapYearCheckSink(n)
+    or
+    // this is a bit of a hack to address cases where a year is normalized and checked, but the
+    // normalized year is never itself assigned to the final year struct
+    //    isLeapYear(getCivilYear(year))
+    //    struct.year = year
+    // This is assuming a user would have done this all on one line though.
+    // setting a variable for the conversion and passing that separately would be more difficult to track
+    // considering this approach good enough for current observed false positives
+    exists(Expr arg |
+      isLeapYearCheckCall(_, arg) and arg.getAChild*() = [n.asExpr(), n.asIndirectExpr()]
+    )
+    or
+    // If as the flow progresses, the value holding a dangerous operation result
+    // is apparently being passed by address to some function, it is more than likely
+    // intended to be modified, and therefore, the definition is killed.
+    exists(Call c | c.getAnArgument().(AddressOfExpr).getAnOperand() = n.asIndirectExpr())
+  }
+
+  /** Block flow out of an operation source to get the "closest" operation to the sink */
+  predicate isBarrierIn(DataFlow::Node n) { isSource(n) }
+
+  predicate isBarrierOut(DataFlow::Node n) { isSink(n) }
+}
+
+module OperationToYearAssignmentFlow = TaintTracking::Global<OperationToYearAssignmentConfig>;
+
+predicate isLeapYearCheckSink(DataFlow::Node sink) {
+  exists(LeapYearGuardCondition lgc |
+    lgc.checkedYearAccess() = [sink.asExpr(), sink.asIndirectExpr()]
+  )
+  or
+  isLeapYearCheckCall(_, [sink.asExpr(), sink.asIndirectExpr()])
+}
+
+predicate yearAssignmentToCheckCommonSteps(DataFlow::Node node1, DataFlow::Node node2) {
+  // flow from a YearFieldAccess to the qualifier
+  node2.asExpr() = node1.asExpr().(YearFieldAccess).getQualifier*()
+  or
+  // getting the 'access' can be tricky at definitions (assignments especially)
+  // as dataflow uses asDefinition not asExpr.
+  // the YearFieldAssignmentNode holds the access in these cases
+  node1.(YearFieldAssignmentNode).getYearFieldAccess().getQualifier() = node2.asExpr()
+  or
+  // flow from a year access qualifier to a year field
+  exists(YearFieldAccess yfa | node2.asExpr() = yfa and node1.asExpr() = yfa.getQualifier())
+  or
+  node1.(YearFieldAssignmentNode).getYearFieldAccess().getQualifier() = node2.asExpr()
+  or
+  // Pass through any intermediate struct
+  exists(Assignment a |
+    a.getRValue() = node1.asExpr() and
+    node2.asExpr() = a.getLValue().(YearFieldAccess).getQualifier*()
+  )
+  or
+  // in cases of t.year = x and the value of x is checked, but the year t.year isn't directly checked
+  // flow from a year assignment node to an RHS if it is an assignment
+  // e.g.,
+  //    t.year = x;
+  //    isLeapYear(x);
+  //    --> at this point there is no flow of t.year to a check, but only its raw value
+  // To detect the flow of 'x' to the isLeapYear check,
+  // flow from t.year to 'x' (at assignment, t.year = x, flow to the RHS to track use-use flow of x)
+  exists(YearFieldAssignmentNode yfan |
+    node1 = yfan and
+    node2.asExpr() = yfan.asDefinition().(Assignment).getRValue()
+  )
+}
+
+/**
+ * A flow configuration from a Year field access to some Leap year check or guard
+ */
+module YearAssignmentToLeapYearCheckConfig implements DataFlow::ConfigSig {
+  predicate isSource(DataFlow::Node source) { source instanceof YearFieldAssignmentNode }
+
+  predicate isSink(DataFlow::Node sink) { isLeapYearCheckSink(sink) }
+
+  predicate isAdditionalFlowStep(DataFlow::Node node1, DataFlow::Node node2) {
+    yearAssignmentToCheckCommonSteps(node1, node2)
+  }
+
+  /**
+   * Enforcing the check must occur in the same call context as the source,
+   * i.e., do not return from the source function and check in a caller.
+   */
+  DataFlow::FlowFeature getAFeature() { result instanceof DataFlow::FeatureHasSourceCallContext }
+}
+
+module YearAssignmentToLeapYearCheckFlow =
+  TaintTracking::Global<YearAssignmentToLeapYearCheckConfig>;
+
+/** Does there exist a flow from the given YearFieldAccess to a Leap Year check or guard? */
+predicate isYearModifiedWithCheck(YearFieldAssignmentNode n) {
+  exists(YearAssignmentToLeapYearCheckFlow::PathNode src |
+    src.isSource() and
+    src.getNode() = n
+  )
+  or
+  // If the time flows to a time conversion whose value/result is checked,
+  // assume the leap year is being handled.
+  exists(YearAssignmentToCheckedTimeConversionFlow::PathNode timeQualSrc |
+    timeQualSrc.isSource() and
+    timeQualSrc.getNode() = n
+  )
+}
+
+/**
+ * An expression which checks the value of a Month field `a->month == 1`.
+ */
+class MonthEqualityCheck extends EqualityOperation {
+  MonthEqualityCheck() { this.getAnOperand() instanceof MonthFieldAccess }
+
+  Expr getExprCompared() {
+    exists(Expr e |
+      e = this.getAnOperand() and
+      not e instanceof MonthFieldAccess and
+      result = e
+    )
+  }
+}
+
+final class FinalMonthEqualityCheck = MonthEqualityCheck;
+
+class MonthEqualityCheckGuard extends GuardCondition, FinalMonthEqualityCheck { }
+
+/**
+ * Verifies if the expression is guarded by a check on the Month property of a date struct, that is NOT February.
+ */
+bindingset[e]
+pragma[inline_late]
+predicate isControlledByMonthEqualityCheckNonFebruary(Expr e) {
+  exists(MonthEqualityCheckGuard monthGuard, Expr compared |
+    monthGuard.controls(e.getBasicBlock(), true) and
+    compared = monthGuard.getExprCompared() and
+    not compared.getValue().toInt() = 2
+  )
+}
+
+/**
+ * Flow from a year field access to a time conversion function
+ * that auto converts feb29 in non-leap year, or through a conversion function that doesn't
+ * auto convert to a sanity check guard of the result for error conditions.
+ */
+module YearAssignmentToCheckedTimeConversionConfig implements DataFlow::StateConfigSig {
+  // Flow state tracks if flow goes through a known time conversion function
+  // see `TimeConversionFunction`.
+  // A valid check with a time conversion function is either the case:
+  // 1) the year flows into a time conversion function, and the time conversion function's result is checked or
+  // 2) the year flows into a time conversion function that auto corrects for leap year, so no check is necessary.
+  class FlowState = boolean;
+
+  predicate isSource(DataFlow::Node source, FlowState state) {
+    source instanceof YearFieldAssignmentNode and
+    state = false
+  }
+
+  predicate isSink(DataFlow::Node sink, FlowState state) {
+    // Case 1: Flow through a time conversion function that requires a check,
+    // and we have arrived at a guard, implying the result was checked for possible error, including leap year error.
+    // state = true indicates the flow went through a time conversion function
+    state = true and
+    (
+      exists(IfStmt ifs | ifs.getCondition().getAChild*() = [sink.asExpr(), sink.asIndirectExpr()])
+      or
+      exists(ConditionalExpr ce |
+        ce.getCondition().getAChild*() = [sink.asExpr(), sink.asIndirectExpr()]
+      )
+      or
+      exists(Loop l | l.getCondition().getAChild*() = [sink.asExpr(), sink.asIndirectExpr()])
+    )
+    or
+    // Case 2: Flow through a time conversion function that auto corrects for leap year, so no check is necessary.
+    // state true or false, as flowing through a time conversion function is not necessary in this instance.
+    state in [true, false] and
+    exists(Call c, TimeConversionFunction f |
+      f.isAutoLeapYearCorrecting() and
+      c.getTarget() = f and
+      c.getAnArgument().getAChild*() = [sink.asExpr(), sink.asIndirectExpr()]
+    )
+  }
+
+  predicate isAdditionalFlowStep(
+    DataFlow::Node node1, FlowState state1, DataFlow::Node node2, FlowState state2
+  ) {
+    state1 in [true, false] and
+    state2 = true and
+    exists(Call c |
+      c.getTarget() instanceof TimeConversionFunction and
+      c.getAnArgument().getAChild*() = [node1.asExpr(), node1.asIndirectExpr()] and
+      node2.asExpr() = c
+    )
+  }
+
+  predicate isAdditionalFlowStep(DataFlow::Node node1, DataFlow::Node node2) {
+    yearAssignmentToCheckCommonSteps(node1, node2)
+  }
+
+  DataFlow::FlowFeature getAFeature() { result instanceof DataFlow::FeatureHasSourceCallContext }
+}
+
+module YearAssignmentToCheckedTimeConversionFlow =
+  DataFlow::GlobalWithState<YearAssignmentToCheckedTimeConversionConfig>;
+
+/**
+ * Finds flow from a parameter of a function to a leap year check.
+ * This is necessary to handle for scenarios like this:
+ *
+ *    year = DANGEROUS_OP // source
+ *    isLeap = isLeapYear(year);
+ *    // logic based on isLeap
+ *    struct.year = year; // sink
+ *
+ * In this case, we may flow a dangerous op to a year assignment, failing
+ * to barrier the flow through a leap year check, as the leap year check
+ * is nested, and dataflow does not progress down into the check and out.
+ * Instead, the point of this flow is to detect isLeapYear's argument
+ * is checked for leap year, making the isLeapYear call a barrier for
+ * the dangerous flow if we flow through the parameter identified to
+ * be checked.
+ */
+module ParameterToLeapYearCheckConfig implements DataFlow::ConfigSig {
+  predicate isSource(DataFlow::Node source) { exists(source.asParameter()) }
+
+  predicate isSink(DataFlow::Node sink) {
+    exists(LeapYearGuardCondition lgc |
+      lgc.checkedYearAccess() = [sink.asExpr(), sink.asIndirectExpr()]
+    )
+  }
+
+  predicate isAdditionalFlowStep(DataFlow::Node node1, DataFlow::Node node2) {
+    // flow from a YearFieldAccess to the qualifier
+    node2.asExpr() = node1.asExpr().(YearFieldAccess).getQualifier*()
+    or
+    // flow from a year access qualifier to a year field
+    exists(YearFieldAccess yfa | node2.asExpr() = yfa and node1.asExpr() = yfa.getQualifier())
+  }
+
+  /**
+   * Enforcing the check must occur in the same call context as the source,
+   * i.e., do not return from the source function and check in a caller.
+   */
+  DataFlow::FlowFeature getAFeature() { result instanceof DataFlow::FeatureHasSourceCallContext }
+}
+
+// NOTE: I do not believe taint flow is necessary here as we should
+// be flowing directyly from some parameter to a leap year check.
+module ParameterToLeapYearCheckFlow = DataFlow::Global<ParameterToLeapYearCheckConfig>;
+
+predicate isLeapYearCheckCall(Call c, Expr arg) {
+  exists(ParameterToLeapYearCheckFlow::PathNode src, Function f, int i |
+    src.isSource() and
+    f.getParameter(i) = src.getNode().asParameter() and
+    c.getTarget() = f and
+    c.getArgument(i) = arg
+  )
+}
+
+class LeapYearGuardCondition extends GuardCondition {
+  Expr yearSinkDiv4;
+  Expr yearSinkDiv100;
+  Expr yearSinkDiv400;
+
+  LeapYearGuardCondition() {
+    exists(
+      LogicalAndExpr andExpr, LogicalOrExpr orExpr, GuardCondition div4Check,
+      GuardCondition div100Check, GuardCondition div400Check, GuardValue gv
+    |
+      // canonical case:
+      // form: `(year % 4 == 0) && (year % 100 != 0 || year % 400 == 0)`
+      //    `!((year % 4 == 0) && (year % 100 != 0 || year % 400 == 0))`
+      //    `!(year % 4) && (year % 100 || !(year % 400))`
+      // Also accepting `((year & 3) == 0) && (year % 100 != 0 || year % 400 == 0)`
+      // and `(year % 4 == 0) && (year % 100 > 0 || year % 400 == 0)`
+      this = andExpr and
+      andExpr.hasOperands(div4Check, orExpr) and
+      orExpr.hasOperands(div100Check, div400Check) and
+      (
+        // year % 4 == 0
+        exists(RemExpr e |
+          div4Check.comparesEq(e, 0, true, gv) and
+          e.getRightOperand().getValue().toInt() = 4 and
+          yearSinkDiv4 = e.getLeftOperand()
+        )
+        or
+        // year & 3 == 0
+        exists(BitwiseAndExpr e |
+          div4Check.comparesEq(e, 0, true, gv) and
+          e.getRightOperand().getValue().toInt() = 3 and
+          yearSinkDiv4 = e.getLeftOperand()
+        )
+      ) and
+      exists(RemExpr e |
+        // year % 100 != 0 or year % 100 > 0
+        (
+          div100Check.comparesEq(e, 0, false, gv) or
+          div100Check.comparesLt(e, 1, false, gv)
+        ) and
+        e.getRightOperand().getValue().toInt() = 100 and
+        yearSinkDiv100 = e.getLeftOperand()
+      ) and
+      // year % 400 == 0
+      exists(RemExpr e |
+        div400Check.comparesEq(e, 0, true, gv) and
+        e.getRightOperand().getValue().toInt() = 400 and
+        yearSinkDiv400 = e.getLeftOperand()
+      )
+      or
+      // Inverted logic case:
+      //  `year % 4 != 0 || (year % 100 == 0 && year % 400 != 0)`
+      // or `year & 3 != 0 || (year % 100 == 0 && year % 400 != 0)`
+      // also accepting `year % 4 > 0 || (year % 100 == 0 && year % 400 > 0)`
+      this = orExpr and
+      orExpr.hasOperands(div4Check, andExpr) and
+      andExpr.hasOperands(div100Check, div400Check) and
+      (
+        // year % 4 != 0 or year % 4 > 0
+        exists(RemExpr e |
+          (
+            div4Check.comparesEq(e, 0, false, gv)
+            or
+            div4Check.comparesLt(e, 1, false, gv)
+          ) and
+          e.getRightOperand().getValue().toInt() = 4 and
+          yearSinkDiv4 = e.getLeftOperand()
+        )
+        or
+        // year & 3 != 0
+        exists(BitwiseAndExpr e |
+          div4Check.comparesEq(e, 0, false, gv) and
+          e.getRightOperand().getValue().toInt() = 3 and
+          yearSinkDiv4 = e.getLeftOperand()
+        )
+      ) and
+      // year % 100 == 0
+      exists(RemExpr e |
+        div100Check.comparesEq(e, 0, true, gv) and
+        e.getRightOperand().getValue().toInt() = 100 and
+        yearSinkDiv100 = e.getLeftOperand()
+      ) and
+      // year % 400 != 0 or year % 400 > 0
+      exists(RemExpr e |
+        (
+          div400Check.comparesEq(e, 0, false, gv)
+          or
+          div400Check.comparesLt(e, 1, false, gv)
+        ) and
+        e.getRightOperand().getValue().toInt() = 400 and
+        yearSinkDiv400 = e.getLeftOperand()
+      )
+    )
+  }
+
+  Expr getYearSinkDiv4() { result = yearSinkDiv4 }
+
+  Expr getYearSinkDiv100() { result = yearSinkDiv100 }
+
+  Expr getYearSinkDiv400() { result = yearSinkDiv400 }
+
+  /**
+   * Gets the variable access that is used in all 3 components of the leap year check
+   * e.g., see getYearSinkDiv4/100/400..
+   * If a field access is used, the qualifier and the field access are both returned
+   * in checked condition.
+   * NOTE: if the year is not checked using the same access in all 3 components, no result is returned.
+   * The typical case observed is a consistent variable access is used. If not, this may indicate a bug.
+   * We could check more accurately with a dataflow analysis, but this is likely sufficient for now.
+   */
+  VariableAccess checkedYearAccess() {
+    exists(Variable var |
+      (
+        this.getYearSinkDiv4().getAChild*() = var.getAnAccess() and
+        this.getYearSinkDiv100().getAChild*() = var.getAnAccess() and
+        this.getYearSinkDiv400().getAChild*() = var.getAnAccess() and
+        result = var.getAnAccess() and
+        (
+          result = this.getYearSinkDiv4().getAChild*() or
+          result = this.getYearSinkDiv100().getAChild*() or
+          result = this.getYearSinkDiv400().getAChild*()
+        )
+      )
+    )
+  }
+}
+
+/**
+ * A difficult case to detect is if a year modification is tied to a month or day modification
+ * and the month or day is safe for leap year.
+ *    e.g.,
+ *          year++;
+ *          month = 1;
+ *          // alternative: day = 15;
+ *        ... values eventually used in the same time struct
+ * If this is even more challenging if the struct the values end up in are not
+ * local (set inter-procedurally).
+ * This configuration looks for constants 1-31 flowing to a month or day assignment.
+ * It is assumed a user of this flow will check if the month/day source and month/day sink
+ * are in the same basic blocks as a year modification source and a year modification sink.
+ * It is also assumed a user will check if the constant source is a value that is ignorable
+ * e.g., if it is 2 and the sink is a month assignment, then it isn't ignorable or
+ * if the value is < 27 and is a day assignment, it is likely ignorable
+ *
+ * Obviously this does not handle all conditions (e.g., the month set in another block).
+ * It is meant to capture the most common cases of false positives.
+ */
+module CandidateConstantToDayOrMonthAssignmentConfig implements DataFlow::ConfigSig {
+  predicate isSource(DataFlow::Node source) {
+    source.asExpr().getValue().toInt() in [1 .. 31] and
+    (
+      exists(Assignment a | a.getRValue() = source.asExpr())
+      or
+      exists(Call c | c.getAnArgument() = source.asExpr())
+    )
+  }
+
+  predicate isSink(DataFlow::Node sink) {
+    exists(Assignment a |
+      (a.getLValue() instanceof MonthFieldAccess or a.getLValue() instanceof DayFieldAccess) and
+      a.getRValue() = sink.asExpr()
+    )
+  }
+}
+
+// NOTE: only data flow here (no taint tracking) as we want the exact
+// constant flowing to the month assignment
+module CandidateConstantToDayOrMonthAssignmentFlow =
+  DataFlow::Global<CandidateConstantToDayOrMonthAssignmentConfig>;
+
+/**
+ * Holds if value the assignment `a` resolves to (`dayOrMonthValSrcExpr`) doesn't represent February,
+ * and/or if it represents a day, is a 'safe' day (meaning the 27th or prior).
+ */
+bindingset[dayOrMonthValSrcExpr]
+predicate isSafeValueForAssignmentOfMonthOrDayValue(Assignment a, Expr dayOrMonthValSrcExpr) {
+  a.getLValue() instanceof MonthFieldAccess and
+  dayOrMonthValSrcExpr.getValue().toInt() != 2
+  or
+  a.getLValue() instanceof DayFieldAccess and
+  dayOrMonthValSrcExpr.getValue().toInt() <= 27
+}
+
+import OperationToYearAssignmentFlow::PathGraph
+
+from OperationToYearAssignmentFlow::PathNode src, OperationToYearAssignmentFlow::PathNode sink
+where
+  OperationToYearAssignmentFlow::flowPath(src, sink) and
+  // Check if a month is set in the same block as the year operation source
+  // and the month value would indicate its set to any other month than february.
+  // Finds if the source year node is in the same block as a source month block
+  // and if the same for the sinks.
+  not exists(DataFlow::Node dayOrMonthValSrc, DataFlow::Node dayOrMonthValSink, Assignment a |
+    CandidateConstantToDayOrMonthAssignmentFlow::flow(dayOrMonthValSrc, dayOrMonthValSink) and
+    a.getRValue() = dayOrMonthValSink.asExpr() and
+    dayOrMonthValSink.getBasicBlock() = sink.getNode().getBasicBlock() and
+    exists(IRBlock dayOrMonthValBB |
+      dayOrMonthValBB = dayOrMonthValSrc.getBasicBlock() and
+      // The source of the day is set in the same block as the source for the year
+      // or the source for the day is set in the same block as the sink for the year
+      dayOrMonthValBB in [
+          src.getNode().getBasicBlock(),
+          sink.getNode().getBasicBlock()
+        ]
+    ) and
+    isSafeValueForAssignmentOfMonthOrDayValue(a, dayOrMonthValSrc.asExpr())
+  )
+select sink, src, sink,
+  "Year field has been modified, but no appropriate check for LeapYear was found."

cpp/ql/src/Likely Bugs/Leap Year/UncheckedReturnValueForTimeFunctions.ql

@@ -44,23 +44,9 @@ class SafeTimeGatheringFunction extends Function {
   }
 }
 
-/**
- * This list of APIs should check for the return value to detect problems during the conversion.
- */
-class TimeConversionFunction extends Function {
-  TimeConversionFunction() {
-    this.getQualifiedName() =
-      [
-        "FileTimeToSystemTime", "SystemTimeToFileTime", "SystemTimeToTzSpecificLocalTime",
-        "SystemTimeToTzSpecificLocalTimeEx", "TzSpecificLocalTimeToSystemTime",
-        "TzSpecificLocalTimeToSystemTimeEx", "RtlLocalTimeToSystemTime",
-        "RtlTimeToSecondsSince1970", "_mkgmtime"
-      ]
-  }
-}
-
 from FunctionCall fcall, TimeConversionFunction trf, Variable var
 where
+  not trf.isAutoLeapYearCorrecting() and
   fcall = trf.getACallToThisFunction() and
   fcall instanceof ExprInVoidContext and
   var.getUnderlyingType() instanceof UnpackedTimeType and

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

@@ -16,17 +16,15 @@
 import cpp
 import semmle.code.cpp.ir.IR
 import semmle.code.cpp.ir.dataflow.MustFlow
-import PathGraph
+import ReturnStackAllocatedMemory::PathGraph
 
 /** Holds if `f` has a name that we interpret as evidence of intentionally returning the value of the stack pointer. */
 predicate intentionallyReturnsStackPointer(Function f) {
   f.getName().toLowerCase().matches(["%stack%", "%sp%"])
 }
 
-class ReturnStackAllocatedMemoryConfig extends MustFlowConfiguration {
-  ReturnStackAllocatedMemoryConfig() { this = "ReturnStackAllocatedMemoryConfig" }
-
-  override predicate isSource(Instruction source) {
+module ReturnStackAllocatedMemoryConfig implements MustFlow::ConfigSig {
+  predicate isSource(Instruction source) {
     exists(Function func |
       // Rule out FPs caused by extraction errors.
       not func.hasErrors() and
@@ -50,7 +48,7 @@ class ReturnStackAllocatedMemoryConfig extends MustFlowConfiguration {
     )
   }
 
-  override predicate isSink(Operand sink) {
+  predicate isSink(Operand sink) {
     // Holds if `sink` is a node that represents the `StoreInstruction` that is subsequently used in
     // a `ReturnValueInstruction`.
     // We use the `StoreInstruction` instead of the instruction that defines the
@@ -72,7 +70,7 @@ class ReturnStackAllocatedMemoryConfig extends MustFlowConfiguration {
   //   int* px = id(&x);
   //  }
   //   ```
-  override predicate allowInterproceduralFlow() { none() }
+  predicate allowInterproceduralFlow() { none() }
 
   /**
    * This configuration intentionally conflates addresses of fields and their object, and pointer offsets
@@ -87,20 +85,22 @@ class ReturnStackAllocatedMemoryConfig extends MustFlowConfiguration {
    * }
    * ```
    */
-  override predicate isAdditionalFlowStep(Operand node1, Instruction node2) {
+  predicate isAdditionalFlowStep(Operand node1, Instruction node2) {
     node2.(FieldAddressInstruction).getObjectAddressOperand() = node1
     or
     node2.(PointerOffsetInstruction).getLeftOperand() = node1
   }
 
-  override predicate isBarrier(Instruction n) { n.getResultType() instanceof ErroneousType }
+  predicate isBarrier(Instruction n) { n.getResultType() instanceof ErroneousType }
 }
 
+module ReturnStackAllocatedMemory = MustFlow::Global<ReturnStackAllocatedMemoryConfig>;
+
 from
-  MustFlowPathNode source, MustFlowPathNode sink, Instruction instr,
-  ReturnStackAllocatedMemoryConfig conf
+  ReturnStackAllocatedMemory::PathNode source, ReturnStackAllocatedMemory::PathNode sink,
+  Instruction instr
 where
-  conf.hasFlowPath(pragma[only_bind_into](source), pragma[only_bind_into](sink)) and
+  ReturnStackAllocatedMemory::flowPath(pragma[only_bind_into](source), pragma[only_bind_into](sink)) and
   source.getInstruction() = instr
 select sink.getInstruction(), source, sink, "May return stack-allocated memory from $@.",
   instr.getAst(), instr.getAst().toString()

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

@@ -15,7 +15,7 @@
 import cpp
 import semmle.code.cpp.ir.IR
 import semmle.code.cpp.ir.dataflow.MustFlow
-import PathGraph
+import UninitializedLocal::PathGraph
 
 /**
  * Auxiliary predicate: Types that don't require initialization
@@ -70,25 +70,26 @@ predicate isSinkImpl(Instruction sink, VariableAccess va) {
   )
 }
 
-class MustFlow extends MustFlowConfiguration {
-  MustFlow() { this = "MustFlow" }
-
-  override predicate isSource(Instruction source) {
+module UninitializedLocalConfig implements MustFlow::ConfigSig {
+  predicate isSource(Instruction source) {
     source instanceof UninitializedInstruction and
     exists(Type t | t = source.getResultType() | not allocatedType(t))
   }
 
-  override predicate isSink(Operand sink) { isSinkImpl(sink.getDef(), _) }
+  predicate isSink(Operand sink) { isSinkImpl(sink.getDef(), _) }
 
-  override predicate allowInterproceduralFlow() { none() }
+  predicate allowInterproceduralFlow() { none() }
 
-  override predicate isBarrier(Instruction instr) { instr instanceof ChiInstruction }
+  predicate isBarrier(Instruction instr) { instr instanceof ChiInstruction }
 }
 
+module UninitializedLocal = MustFlow::Global<UninitializedLocalConfig>;
+
 from
-  VariableAccess va, LocalVariable v, MustFlow conf, MustFlowPathNode source, MustFlowPathNode sink
+  VariableAccess va, LocalVariable v, UninitializedLocal::PathNode source,
+  UninitializedLocal::PathNode sink
 where
-  conf.hasFlowPath(source, sink) and
+  UninitializedLocal::flowPath(source, sink) and
   isSinkImpl(sink.getInstruction(), va) and
   v = va.getTarget()
 select va, source, sink, "The variable $@ may not be initialized at this access.", v, v.getName()

cpp/ql/src/Likely Bugs/OO/UnsafeUseOfThis.ql

@@ -17,16 +17,16 @@
 import cpp
 import semmle.code.cpp.ir.IR
 import semmle.code.cpp.ir.dataflow.MustFlow
-import PathGraph
+import UnsafeUseOfThis::PathGraph
 
-class UnsafeUseOfThisConfig extends MustFlowConfiguration {
-  UnsafeUseOfThisConfig() { this = "UnsafeUseOfThisConfig" }
+module UnsafeUseOfThisConfig implements MustFlow::ConfigSig {
+  predicate isSource(Instruction source) { isSource(source, _, _) }
 
-  override predicate isSource(Instruction source) { isSource(source, _, _) }
-
-  override predicate isSink(Operand sink) { isSink(sink, _) }
+  predicate isSink(Operand sink) { isSink(sink, _) }
 }
 
+module UnsafeUseOfThis = MustFlow::Global<UnsafeUseOfThisConfig>;
+
 /** Holds if `sink` is a `this` pointer used by the call instruction `call`. */
 predicate isSink(Operand sink, CallInstruction call) {
   exists(PureVirtualFunction func |
@@ -66,19 +66,17 @@ predicate isSource(InitializeParameterInstruction source, string msg, Class c) {
  * - `msg` is a string describing whether `source` is from a constructor or destructor.
  */
 predicate flows(
-  MustFlowPathNode source, string msg, Class sourceClass, MustFlowPathNode sink,
+  UnsafeUseOfThis::PathNode source, string msg, Class sourceClass, UnsafeUseOfThis::PathNode sink,
   CallInstruction call
 ) {
-  exists(UnsafeUseOfThisConfig conf |
-    conf.hasFlowPath(source, sink) and
-    isSource(source.getInstruction(), msg, sourceClass) and
-    isSink(sink.getInstruction().getAUse(), call)
-  )
+  UnsafeUseOfThis::flowPath(source, sink) and
+  isSource(source.getInstruction(), msg, sourceClass) and
+  isSink(sink.getInstruction().getAUse(), call)
 }
 
 from
-  MustFlowPathNode source, MustFlowPathNode sink, CallInstruction call, string msg,
-  Class sourceClass
+  UnsafeUseOfThis::PathNode source, UnsafeUseOfThis::PathNode sink, CallInstruction call,
+  string msg, Class sourceClass
 where
   flows(source, msg, sourceClass, sink, call) and
   // Only raise an alert if there is no override of the pure virtual function in any base class.

cpp/ql/src/Security/CWE/CWE-089/SqlTainted.ql

@@ -45,14 +45,15 @@ module SqlTaintedConfig implements DataFlow::ConfigSig {
 
   predicate isBarrier(DataFlow::Node node) {
     node.asExpr().getUnspecifiedType() instanceof IntegralType
-  }
-
-  predicate isBarrierIn(DataFlow::Node node) {
+    or
     exists(SqlBarrierFunction sql, int arg, FunctionInput input |
       node.asIndirectArgument() = sql.getACallToThisFunction().getArgument(arg) and
       input.isParameterDeref(arg) and
       sql.barrierSqlArgument(input, _)
     )
+    or
+    // barrier defined using models-as-data
+    barrierNode(node, "sql-injection")
   }
 
   predicate observeDiffInformedIncrementalMode() { any() }

cpp/ql/src/change-notes/released/1.5.10.md

@@ -0,0 +1,3 @@
+## 1.5.10
+
+No user-facing changes.