Merge branch 'main' into py-restframework

.github/dependabot.yml

@@ -17,3 +17,26 @@ updates:
     ignore:
       - dependency-name: '*'
         update-types: ['version-update:semver-patch', 'version-update:semver-minor']
+
+  - package-ecosystem: "gomod"
+    directory: "go/extractor"
+    schedule:
+      interval: "daily"
+    allow:
+      - dependency-name: "golang.org/x/mod"
+      - dependency-name: "golang.org/x/tools"
+    groups:
+      extractor-dependencies:
+        patterns:
+          - "golang.org/x/*"
+    reviewers:
+      - "github/codeql-go"
+
+  - package-ecosystem: "gomod"
+    directory: "go/ql/test"
+    schedule:
+      interval: "monthly"
+    ignore:
+      - dependency-name: "*"
+    reviewers:
+      - "github/codeql-go"

.github/workflows/check-change-note.yml

@@ -15,20 +15,34 @@ on:
 
 jobs:
   check-change-note:
+    env: 
+      REPO: ${{ github.repository }}
+      PULL_REQUEST_NUMBER: ${{ github.event.number }}
+      GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
     runs-on: ubuntu-latest
     steps:
+
       - name: Fail if no change note found. To fix, either add one, or add the `no-change-note-required` label.
         if: |
           github.event.pull_request.draft == false &&
           !contains(github.event.pull_request.labels.*.name, 'no-change-note-required')
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
         run: |
-          gh api 'repos/${{github.repository}}/pulls/${{github.event.number}}/files' --paginate --jq 'any(.[].filename ; test("/change-notes/.*[.]md$"))' |
-          grep true -c
+          change_note_files=$(gh api "repos/$REPO/pulls/$PULL_REQUEST_NUMBER/files" --paginate --jq '.[].filename | select(test("/change-notes/.*[.]md$"))')
+          
+          if [ -z "$change_note_files" ]; then
+            echo "No change note found. Either add one, or add the 'no-change-note-required' label."
+            exit 1
+          fi
+
+          echo "Change notes found:"
+          echo "$change_note_files"
+
       - name: Fail if the change note filename doesn't match the expected format. The file name must be of the form 'YYYY-MM-DD.md', 'YYYY-MM-DD-{title}.md', where '{title}' is arbitrary text, or released/x.y.z.md for released change-notes
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
         run: |
-          gh api 'repos/${{github.repository}}/pulls/${{github.event.number}}/files' --paginate --jq '[.[].filename | select(test("/change-notes/.*[.]md$"))] | all(test("/change-notes/[0-9]{4}-[0-9]{2}-[0-9]{2}.*[.]md$") or test("/change-notes/released/[0-9]*[.][0-9]*[.][0-9]*[.]md$"))' |
-          grep true -c
+          bad_change_note_file_names=$(gh api "repos/$REPO/pulls/$PULL_REQUEST_NUMBER/files" --paginate --jq '[.[].filename | select(test("/change-notes/.*[.]md$"))][] | select((test("/change-notes/[0-9]{4}-[0-9]{2}-[0-9]{2}.*[.]md$") or test("/change-notes/released/[0-9]*[.][0-9]*[.][0-9]*[.]md$")) | not)')
+
+          if [ -n "$bad_change_note_file_names" ]; then
+            echo "The following change note file names are invalid:"
+            echo "$bad_change_note_file_names"
+            exit 1
+          fi

.github/workflows/compile-queries.yml

@@ -29,9 +29,9 @@ jobs:
         # run with --check-only if running in a PR (github.sha != main)
         if : ${{ github.event_name == 'pull_request' }}
         shell: bash
-        run: codeql query compile -q -j0 */ql/{src,examples} --keep-going --warnings=error --check-only --compilation-cache "${{ steps.query-cache.outputs.cache-dir }}"
+        run: codeql query compile -q -j0 */ql/{src,examples} --keep-going --warnings=error --check-only --compilation-cache "${{ steps.query-cache.outputs.cache-dir }}" --compilation-cache-size=500
       - name: compile queries - full
         # do full compile if running on main - this populates the cache
         if : ${{ github.event_name != 'pull_request' }}
         shell: bash
-        run: codeql query compile -q -j0 */ql/{src,examples} --keep-going --warnings=error --compilation-cache "${{ steps.query-cache.outputs.cache-dir }}"
+        run: codeql query compile -q -j0 */ql/{src,examples} --keep-going --warnings=error --compilation-cache "${{ steps.query-cache.outputs.cache-dir }}" --compilation-cache-size=500

.github/workflows/csharp-qltest.yml

@@ -91,7 +91,7 @@ jobs:
         run: |
           # Generate (Asp)NetCore stubs
           STUBS_PATH=stubs_output
-          python3 ql/src/Stubs/make_stubs_nuget.py webapp Swashbuckle.AspNetCore.Swagger latest "$STUBS_PATH"
+          python3 ql/src/Stubs/make_stubs_nuget.py webapp Swashbuckle.AspNetCore.Swagger 6.5.0 "$STUBS_PATH"
           rm -rf ql/test/resources/stubs/_frameworks
           # Update existing stubs in the repo with the freshly generated ones
           mv "$STUBS_PATH/output/stubs/_frameworks" ql/test/resources/stubs/

config/identical-files.json

@@ -498,22 +498,6 @@
     "ruby/ql/lib/codeql/ruby/frameworks/data/internal/ApiGraphModelsExtensions.qll",
     "python/ql/lib/semmle/python/frameworks/data/internal/ApiGraphModelsExtensions.qll"
   ],
-  "TaintedFormatStringQuery Ruby/JS": [
-    "javascript/ql/lib/semmle/javascript/security/dataflow/TaintedFormatStringQuery.qll",
-    "ruby/ql/lib/codeql/ruby/security/TaintedFormatStringQuery.qll"
-  ],
-  "TaintedFormatStringCustomizations Ruby/JS": [
-    "javascript/ql/lib/semmle/javascript/security/dataflow/TaintedFormatStringCustomizations.qll",
-    "ruby/ql/lib/codeql/ruby/security/TaintedFormatStringCustomizations.qll"
-  ],
-  "HttpToFileAccessQuery JS/Ruby": [
-    "javascript/ql/lib/semmle/javascript/security/dataflow/HttpToFileAccessQuery.qll",
-    "ruby/ql/lib/codeql/ruby/security/HttpToFileAccessQuery.qll"
-  ],
-  "HttpToFileAccessCustomizations JS/Ruby": [
-    "javascript/ql/lib/semmle/javascript/security/dataflow/HttpToFileAccessCustomizations.qll",
-    "ruby/ql/lib/codeql/ruby/security/HttpToFileAccessCustomizations.qll"
-  ],
   "Typo database": [
     "javascript/ql/src/Expressions/TypoDatabase.qll",
     "ql/ql/src/codeql_ql/style/TypoDatabase.qll"

cpp/BUILD.bazel

@@ -1,12 +1,17 @@
-package(default_visibility = ["//visibility:public"])
-
 load("@rules_pkg//:mappings.bzl", "pkg_filegroup")
 
+package(default_visibility = ["//visibility:public"])
+
 alias(
     name = "dbscheme",
     actual = "//cpp/ql/lib:dbscheme",
 )
 
+alias(
+    name = "dbscheme-stats",
+    actual = "//cpp/ql/lib:dbscheme-stats",
+)
+
 pkg_filegroup(
     name = "db-files",
     srcs = [

cpp/downgrades/5b388693c66db1e7dc2e76a90aa67a2b6eb74f0f/builtintypes.ql

@@ -0,0 +1,19 @@
+class BuiltinType extends @builtintype {
+  string toString() { none() }
+}
+
+from BuiltinType type, string name, int kind, int kind_new, int size, int sign, int alignment
+where
+  builtintypes(type, name, kind, size, sign, alignment) and
+  if
+    type instanceof @fp16 or
+    type instanceof @std_bfloat16 or
+    type instanceof @std_float16 or
+    type instanceof @complex_std_float32 or
+    type instanceof @complex_float32x or
+    type instanceof @complex_std_float64 or
+    type instanceof @complex_float64x or
+    type instanceof @complex_std_float128
+  then kind_new = 2
+  else kind_new = kind
+select type, name, kind_new, size, sign, alignment

cpp/downgrades/5b388693c66db1e7dc2e76a90aa67a2b6eb74f0f/upgrade.properties

@@ -0,0 +1,3 @@
+description: Introduce new floating-point types from C23 and C++23
+compatibility: backwards
+builtintypes.rel: run builtintypes.qlo

cpp/ql/lib/BUILD.bazel

@@ -1,7 +1,7 @@
-package(default_visibility = ["//cpp:__pkg__"])
-
 load("@rules_pkg//:mappings.bzl", "pkg_files")
 
+package(default_visibility = ["//cpp:__pkg__"])
+
 pkg_files(
     name = "dbscheme",
     srcs = ["semmlecode.cpp.dbscheme"],

cpp/ql/lib/CHANGELOG.md

@@ -1,3 +1,39 @@
+## 0.10.1
+
+### Minor Analysis Improvements
+
+* Deleted the deprecated `AnalysedString` class, use the new name `AnalyzedString`.
+* Deleted the deprecated `isBarrierGuard` predicate from the dataflow library and its uses, use `isBarrier` and the `BarrierGuard` module instead.
+
+## 0.10.0
+
+### Minor Analysis Improvements
+
+* Functions that do not return due to calling functions that don't return (e.g. `exit`) are now detected as
+ non-returning in the IR and dataflow.
+* Treat functions that reach the end of the function as returning in the IR.
+  They used to be treated as unreachable but it is allowed in C. 
+* The `DataFlow::asDefiningArgument` predicate now takes its argument from the range starting at `1` instead of `2`. Queries that depend on the single-parameter version of `DataFlow::asDefiningArgument` should have their arguments updated accordingly.
+
+## 0.9.3
+
+No user-facing changes.
+
+## 0.9.2
+
+### Deprecated APIs
+
+* `getAllocatorCall` on `DeleteExpr` and `DeleteArrayExpr` has been deprecated. `getDeallocatorCall` should be used instead.
+
+### New Features
+
+* Added `DeleteOrDeleteArrayExpr` as a super type of `DeleteExpr` and `DeleteArrayExpr`
+
+### Minor Analysis Improvements
+
+* `delete` and `delete[]` are now modeled as calls to the relevant `operator delete` in the IR. In the case of a dynamic delete call a new instruction `VirtualDeleteFunctionAddress` is used to represent a function that dispatches to the correct delete implementation.
+* Only the 2 level indirection of `argv` (corresponding to `**argv`) is consided for `FlowSource`.
+
 ## 0.9.1
 
 No user-facing changes.

cpp/ql/lib/change-notes/2023-08-24-no-taint-argv-indirections.md

@@ -1,4 +0,0 @@
----
-category: minorAnalysis
----
-* Only the 2 level indirection of `argv` (corresponding to `**argv`) is consided for `FlowSource`.

cpp/ql/lib/change-notes/2023-08-25-delete-or-delete-array.md

@@ -1,4 +0,0 @@
----
-category: feature
----
-* Added `DeleteOrDeleteArrayExpr` as a super type of `DeleteExpr` and `DeleteArrayExpr`

cpp/ql/lib/change-notes/2023-08-25-getAllocatorCall-deprecated.md

@@ -1,4 +0,0 @@
----
-category: deprecated
----
-* `getAllocatorCall` on `DeleteExpr` and `DeleteArrayExpr` has been deprecated. `getDeallocatorCall` should be used instead.

cpp/ql/lib/change-notes/2023-08-29-delete-ir.md

@@ -1,4 +0,0 @@
----
-category: minorAnalysis
----
-* `delete` and `delete[]` are now modeled as calls to the relevant `operator delete` in the IR. In the case of a dynamic delete call a new instruction `VirtualDeleteFunctionAddress` is used to represent a function that dispatches to the correct delete implementation.

cpp/ql/lib/change-notes/2023-09-04-more-floating-point-types.md

@@ -0,0 +1,4 @@
+---
+category: minorAnalysis
+---
+* Added support for new floating-point types in C23 and C++23.

cpp/ql/lib/change-notes/2023-09-06-as-defining-argument-off-by-one-fix.md

@@ -1,4 +0,0 @@
----
-category: minorAnalysis
----
-* The `DataFlow::asDefiningArgument` predicate now takes its argument from the range starting at `1` instead of `2`. Queries that depend on the single-parameter version of `DataFlow::asDefiningArgument` should have their arguments updated accordingly.

cpp/ql/lib/change-notes/2023-09-07-return-from-end.md

@@ -1,5 +0,0 @@
----
-category: minorAnalysis
----
-* Treat functions that reach the end of the function as returning in the IR.
-  They used to be treated as unreachable but it is allowed in C. 

cpp/ql/lib/change-notes/2023-09-08-more-unreachble.md

@@ -1,5 +0,0 @@
----
-category: minorAnalysis
----
-* Functions that do not return due to calling functions that don't return (e.g. `exit`) are now detected as
- non-returning in the IR and dataflow.

cpp/ql/lib/change-notes/2023-10-12-additional-call-targets.md

@@ -0,0 +1,4 @@
+---
+category: feature
+---
+* Added a new class `AdditionalCallTarget` for specifying additional call targets.

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

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

cpp/ql/lib/change-notes/released/0.10.1.md

@@ -0,0 +1,6 @@
+## 0.10.1
+
+### Minor Analysis Improvements
+
+* Deleted the deprecated `AnalysedString` class, use the new name `AnalyzedString`.
+* Deleted the deprecated `isBarrierGuard` predicate from the dataflow library and its uses, use `isBarrier` and the `BarrierGuard` module instead.

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

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

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

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

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

@@ -1,2 +1,2 @@
 ---
-lastReleaseVersion: 0.9.1
+lastReleaseVersion: 0.10.1

cpp/ql/lib/experimental/cryptography/Concepts.qll

@@ -0,0 +1,3 @@
+import experimental.cryptography.CryptoArtifact
+import experimental.cryptography.CryptoAlgorithmNames
+import experimental.cryptography.modules.OpenSSL as OpenSSL

cpp/ql/lib/experimental/cryptography/CryptoAlgorithmNames.qll

@@ -0,0 +1,239 @@
+/**
+ * Names of known cryptographic algorithms.
+ * The names are standardized into upper-case, no spaces, dashes or underscores.
+ */
+
+/**
+ * Returns a string to represent generally unknown algorithms.
+ * Predicate is to be used to get a consistent string representation
+ * for unknown algorithms.
+ */
+string unknownAlgorithm() { result = "UNKNOWN" }
+
+string getHashType() { result = "HASH" }
+
+string getSymmetricEncryptionType() { result = "SYMMETRIC_ENCRYPTION" }
+
+string getAsymmetricEncryptionType() { result = "ASYMMETRIC_ENCRYPTION" }
+
+string getKeyDerivationType() { result = "KEY_DERIVATION" }
+
+string getCipherBlockModeType() { result = "BLOCK_MODE" }
+
+string getSymmetricPaddingType() { result = "SYMMETRIC_PADDING" }
+
+string getAsymmetricPaddingType() { result = "ASYMMETRIC_PADDING" }
+
+string getEllipticCurveType() { result = "ELLIPTIC_CURVE" }
+
+string getSignatureType() { result = "SIGNATURE" }
+
+string getKeyExchangeType() { result = "KEY_EXCHANGE" }
+
+string getAsymmetricType() {
+  result in [
+      getAsymmetricEncryptionType(), getSignatureType(), getKeyExchangeType(),
+      getEllipticCurveType()
+    ]
+}
+
+predicate isKnownType(string algType) {
+  algType in [
+      getHashType(), getSymmetricEncryptionType(), getAsymmetricEncryptionType(),
+      getKeyDerivationType(), getCipherBlockModeType(), getSymmetricPaddingType(),
+      getAsymmetricPaddingType(), getEllipticCurveType(), getSignatureType(), getKeyExchangeType()
+    ]
+}
+
+predicate isKnownAlgorithm(string name) { isKnownAlgorithm(name, _) }
+
+predicate isKnownAlgorithm(string name, string algType) {
+  isHashingAlgorithm(name) and algType = "HASH"
+  or
+  isEncryptionAlgorithm(name, algType) and
+  algType in ["SYMMETRIC_ENCRYPTION", "ASYMMETRIC_ENCRYPTION"]
+  or
+  isKeyDerivationAlgorithm(name) and algType = "KEY_DERIVATION"
+  or
+  isCipherBlockModeAlgorithm(name) and algType = "BLOCK_MODE"
+  or
+  isPaddingAlgorithm(name, algType) and algType in ["SYMMETRIC_PADDING", "ASYMMETRIC_PADDING"]
+  or
+  isEllipticCurveAlgorithm(name) and algType = "ELLIPTIC_CURVE"
+  or
+  isSignatureAlgorithm(name) and algType = "SIGNATURE"
+  or
+  isKeyExchangeAlgorithm(name) and algType = "KEY_EXCHANGE"
+}
+
+/**
+ * Holds if `name` is a known hashing algorithm in the model/library.
+ */
+predicate isHashingAlgorithm(string name) {
+  name =
+    [
+      "BLAKE2", "BLAKE2B", "BLAKE2S", "SHA2", "SHA224", "SHA256", "SHA384", "SHA512", "SHA512224",
+      "SHA512256", "SHA3", "SHA3224", "SHA3256", "SHA3384", "SHA3512", "SHAKE128", "SHAKE256",
+      "SM3", "WHIRLPOOL", "POLY1305", "HAVEL128", "MD2", "MD4", "MD5", "PANAMA", "RIPEMD",
+      "RIPEMD128", "RIPEMD256", "RIPEMD160", "RIPEMD320", "SHA0", "SHA1", "SHA", "MGF1", "MGF1SHA1",
+      "MDC2", "SIPHASH"
+    ]
+}
+
+predicate isEncryptionAlgorithm(string name, string algType) {
+  isAsymmetricEncryptionAlgorithm(name) and algType = "ASYMMETRIC_ENCRYPTION"
+  or
+  isSymmetricEncryptionAlgorithm(name) and algType = "SYMMETRIC_ENCRYPTION"
+}
+
+predicate isEncryptionAlgorithm(string name) { isEncryptionAlgorithm(name, _) }
+
+/**
+ * Holds if `name` corresponds to a known symmetric encryption algorithm.
+ */
+predicate isSymmetricEncryptionAlgorithm(string name) {
+  // NOTE: AES is meant to caputure all possible key lengths
+  name =
+    [
+      "AES", "AES128", "AES192", "AES256", "ARIA", "BLOWFISH", "BF", "ECIES", "CAST", "CAST5",
+      "CAMELLIA", "CAMELLIA128", "CAMELLIA192", "CAMELLIA256", "CHACHA", "CHACHA20",
+      "CHACHA20POLY1305", "GOST", "GOSTR34102001", "GOSTR341094", "GOSTR341194", "GOST2814789",
+      "GOSTR341194", "GOST2814789", "GOST28147", "GOSTR341094", "GOST89", "GOST94", "GOST34102012",
+      "GOST34112012", "IDEA", "RABBIT", "SEED", "SM4", "DES", "DESX", "3DES", "TDES", "2DES",
+      "DES3", "TRIPLEDES", "TDEA", "TRIPLEDEA", "ARC2", "RC2", "ARC4", "RC4", "ARCFOUR", "ARC5",
+      "RC5", "MAGMA", "KUZNYECHIK"
+    ]
+}
+
+/**
+ * Holds if `name` corresponds to a known key derivation algorithm.
+ */
+predicate isKeyDerivationAlgorithm(string name) {
+  name =
+    [
+      "ARGON2", "CONCATKDF", "CONCATKDFHASH", "CONCATKDFHMAC", "KBKDFCMAC", "BCRYPT", "HKDF",
+      "HKDFEXPAND", "KBKDF", "KBKDFHMAC", "PBKDF1", "PBKDF2", "PBKDF2HMAC", "PKCS5", "SCRYPT",
+      "X963KDF", "EVPKDF"
+    ]
+}
+
+/**
+ * Holds if `name` corresponds to a known cipher block mode
+ */
+predicate isCipherBlockModeAlgorithm(string name) {
+  name = ["CBC", "GCM", "CCM", "CFB", "OFB", "CFB8", "CTR", "OPENPGP", "XTS", "EAX", "SIV", "ECB"]
+}
+
+/**
+ * Holds if `name` corresponds to a known padding algorithm
+ */
+predicate isPaddingAlgorithm(string name, string algType) {
+  isSymmetricPaddingAlgorithm(name) and algType = "SYMMETRIC_PADDING"
+  or
+  isAsymmetricPaddingAlgorithm(name) and algType = "ASYMMETRIC_PADDING"
+}
+
+/**
+ * holds if `name` corresponds to a known symmetric padding algorithm
+ */
+predicate isSymmetricPaddingAlgorithm(string name) { name = ["PKCS7", "ANSIX923"] }
+
+/**
+ * Holds if `name` corresponds to a known asymmetric padding algorithm
+ */
+predicate isAsymmetricPaddingAlgorithm(string name) { name = ["OAEP", "PKCS1V15", "PSS", "KEM"] }
+
+predicate isBrainpoolCurve(string curveName, int keySize) {
+  // ALL BRAINPOOL CURVES
+  keySize in [160, 192, 224, 256, 320, 384, 512] and
+  (
+    curveName = "BRAINPOOLP" + keySize.toString() + "R1"
+    or
+    curveName = "BRAINPOOLP" + keySize.toString() + "T1"
+  )
+}
+
+predicate isSecCurve(string curveName, int keySize) {
+  // ALL SEC CURVES
+  keySize in [112, 113, 128, 131, 160, 163, 192, 193, 224, 233, 239, 256, 283, 384, 409, 521, 571] and
+  exists(string suff | suff in ["R1", "R2", "K1"] |
+    curveName = "SECT" + keySize.toString() + suff or
+    curveName = "SECP" + keySize.toString() + suff
+  )
+}
+
+predicate isC2Curve(string curveName, int keySize) {
+  // ALL C2 CURVES
+  keySize in [163, 176, 191, 208, 239, 272, 304, 359, 368, 431] and
+  exists(string pre, string suff |
+    pre in ["PNB", "ONB", "TNB"] and suff in ["V1", "V2", "V3", "V4", "V5", "W1", "R1"]
+  |
+    curveName = "C2" + pre + keySize.toString() + suff
+  )
+}
+
+predicate isPrimeCurve(string curveName, int keySize) {
+  // ALL PRIME CURVES
+  keySize in [192, 239, 256] and
+  exists(string suff | suff in ["V1", "V2", "V3"] | curveName = "PRIME" + keySize.toString() + suff)
+}
+
+predicate isEllipticCurveAlgorithm(string curveName) { isEllipticCurveAlgorithm(curveName, _) }
+
+/**
+ * Holds if `name` corresponds to a known elliptic curve.
+ */
+predicate isEllipticCurveAlgorithm(string curveName, int keySize) {
+  isSecCurve(curveName, keySize)
+  or
+  isBrainpoolCurve(curveName, keySize)
+  or
+  isC2Curve(curveName, keySize)
+  or
+  isPrimeCurve(curveName, keySize)
+  or
+  curveName = "ES256" and keySize = 256
+  or
+  curveName = "CURVE25519" and keySize = 255
+  or
+  curveName = "X25519" and keySize = 255
+  or
+  curveName = "ED25519" and keySize = 255
+  or
+  curveName = "CURVE448" and keySize = 448 // TODO: need to check the key size
+  or
+  curveName = "ED448" and keySize = 448
+  or
+  curveName = "X448" and keySize = 448
+  or
+  curveName = "NUMSP256T1" and keySize = 256
+  or
+  curveName = "NUMSP384T1" and keySize = 384
+  or
+  curveName = "NUMSP512T1" and keySize = 512
+  or
+  curveName = "SM2" and keySize in [256, 512]
+}
+
+/**
+ * Holds if `name` corresponds to a known signature algorithm.
+ */
+predicate isSignatureAlgorithm(string name) {
+  name =
+    [
+      "DSA", "ECDSA", "EDDSA", "ES256", "ES256K", "ES384", "ES512", "ED25519", "ED448", "ECDSA256",
+      "ECDSA384", "ECDSA512"
+    ]
+}
+
+/**
+ * Holds if `name` is a key exchange algorithm.
+ */
+predicate isKeyExchangeAlgorithm(string name) {
+  name = ["ECDH", "DH", "DIFFIEHELLMAN", "X25519", "X448"]
+}
+
+/**
+ * Holds if `name` corresponds to a known asymmetric encryption.
+ */
+predicate isAsymmetricEncryptionAlgorithm(string name) { name = ["RSA"] }

cpp/ql/lib/experimental/cryptography/CryptoArtifact.qll

@@ -0,0 +1,316 @@
+import cpp
+private import experimental.cryptography.CryptoAlgorithmNames
+import semmle.code.cpp.ir.dataflow.TaintTracking
+
+/*
+ * A cryptographic artifact is a DataFlow::Node associated with some
+ * operation, algorithm, or any other aspect of cryptography.
+ */
+
+abstract class CryptographicArtifact extends Expr { }
+
+// /**
+//  * Associates a symmetric encryption algorithm with a block mode.
+//  * The DataFlow::Node representing this association should be the
+//  * point where the algorithm and block mode are combined.
+//  * This may be at the call to encryption or in the construction
+//  * of an object prior to encryption.
+//  */
+// abstract class SymmetricCipher extends CryptographicArtifact{
+//   abstract SymmetricEncryptionAlgorithm getEncryptionAlgorithm();
+//   abstract BlockMode getBlockMode();
+//   final predicate hasBlockMode(){
+//     exists(this.getBlockMode())
+//   }
+// }
+// /**
+//  * A cryptographic operation is a method call that invokes a cryptographic
+//  * algorithm (encrypt/decrypt) or a function in support of a cryptographic algorithm
+//  * (key generation).
+//  *
+//  * Since operations are related to or in support of algorithms, operations must
+//  * provide a reference to their associated algorithm. Often operataions themselves
+//  * encapsulate algorithms, so operations can also extend CryptographicAlgorithm
+//  * and refer to themselves as the target algorithm.
+//  */
+// abstract class CryptographicOperation extends CryptographicArtifact, Call{
+//   // bindingset[paramName, ind]
+//   // final DataFlow::Node getParameterSource(int ind, string paramName){
+//   //   result = Utils::getUltimateSrcFromApiNode(this.(API::CallNode).getParameter(ind, paramName))
+//   // }
+//   final string getAlgorithmName(){
+//     if exists(this.getAlgorithm().getName())
+//     then result = this.getAlgorithm().getName()
+//     else result = unknownAlgorithm()
+//   }
+//   final predicate hasAlgorithm(){
+//     exists(this.getAlgorithm())
+//   }
+//   final predicate isUnknownAlgorithm(){
+//     this.getAlgorithmName() = unknownAlgorithm()
+//     or
+//     not this.hasAlgorithm()
+//   }
+//   // TODO: this might have to be parameterized by a configuration source for
+//   //       situations where an operation is passed an algorithm
+//   abstract CryptographicAlgorithm getAlgorithm();
+// }
+// /** A key generation operation for asymmetric keys */
+// abstract class KeyGen extends CryptographicOperation{
+//   int getAKeySizeInBits(){
+//     result = getKeySizeInBits(_)
+//   }
+//   final predicate hasKeySize(Expr configSrc){
+//     exists(this.getKeySizeInBits(configSrc))
+//   }
+//   final predicate hasKeySize(){
+//     exists(this.getAKeySizeInBits())
+//   }
+//   abstract Expr getKeyConfigSrc();
+//   abstract int getKeySizeInBits(Expr configSrc);
+// }
+abstract class CryptographicOperation extends CryptographicArtifact, Call { }
+
+abstract class KeyGeneration extends CryptographicOperation {
+  // TODO: what if the algorithm is UNKNOWN?
+  abstract Expr getKeyConfigurationSource(CryptographicAlgorithm alg);
+
+  abstract CryptographicAlgorithm getAlgorithm();
+
+  int getKeySizeInBits(CryptographicAlgorithm alg) {
+    result = this.getKeyConfigurationSource(alg).(Literal).getValue().toInt()
+  }
+
+  predicate hasConstantKeySize(CryptographicAlgorithm alg) { exists(this.getKeySizeInBits(alg)) }
+
+  predicate hasKeyConfigurationSource(CryptographicAlgorithm alg) {
+    exists(this.getKeyConfigurationSource(alg))
+  }
+
+  Expr getAKeyConfigurationSource() { result = this.getKeyConfigurationSource(_) }
+}
+
+abstract class AsymmetricKeyGeneration extends KeyGeneration { }
+
+abstract class SymmetricKeyGeneration extends KeyGeneration { }
+
+/**
+ * A cryptographic algorithm is a `CryptographicArtifact`
+ * representing a cryptographic algorithm (see `CryptoAlgorithmNames.qll`).
+ * Cryptographic algorithms can be functions referencing common crypto algorithms (e.g., hashlib.md5)
+ * or strings that are used in cryptographic operation configurations (e.g., hashlib.new("md5")).
+ * Cryptogrpahic algorithms may also be operations that wrap or abstract one or
+ * more algorithms (e.g., cyrptography.fernet.Fernet and AES, CBC and PKCS7).
+ *
+ * In principle, this class should model the location where an algorithm enters the program, not
+ * necessarily where it is used.
+ */
+abstract class CryptographicAlgorithm extends CryptographicArtifact {
+  abstract string getName();
+
+  abstract string getAlgType();
+
+  //  string getAlgType(){
+  //   if this instanceof HashAlgorithm then result = getHashType()
+  //   else if this instanceof KeyDerivationAlgorithm then result = getKeyDerivationType()
+  //   else if this instanceof SymmetricEncryptionAlgorithm then result = getSymmetricEncryptionType()
+  //   else if this instanceof AsymmetricEncryptionAlgorithm then result = getAsymmetricEncryptionType()
+  //   else if this instanceof SymmetricEncryptionAlgorithm then result = getSymmetricPaddingType()
+  //   else if this instanceof AsymmetricEncryptionAlgorithm then result = getAsymmetricPaddingType()
+  //   else if this instanceof EllipticCurveAlgorithm then result = getEllipticCurveType()
+  //   else if this instanceof BlockMode then result = getCipherBlockModeType()
+  //   else if this instanceof KeyExchangeAlgorithm then result = getKeyExchangeType()
+  //   else if this instanceof SigningAlgorithm then result = getSignatureType()
+  //   else result = unknownAlgorithm()
+  // }
+  // TODO: handle case where name isn't known, not just unknown?
+  /**
+   * Normalizes a raw name into a normalized name as found in `CryptoAlgorithmNames.qll`.
+   * Subclassess should override for more api-specific normalization.
+   * By deafult, converts a raw name to upper-case with no hyphen, underscore, hash, or space.
+   */
+  bindingset[s]
+  string normalizeName(string s) {
+    exists(string normStr | normStr = s.toUpperCase().regexpReplaceAll("[-_ ]|/", "") |
+      result = normStr and isKnownAlgorithm(result)
+      or
+      result = unknownAlgorithm() and not isKnownAlgorithm(normStr)
+    )
+  }
+
+  abstract Expr configurationSink();
+
+  predicate hasConfigurationSink() { exists(this.configurationSink()) }
+}
+
+abstract class HashAlgorithm extends CryptographicAlgorithm {
+  final string getHashName() {
+    if exists(string n | n = this.getName() and isHashingAlgorithm(n))
+    then isHashingAlgorithm(result) and result = this.getName()
+    else result = unknownAlgorithm()
+  }
+
+  override string getAlgType() { result = getHashType() }
+}
+
+abstract class KeyDerivationAlgorithm extends CryptographicAlgorithm {
+  final string getKDFName() {
+    if exists(string n | n = this.getName() and isKeyDerivationAlgorithm(n))
+    then isKeyDerivationAlgorithm(result) and result = this.getName()
+    else result = unknownAlgorithm()
+  }
+
+  override string getAlgType() { result = getKeyDerivationType() }
+}
+
+// abstract class KeyDerivationOperation extends CryptographicOperation{
+//   DataFlow::Node getIterationSizeSrc(){
+//     none()
+//   }
+//   DataFlow::Node getSaltConfigSrc(){
+//     none()
+//   }
+//   DataFlow::Node getHashConfigSrc(){
+//     none()
+//   }
+//   // TODO: get encryption algorithm for CBC-based KDF?
+//   DataFlow::Node getDerivedKeySizeSrc(){
+//     none()
+//   }
+//   DataFlow::Node getModeSrc(){
+//     none()
+//   }
+//   // TODO: add more to cover all the parameters of most KDF operations? Perhaps subclass for each type?
+//   abstract predicate requiresIteration();
+//   abstract predicate requiresSalt();
+//   abstract predicate requiresHash();
+//   //abstract predicate requiresKeySize(); // Going to assume all requires a size
+//   abstract predicate requiresMode();
+// }
+abstract class EncryptionAlgorithm extends CryptographicAlgorithm {
+  final predicate isAsymmetric() { this instanceof AsymmetricEncryptionAlgorithm }
+
+  final predicate isSymmetric() { not this.isAsymmetric() }
+  // NOTE: DO_NOT add getEncryptionName here, we rely on the fact the parent
+  //       class does not have this common predicate.
+}
+
+/**
+ * A parent class to represent any algorithm for which
+ * asymmetric cryptography is involved.
+ * Intended to be distinct from AsymmetricEncryptionAlgorithm
+ * which is intended only for asymmetric algorithms that specifically encrypt.
+ */
+abstract class AsymmetricAlgorithm extends CryptographicAlgorithm { }
+
+/**
+ * Algorithms directly or indirectly related to asymmetric encryption,
+ * e.g., RSA, DSA, but also RSA padding algorithms
+ */
+abstract class AsymmetricEncryptionAlgorithm extends AsymmetricAlgorithm, EncryptionAlgorithm {
+  final string getEncryptionName() {
+    if exists(string n | n = this.getName() and isAsymmetricEncryptionAlgorithm(n))
+    then isAsymmetricEncryptionAlgorithm(result) and result = this.getName()
+    else result = unknownAlgorithm()
+  }
+
+  override string getAlgType() { result = getAsymmetricEncryptionType() }
+}
+
+/**
+ * Algorithms directly or indirectly related to symmetric encryption,
+ * e.g., AES, DES, but also block modes and padding
+ */
+abstract class SymmetricEncryptionAlgorithm extends EncryptionAlgorithm {
+  final string getEncryptionName() {
+    if exists(string n | n = this.getName() and isSymmetricEncryptionAlgorithm(n))
+    then isSymmetricEncryptionAlgorithm(result) and result = this.getName()
+    else result = unknownAlgorithm()
+  }
+
+  // TODO: add a stream cipher predicate?
+  override string getAlgType() { result = getSymmetricEncryptionType() }
+}
+
+// Used only to categorize all padding into a single object,
+// DO_NOT add predicates here. Only for categorization purposes.
+abstract class PaddingAlgorithm extends CryptographicAlgorithm { }
+
+abstract class SymmetricPadding extends PaddingAlgorithm {
+  final string getPaddingName() {
+    if exists(string n | n = this.getName() and isSymmetricPaddingAlgorithm(n))
+    then isSymmetricPaddingAlgorithm(result) and result = this.getName()
+    else result = unknownAlgorithm()
+  }
+
+  override string getAlgType() { result = getSymmetricPaddingType() }
+}
+
+abstract class AsymmetricPadding extends PaddingAlgorithm {
+  final string getPaddingName() {
+    if exists(string n | n = this.getName() and isAsymmetricPaddingAlgorithm(n))
+    then isAsymmetricPaddingAlgorithm(result) and result = this.getName()
+    else result = unknownAlgorithm()
+  }
+
+  override string getAlgType() { result = getAsymmetricPaddingType() }
+}
+
+abstract class EllipticCurveAlgorithm extends AsymmetricAlgorithm {
+  final string getCurveName() {
+    if exists(string n | n = this.getName() and isEllipticCurveAlgorithm(n))
+    then isEllipticCurveAlgorithm(result) and result = this.getName()
+    else result = unknownAlgorithm()
+  }
+
+  final int getCurveBitSize() { isEllipticCurveAlgorithm(this.getCurveName(), result) }
+
+  override string getAlgType() { result = getEllipticCurveType() }
+}
+
+abstract class BlockModeAlgorithm extends CryptographicAlgorithm {
+  final string getBlockModeName() {
+    if exists(string n | n = this.getName() and isCipherBlockModeAlgorithm(n))
+    then isCipherBlockModeAlgorithm(result) and result = this.getName()
+    else result = unknownAlgorithm()
+  }
+
+  /**
+   * Gets the source of the IV configuration.
+   */
+  abstract Expr getIVorNonce();
+
+  final predicate hasIVorNonce() { exists(this.getIVorNonce()) }
+
+  override string getAlgType() { result = getCipherBlockModeType() }
+}
+
+// abstract class KeyWrapOperation extends CryptographicOperation{
+// }
+abstract class AuthenticatedEncryptionAlgorithm extends SymmetricEncryptionAlgorithm {
+  final string getAuthticatedEncryptionName() {
+    if exists(string n | n = this.getName() and isSymmetricEncryptionAlgorithm(n))
+    then isSymmetricEncryptionAlgorithm(result) and result = this.getName()
+    else result = unknownAlgorithm()
+  }
+}
+
+abstract class KeyExchangeAlgorithm extends AsymmetricAlgorithm {
+  final string getKeyExchangeName() {
+    if exists(string n | n = this.getName() and isKeyExchangeAlgorithm(n))
+    then isKeyExchangeAlgorithm(result) and result = this.getName()
+    else result = unknownAlgorithm()
+  }
+
+  override string getAlgType() { result = getKeyExchangeType() }
+}
+
+abstract class SigningAlgorithm extends AsymmetricAlgorithm {
+  final string getSigningName() {
+    if exists(string n | n = this.getName() and isSignatureAlgorithm(n))
+    then isSignatureAlgorithm(result) and result = this.getName()
+    else result = unknownAlgorithm()
+  }
+
+  override string getAlgType() { result = getSignatureType() }
+}

cpp/ql/lib/experimental/cryptography/modules/OpenSSL.qll

@@ -0,0 +1,718 @@
+import cpp
+import experimental.cryptography.CryptoAlgorithmNames
+import experimental.cryptography.CryptoArtifact
+import experimental.cryptography.utils.OpenSSL.CryptoFunction
+import experimental.cryptography.utils.OpenSSL.AlgorithmSink
+import experimental.cryptography.utils.OpenSSL.PassthroughFunction
+import experimental.cryptography.utils.OpenSSL.CryptoAlgorithm
+import experimental.cryptography.CryptoArtifact
+// import semmle.code.cpp.ir.dataflow.TaintTracking
+import semmle.code.cpp.ir.dataflow.DataFlow
+
+/**
+ *  Problematic case in OpenSSL speed.c
+ *    static const char *names[ALGOR_NUM] = {
+ *        "md2", "mdc2", "md4", "md5", "sha1", "rmd160",
+ *        "sha256", "sha512", "whirlpool", "hmac(md5)",
+ *        "des-cbc", "des-ede3", "rc4", "idea-cbc", "seed-cbc",
+ *        "rc2-cbc", "rc5-cbc", "blowfish", "cast-cbc",
+ *        "aes-128-cbc", "aes-192-cbc", "aes-256-cbc",
+ *        "camellia-128-cbc", "camellia-192-cbc", "camellia-256-cbc",
+ *        "evp", "ghash", "rand", "cmac"
+ *    };
+ *
+ *    Every entry is considered a block mode, hash, and symmetric encryption algorithm
+ *    getEncryptionName for example, will return unknown
+ */
+predicate nodeToExpr(DataFlow::Node node, Expr e) {
+  e = node.asExpr() or e = node.asIndirectArgument()
+}
+
+Expr getExprFromNode(DataFlow::Node node) { nodeToExpr(node, result) }
+
+DataFlow::Node getNodeFromExpr(Expr e) { nodeToExpr(result, e) }
+
+predicate isEVP_PKEY_CTX(Type t) { t.getUnderlyingType().stripType().getName() = "evp_pkey_ctx_st" }
+
+/**
+ * An expression representing an EVP_PKEY_CTX* at the location of a
+ * known AlgorithmSinkArgument.
+ * The EVP_PKEY_CTX* represents the location where the CTX is tied to the algorithm,
+ * and can be used as a source for tracing EVP_PKEY_CTX to other operations.
+ */
+class Known_EVP_PKEY_CTX_Ptr_Source extends Expr {
+  Known_EVP_PKEY_CTX_Ptr_Source() {
+    isEVP_PKEY_CTX(this.getUnderlyingType()) and
+    this.getUnderlyingType() instanceof PointerType and
+    exists(AlgorithmSinkArgument arg, Call sinkCall |
+      arg.getSinkCall() = sinkCall and
+      sinkCall.getAnArgument() = this
+      or
+      this = sinkCall
+    )
+  }
+}
+
+// module CTXFlow implements DataFlow::ConfigSig{
+//     predicate isSource(DataFlow::Node source) {
+//         // ASSUMPTION: at a sink, an algorithm is converted into a CTX through a return of the call only
+//         //             and is the primary source of interest for CTX tracing
+//         source.asExpr() instanceof AlgorithmSinkArgument
+//     }
+//     predicate isSink(DataFlow::Node sink){
+//         sink.asExpr() instanceof CTXSink
+//     }
+//     predicate isAdditionalFlowStep(DataFlow::Node node1, DataFlow::Node node2) {
+//         //  cls.getName() = "asn1_object_st" flow out on any EVP_PKEY_CTX which is "evp_pkey_ctx_st"
+//         exists(Call c |
+//             isEVP_PKEY_CTX(c.getUnderlyingType()) and
+//             node1.asExpr() = c.getAnArgument() and c = node2.asExpr())
+//     }
+// }
+// module CTXFlowConfig = DataFlow::Global<CTXFlow>;
+// TODO: currently only handles tracing from literals to sinks
+module LiteralAlgorithmTracerConfig implements DataFlow::ConfigSig {
+  predicate isSource(DataFlow::Node source) {
+    source.asExpr() instanceof Literal and
+    // Optimization to reduce literal tracing on integers to only those that are known/relevant NIDs.
+    (
+      exists(source.asExpr().getValue().toInt())
+      implies
+      source.asExpr().getValue().toInt() < getNIDMax()
+    ) and
+    // False positives observed inside OBJ_nid2* and OBJ_sn2* functions where NULL is a possible assignment.
+    // While this is a concern, it only occurs if the object being referenced is NULL to begin with
+    // Perhaps a different query should be used to find these caes if they represent a threat.
+    // Filter out any open ssl function source in a function namae Obj_*
+    // False positives in OpenSSL also observed for CRYPTO_strndup (filtering any CRYPTO_* function)
+    // due to setting a null byte in the string
+    (
+      isPossibleOpenSSLFunction(source.getEnclosingCallable())
+      implies
+      (
+        not source.getEnclosingCallable().getName().matches("OBJ_%") and
+        not source.getEnclosingCallable().getName().matches("CRYPTO_%")
+      )
+    )
+  }
+
+  predicate isSink(DataFlow::Node sink) {
+    // A sink is a call to a function that takes an algorithm as an argument
+    // must include checks for asIndirectArgument since the input may be a pointer to an object
+    // and the member of the object holds the algorithm on the trace.
+    getExprFromNode(sink) instanceof AlgorithmSinkArgument
+  }
+
+  predicate isAdditionalFlowStep(DataFlow::Node node1, DataFlow::Node node2) {
+    knownPassThroughStep(node1, node2)
+  }
+
+  predicate isBarrier(DataFlow::Node node) {
+    // If the node is the 'next' argument of a isCallPassThrough, it is only allowed if it is an out parameter
+    // i.e., a defining argument. This barrier says that if the node is an expression not an out parameter, it is filtered.
+    // Out arguments will not be filtered.
+    exists(Call c | knownPassthoughCall(c, _, node.asExpr()) and c.getAnArgument() = node.asExpr())
+    or
+    // False positive reducer, don't flow out through argv
+    node.asVariable().hasName("argv")
+    or
+    node.asIndirectVariable().hasName("argv")
+  }
+
+  predicate allowImplicitRead(DataFlow::Node node, DataFlow::ContentSet c) {
+    // Assume a read on crypto identifying field for any object of type asn1_object_st (i.e., ASN1_OBJECT)
+    exists(Class cls | cls.getName() = "asn1_object_st" |
+      node.getType().getUnspecifiedType().stripType() = cls and
+      c.(DataFlow::FieldContent).getField() = cls.getAMember() and
+      c.(DataFlow::FieldContent).getField().getName() in ["nid", "sn", "ln"]
+    )
+  }
+}
+
+module LiteralAlgorithmTracer = DataFlow::Global<LiteralAlgorithmTracerConfig>;
+
+/**
+ * `source` is an expression that is a source of an algorithm of type `algType`.
+ * `algType` may be `UNKONWN`.
+ *  See CryptoAlgorithmNames for other possible values of `algType`.
+ */
+bindingset[sinkAlgType]
+predicate hasLiteralPathToAlgSink(DataFlow::Node source, DataFlow::Node sink, string sinkAlgType) {
+  LiteralAlgorithmTracer::flow(source, sink) and
+  getExprFromNode(sink).(AlgorithmSinkArgument).algType() = sinkAlgType
+}
+
+private predicate knownTracedAlgorithm(Literal e, string srcSinkType) {
+  knownTracedAlgorithm(e, srcSinkType, srcSinkType)
+}
+
+private predicate knownTracedAlgorithm(Literal e, string srcType, string sinkType) {
+  resolveAlgorithmFromLiteral(e, _, srcType) and
+  hasLiteralPathToAlgSink(DataFlow::exprNode(e), _, sinkType) and
+  isKnownType(sinkType) and
+  isKnownType(srcType)
+}
+
+private predicate unknownTracedLiteralAlgorithm(Literal e, string srcSinkType) {
+  // Asymmetric special case:
+  // Since asymmetric algorithm sinks are used for various categories of asymmetric algorithms
+  // an asymmetric algorithm is only unknown if there is no trace from any asymmetric type to the given srcSinkType sink
+  if getAsymmetricType() = srcSinkType
+  then forall(string t | t = getAsymmetricType() | unknownTracedLiteralAlgorithm(e, t, srcSinkType))
+  else unknownTracedLiteralAlgorithm(e, srcSinkType, srcSinkType)
+}
+
+private predicate unknownTracedLiteralAlgorithm(Literal e, string srcType, string sinkType) {
+  // the literal resolves to an algorithm, but not to the sinktype
+  // or generally doesn't resolve to any algorithm type
+  // this case covers 'nonsense' cases e.g., use RSA for symmetric encryption
+  not resolveAlgorithmFromLiteral(e, _, srcType) and
+  isValidAlgorithmLiteral(e) and
+  hasLiteralPathToAlgSink(DataFlow::exprNode(e), _, sinkType) and
+  isKnownType(sinkType) and
+  isKnownType(srcType)
+}
+
+private predicate unknownTracedNonLiteralAlgorithm(AlgorithmSinkArgument e, string srcSinkType) {
+  // Asymmetric special case:
+  // Since asymmetric algorithm sinks are used for various categories of asymmetric algorithms
+  // an asymmetric algorithm is only unknown if there is no trace from any asymmetric type to the given srcSinkType sink
+  if getAsymmetricType() = srcSinkType
+  then
+    forall(string t | t = getAsymmetricType() | unknownTracedNonLiteralAlgorithm(e, t, srcSinkType))
+  else unknownTracedNonLiteralAlgorithm(e, srcSinkType, srcSinkType)
+}
+
+private predicate unknownTracedNonLiteralAlgorithm(
+  AlgorithmSinkArgument e, string srcType, string sinkType
+) {
+  not hasLiteralPathToAlgSink(_, getNodeFromExpr(e), srcType) and
+  LiteralAlgorithmTracerConfig::isSink(getNodeFromExpr(e)) and
+  e.algType() = sinkType and
+  isKnownType(srcType) and
+  isKnownType(sinkType)
+}
+
+private predicate functionAlgorithm(Call c, string algType) {
+  isOpenSSLCryptoFunctionCall(c, _, algType)
+}
+
+abstract class OpenSSLTracedAlgorithm extends CryptographicAlgorithm {
+  override string getName() { resolveAlgorithmFromLiteral(this, result, this.getAlgType()) }
+
+  override Expr configurationSink() {
+    exists(DataFlow::Node sink |
+      hasLiteralPathToAlgSink(DataFlow::exprNode(this), sink, this.getAlgType())
+    |
+      result = getExprFromNode(sink)
+    )
+  }
+}
+
+abstract class OpenSSLFunctionAlgorithm extends CryptographicAlgorithm {
+  override string getName() { isOpenSSLCryptoFunctionCall(this, result, this.getAlgType()) }
+
+  override Expr configurationSink() { result = this }
+}
+
+abstract class OpenSSLUnknownTracedLiteralAlgorithm extends CryptographicAlgorithm {
+  override string getName() { result = unknownAlgorithm() }
+
+  override Expr configurationSink() {
+    exists(DataFlow::Node sink |
+      hasLiteralPathToAlgSink(DataFlow::exprNode(this), sink, this.getAlgType())
+    |
+      result = getExprFromNode(sink)
+    )
+  }
+}
+
+abstract class OpenSSLUnknownTracedNonLiteralAlgorithm extends CryptographicAlgorithm {
+  override string getName() { result = unknownAlgorithm() }
+
+  override Expr configurationSink() { result = this }
+}
+
+module SymmetricEncryption {
+  abstract class OpenSSLSymmetricEncryptionAlgorithm extends SymmetricEncryptionAlgorithm { }
+
+  class OpenSSLSymmetricEncryptionTracedAlgorithm extends OpenSSLTracedAlgorithm,
+    OpenSSLSymmetricEncryptionAlgorithm
+  {
+    OpenSSLSymmetricEncryptionTracedAlgorithm() {
+      knownTracedAlgorithm(this, getSymmetricEncryptionType())
+    }
+  }
+
+  class OpenSSLSymmetricEncryptionFunctionAlgorithm extends OpenSSLFunctionAlgorithm,
+    OpenSSLSymmetricEncryptionAlgorithm
+  {
+    OpenSSLSymmetricEncryptionFunctionAlgorithm() {
+      functionAlgorithm(this, getSymmetricEncryptionType())
+    }
+  }
+
+  class OpenSSLSymmetricEncryptionTracedUnknownLiteralAlgorithm extends OpenSSLUnknownTracedLiteralAlgorithm,
+    OpenSSLSymmetricEncryptionAlgorithm
+  {
+    OpenSSLSymmetricEncryptionTracedUnknownLiteralAlgorithm() {
+      unknownTracedLiteralAlgorithm(this, getSymmetricEncryptionType())
+    }
+  }
+
+  class OpenSSLSymmetricEncryptionUnknownNonLiteralTracedAlgorithm extends OpenSSLUnknownTracedNonLiteralAlgorithm,
+    OpenSSLSymmetricEncryptionAlgorithm
+  {
+    OpenSSLSymmetricEncryptionUnknownNonLiteralTracedAlgorithm() {
+      unknownTracedNonLiteralAlgorithm(this, getSymmetricEncryptionType())
+    }
+  }
+}
+
+module BlockModes {
+  /**
+   * In OpenSSL, block modes are associated directly with symmetric encryption algorithms.
+   * As such, OpenSSLBLockModes are modeled as extensions of any openssl symmetric encryption algorithm
+   */
+  class OpenSSLBlockModeAlgorithm extends BlockModeAlgorithm, Expr instanceof SymmetricEncryption::OpenSSLSymmetricEncryptionAlgorithm
+  {
+    OpenSSLBlockModeAlgorithm() {
+      //two cases, either the block mode is a literal or it is a function call
+      resolveAlgorithmFromLiteral(this, _, "BLOCK_MODE")
+      or
+      isOpenSSLCryptoFunctionCall(this, _, "BLOCK_MODE")
+    }
+
+    override string getName() {
+      resolveAlgorithmFromLiteral(this, result, "BLOCK_MODE")
+      or
+      isOpenSSLCryptoFunctionCall(this, result, "BLOCK_MODE")
+    }
+
+    override Expr configurationSink() {
+      result = this.(SymmetricEncryption::OpenSSLSymmetricEncryptionAlgorithm).configurationSink()
+    }
+
+    override Expr getIVorNonce() {
+      // TODO
+      none()
+    }
+  }
+
+  class UnknownOpenSSLBlockModeAlgorithm extends BlockModeAlgorithm, Expr instanceof SymmetricEncryption::OpenSSLSymmetricEncryptionAlgorithm
+  {
+    UnknownOpenSSLBlockModeAlgorithm() {
+      //two cases, either the block mode is a literal or it is a function call
+      not resolveAlgorithmFromLiteral(this, _, "BLOCK_MODE") and
+      not isOpenSSLCryptoFunctionCall(this, _, "BLOCK_MODE")
+    }
+
+    override string getName() { result = unknownAlgorithm() }
+
+    override Expr configurationSink() {
+      result = this.(SymmetricEncryption::OpenSSLSymmetricEncryptionAlgorithm).configurationSink()
+    }
+
+    override Expr getIVorNonce() { none() }
+  }
+}
+
+module Hashes {
+  abstract class OpenSSLHashAlgorithm extends HashAlgorithm { }
+
+  class OpenSSLHashTracedAlgorithm extends OpenSSLTracedAlgorithm, OpenSSLHashAlgorithm {
+    OpenSSLHashTracedAlgorithm() { knownTracedAlgorithm(this, getHashType()) }
+  }
+
+  class OpenSSLHashFunctionAlgorithm extends OpenSSLFunctionAlgorithm, OpenSSLHashAlgorithm {
+    OpenSSLHashFunctionAlgorithm() { functionAlgorithm(this, getHashType()) }
+  }
+
+  class OpenSSLHashTracedUnknownLiteralAlgorithm extends OpenSSLUnknownTracedLiteralAlgorithm,
+    OpenSSLHashAlgorithm
+  {
+    OpenSSLHashTracedUnknownLiteralAlgorithm() {
+      unknownTracedLiteralAlgorithm(this, getHashType())
+    }
+  }
+
+  class OpenSSLHashUnknownNonLiteralTracedAlgorithm extends OpenSSLUnknownTracedNonLiteralAlgorithm,
+    OpenSSLHashAlgorithm
+  {
+    OpenSSLHashUnknownNonLiteralTracedAlgorithm() {
+      unknownTracedNonLiteralAlgorithm(this, getHashType())
+    }
+  }
+
+  class OpenSSLNullHash extends HashAlgorithm {
+    OpenSSLNullHash() {
+      exists(Call c |
+        this = c and
+        isPossibleOpenSSLFunction(c.getTarget()) and
+        c.getTarget().getName() in ["EVP_md_null"]
+      )
+    }
+
+    override string getName() { result = unknownAlgorithm() }
+
+    override Expr configurationSink() { result = this }
+  }
+}
+
+module EllipticCurves {
+  // TODO: need to address EVP_PKEY_Q_keygen where the type is "EC" but the curve is UNKNOWN?
+  class OpenSSLEllipticCurveTracedAlgorithm extends OpenSSLTracedAlgorithm, EllipticCurveAlgorithm {
+    OpenSSLEllipticCurveTracedAlgorithm() { knownTracedAlgorithm(this, getEllipticCurveType()) }
+  }
+
+  class OpenSSLEllipticCurveFunctionAlgorithm extends OpenSSLFunctionAlgorithm,
+    EllipticCurveAlgorithm
+  {
+    OpenSSLEllipticCurveFunctionAlgorithm() { functionAlgorithm(this, getEllipticCurveType()) }
+  }
+
+  class OpenSSLEllipticCurveTracedUnknownLiteralAlgorithm extends OpenSSLUnknownTracedLiteralAlgorithm,
+    EllipticCurveAlgorithm
+  {
+    OpenSSLEllipticCurveTracedUnknownLiteralAlgorithm() {
+      unknownTracedLiteralAlgorithm(this, getEllipticCurveType())
+    }
+  }
+
+  class OpenSSLEllipticCurvehUnknownNonLiteralTracedAlgorithm extends OpenSSLUnknownTracedNonLiteralAlgorithm,
+    EllipticCurveAlgorithm
+  {
+    OpenSSLEllipticCurvehUnknownNonLiteralTracedAlgorithm() {
+      unknownTracedNonLiteralAlgorithm(this, getEllipticCurveType())
+    }
+  }
+
+  // https://www.openssl.org/docs/manmaster/man3/EC_KEY_new_ex.html
+  class OpenSSLNullEllipticCurve extends EllipticCurveAlgorithm {
+    OpenSSLNullEllipticCurve() {
+      exists(Call c |
+        this = c and
+        isPossibleOpenSSLFunction(c.getTarget()) and
+        c.getTarget().getName() in ["EC_KEY_new", "EC_KEY_new_ex"]
+      )
+    }
+
+    override string getName() { result = unknownAlgorithm() }
+
+    override Expr configurationSink() { result = this }
+  }
+}
+
+module AsymmetricEncryption {
+  class OpenSSLAsymmetricEncryptionTracedAlgorithm extends OpenSSLTracedAlgorithm,
+    AsymmetricEncryptionAlgorithm
+  {
+    OpenSSLAsymmetricEncryptionTracedAlgorithm() {
+      knownTracedAlgorithm(this, getAsymmetricEncryptionType())
+    }
+  }
+
+  class OpenSSLAsymmetricEncryptionFunctionAlgorithm extends OpenSSLFunctionAlgorithm,
+    AsymmetricEncryptionAlgorithm
+  {
+    OpenSSLAsymmetricEncryptionFunctionAlgorithm() {
+      functionAlgorithm(this, getAsymmetricEncryptionType())
+    }
+  }
+
+  class OpenSSLAsymmetricEncryptionTracedUnknownLiteralAlgorithm extends OpenSSLUnknownTracedLiteralAlgorithm,
+    AsymmetricEncryptionAlgorithm
+  {
+    OpenSSLAsymmetricEncryptionTracedUnknownLiteralAlgorithm() {
+      unknownTracedLiteralAlgorithm(this, getAsymmetricEncryptionType())
+    }
+  }
+
+  class OpenSSLAsymmetricEncryptionUnknownNonLiteralTracedAlgorithm extends OpenSSLUnknownTracedNonLiteralAlgorithm,
+    AsymmetricEncryptionAlgorithm
+  {
+    OpenSSLAsymmetricEncryptionUnknownNonLiteralTracedAlgorithm() {
+      unknownTracedNonLiteralAlgorithm(this, getAsymmetricEncryptionType())
+    }
+  }
+}
+
+module SigningAlgorithms {
+  class OpenSSLSignatureTracedAlgorithm extends OpenSSLTracedAlgorithm, SigningAlgorithm {
+    OpenSSLSignatureTracedAlgorithm() { knownTracedAlgorithm(this, getSignatureType()) }
+  }
+
+  class OpenSSLSignatureFunctionAlgorithm extends OpenSSLFunctionAlgorithm, SigningAlgorithm {
+    OpenSSLSignatureFunctionAlgorithm() { functionAlgorithm(this, getSignatureType()) }
+  }
+
+  class OpenSSLSignatureTracedUnknownLiteralAlgorithm extends OpenSSLUnknownTracedLiteralAlgorithm,
+    SigningAlgorithm
+  {
+    OpenSSLSignatureTracedUnknownLiteralAlgorithm() {
+      unknownTracedLiteralAlgorithm(this, getSignatureType())
+    }
+  }
+
+  class OpenSSLSignatureUnknownNonLiteralTracedAlgorithm extends OpenSSLUnknownTracedNonLiteralAlgorithm,
+    SigningAlgorithm
+  {
+    OpenSSLSignatureUnknownNonLiteralTracedAlgorithm() {
+      unknownTracedNonLiteralAlgorithm(this, getSignatureType())
+    }
+  }
+}
+
+module KeyExchange {
+  class OpenSSLKeyExchangeTracedAlgorithm extends OpenSSLTracedAlgorithm, KeyExchangeAlgorithm {
+    OpenSSLKeyExchangeTracedAlgorithm() { knownTracedAlgorithm(this, getKeyExchangeType()) }
+  }
+
+  class OpenSSLKeyExchangeFunctionAlgorithm extends OpenSSLFunctionAlgorithm, KeyExchangeAlgorithm {
+    OpenSSLKeyExchangeFunctionAlgorithm() { functionAlgorithm(this, getKeyExchangeType()) }
+  }
+
+  class OpenSSLKeyExchangeTracedUnknownLiteralAlgorithm extends OpenSSLUnknownTracedLiteralAlgorithm,
+    KeyExchangeAlgorithm
+  {
+    OpenSSLKeyExchangeTracedUnknownLiteralAlgorithm() {
+      unknownTracedLiteralAlgorithm(this, getKeyExchangeType())
+    }
+  }
+
+  class OpenSSLKeyExchangeUnknownNonLiteralTracedAlgorithm extends OpenSSLUnknownTracedNonLiteralAlgorithm,
+    KeyExchangeAlgorithm
+  {
+    OpenSSLKeyExchangeUnknownNonLiteralTracedAlgorithm() {
+      unknownTracedNonLiteralAlgorithm(this, getKeyExchangeType())
+    }
+  }
+}
+
+module KeyGeneration {
+  /**
+   * Functions that explicitly set key generation parameters.
+   * `sizeInd` is the parameter specifying the size of the key.
+   * `outInd` is the parameter or return value that the key is written to.
+   * `outInd` is -1 if the key is written to the return value.
+   */
+  predicate isAsymmetricKeyGenExplicitAlgorithm(Function func, int sizeInd, int outInd) {
+    isPossibleOpenSSLFunction(func) and
+    exists(string name | func.hasGlobalName(name) |
+      name in [
+          "EVP_PKEY_CTX_set_dsa_paramgen_bits", "DSA_generate_parameters_ex",
+          "EVP_PKEY_CTX_set_rsa_keygen_bits", "RSA_generate_key_ex", "RSA_generate_key_fips",
+          "EVP_PKEY_CTX_set_dh_paramgen_prime_len", "DH_generate_parameters_ex"
+        ] and
+      sizeInd = 1 and
+      outInd = 0
+      or
+      name in ["DSA_generate_parameters", "RSA_generate_key", "DH_generate_parameters"] and
+      sizeInd = 0 and
+      outInd = -1
+    ) and
+    exists(Type t |
+      (
+        if sizeInd = -1
+        then t = func.getType().getUnderlyingType()
+        else t = func.getParameter(sizeInd).getUnderlyingType()
+      ) and
+      t instanceof IntegralType and
+      not t instanceof CharType
+    )
+  }
+
+  module AsymExplicitAlgKeyLengthFlowConfig implements DataFlow::ConfigSig {
+    predicate isSource(DataFlow::Node node) {
+      // Optimizations to avoid tracing all integers
+      node.asExpr().(Literal).getValue().toInt() > 0 and // exclude sentinel values
+      node.asExpr().(Literal).getValue().toInt() < 8500
+    }
+
+    predicate isSink(DataFlow::Node node) {
+      exists(FunctionCall c, int sizeInd |
+        isAsymmetricKeyGenExplicitAlgorithm(c.getTarget(), sizeInd, _) and
+        c.getArgument(sizeInd) = node.asExpr()
+      )
+    }
+  }
+
+  module AsymExplicitAlgKeyLengthFlow = DataFlow::Global<AsymExplicitAlgKeyLengthFlowConfig>;
+
+  class OpenSSLAsymmetricKeyGenTiedToAlgorithm extends AsymmetricKeyGeneration {
+    OpenSSLAsymmetricKeyGenTiedToAlgorithm() {
+      exists(Call c |
+        this = c and
+        isPossibleOpenSSLFunction(c.getTarget()) and
+        isAsymmetricKeyGenExplicitAlgorithm(c.getTarget(), _, _)
+      )
+    }
+
+    override CryptographicAlgorithm getAlgorithm() { result = this }
+
+    override Expr getKeyConfigurationSource(CryptographicAlgorithm alg) {
+      alg = this and
+      exists(int sizeInd |
+        isAsymmetricKeyGenExplicitAlgorithm(this.getTarget(), sizeInd, _) and
+        AsymExplicitAlgKeyLengthFlow::flow(DataFlow::exprNode(result),
+          DataFlow::exprNode(this.getArgument(sizeInd)))
+      )
+    }
+  }
+
+  module Length_to_RSA_EVP_PKEY_Q_keygen_Config implements DataFlow::ConfigSig {
+    predicate isSource(DataFlow::Node node) {
+      // Optimizations to avoid tracing all integers
+      node.asExpr().(Literal).getValue().toInt() > 0 and // exclude sentinel values
+      node.asExpr().(Literal).getValue().toInt() < 5000
+    }
+
+    predicate isSink(DataFlow::Node node) {
+      exists(FunctionCall c |
+        c.getTarget().getName() = "EVP_PKEY_Q_keygen" and
+        isPossibleOpenSSLFunction(c.getTarget()) and
+        c.getArgument(3) = node.asExpr()
+      )
+    }
+  }
+
+  module Length_to_RSA_EVP_PKEY_Q_keygen_Flow =
+    DataFlow::Global<Length_to_RSA_EVP_PKEY_Q_keygen_Config>;
+
+  class OpenSSL_RSA_EVP_PKEY_Q_keygen extends AsymmetricKeyGeneration {
+    OpenSSL_RSA_EVP_PKEY_Q_keygen() {
+      exists(Call c |
+        this = c and
+        isPossibleOpenSSLFunction(c.getTarget()) and
+        this.getTarget().getName() = "EVP_PKEY_Q_keygen" and
+        this.getArgument(3).getUnderlyingType() instanceof IntegralType
+      )
+    }
+
+    override CryptographicAlgorithm getAlgorithm() {
+      result.configurationSink().(AlgorithmSinkArgument).getSinkCall() = this
+    }
+
+    override Expr getKeyConfigurationSource(CryptographicAlgorithm alg) {
+      alg = this.getAlgorithm() and
+      Length_to_RSA_EVP_PKEY_Q_keygen_Flow::flow(DataFlow::exprNode(result),
+        DataFlow::exprNode(this.getArgument(3)))
+    }
+  }
+
+  predicate isKeyGenOperationWithNoSize(Function func) {
+    isPossibleOpenSSLFunction(func) and
+    exists(string name | func.hasGlobalName(name) |
+      name in ["EVP_PKEY_keygen", "DSA_generate_key", "DH_generate_key", "EVP_PKEY_generate"]
+    )
+  }
+
+  module KeyGenKeySizeInitToKeyGenConfig implements DataFlow::ConfigSig {
+    predicate isSource(DataFlow::Node node) {
+      exists(Call c, Function func, int outInd |
+        isAsymmetricKeyGenExplicitAlgorithm(func, _, outInd) and
+        c.getTarget() = func
+      |
+        if outInd = -1 then node.asExpr() = c else node.asExpr() = c.getArgument(outInd)
+      )
+    }
+
+    predicate isSink(DataFlow::Node node) {
+      exists(Call c |
+        isKeyGenOperationWithNoSize(c.getTarget()) and c.getAnArgument() = node.asExpr()
+      )
+    }
+  }
+
+  module KeyGenKeySizeInitToKeyGenFlow = DataFlow::Global<KeyGenKeySizeInitToKeyGenConfig>;
+
+  predicate isEVP_PKEY_CTX_Source(DataFlow::Node node, CryptographicAlgorithm alg) {
+    exists(Call c |
+      alg.configurationSink().(AlgorithmSinkArgument).getSinkCall() = c and
+      (
+        node.asExpr() = c
+        or
+        node.asExpr() = c.getAnArgument()
+        or
+        node.asDefiningArgument() = c.getAnArgument()
+      )
+    ) and
+    (
+      node.asExpr() instanceof Known_EVP_PKEY_CTX_Ptr_Source
+      or
+      node.asDefiningArgument() instanceof Known_EVP_PKEY_CTX_Ptr_Source
+    )
+  }
+
+  predicate isKeyGen_EVP_PKEY_CTX_Sink(DataFlow::Node node, Call c) {
+    isKeyGenOperationWithNoSize(c.getTarget()) and nodeToExpr(node, c.getAnArgument())
+  }
+
+  /**
+   * Trace from EVP_PKEY_CTX* at algorithm sink to keygen,
+   * users can then extrapolatae the matching algorithm from the alg sink to the keygen
+   */
+  module EVP_PKEY_CTX_Ptr_Source_to_KeyGenOperationWithNoSize implements DataFlow::ConfigSig {
+    predicate isSource(DataFlow::Node source) { isEVP_PKEY_CTX_Source(source, _) }
+
+    predicate isSink(DataFlow::Node sink) { isKeyGen_EVP_PKEY_CTX_Sink(sink, _) }
+  }
+
+  module EVP_PKEY_CTX_Ptr_Source_to_KeyGenOperationWithNoSize_Flow =
+    DataFlow::Global<EVP_PKEY_CTX_Ptr_Source_to_KeyGenOperationWithNoSize>;
+
+  /**
+   * UNKNOWN key sizes to general purpose key generation functions (i.e., that take in no key size and assume
+   * is it set on context prior to the call). No path from a key configuration to these operations
+   * means the key size is UNKNOWN, or more precisely the key size is DEFAULT but
+   * the defaults can change with each version of OpenSSL, we simply assume the size is generally UNKNOWN.
+   * ASSUMPTION/TODO: we currently model all known locations where a key size is set explicitly.
+   *                 When a key is set implicitly, this usually means a key generation operation
+   *                 is called where the operation takes in no key size, and no flow to this operation
+   *                 initializes the context with a key size.
+   *                 Currently, without a definitive source (set of sources) to start tracing from, we cannot determine
+   *                 determine if a single path exists that initializes the context with a key size and another that doesn't.
+   *                 Rather than attempt to model all possible sources, we assume that if no path
+   *                 from a key config location reaches a generic key generation operation, then the key size is not set.
+   *                 NOTE: while this is true, it is possible a key size is set in one path, but not in another
+   *                 meaning this approach (and other similar approaches used in this model for UNKNOWN)
+   *                 can produce false negatives.
+   */
+  class OpenSSLDefaultKeyGeneration extends AsymmetricKeyGeneration {
+    OpenSSLDefaultKeyGeneration() {
+      // this is a call to a function matching isKeyGenOperationWithNoSize
+      // and there is no flow from a key configuration source to this call
+      exists(Call c |
+        this = c and
+        isKeyGenOperationWithNoSize(this.getTarget()) and
+        not exists(DataFlow::Node src, DataFlow::Node sink |
+          KeyGenKeySizeInitToKeyGenFlow::flow(src, sink) and
+          nodeToExpr(sink, this.getAnArgument())
+        )
+      )
+    }
+
+    override CryptographicAlgorithm getAlgorithm() {
+      if this.getTarget().getName() in ["DSA_generate_key", "DH_generate_key"]
+      then result = this
+      else
+        // NOTE/ASSUMPTION: EVP_PKEY_keygen, EVP_PKEY_generate assume only other possibilities,
+        //        each take in a CTX as the first arg, need to trace from an alg sink from this CTX param
+        // get every alg sink, get the corresponding call, trace out on any CTX type variable
+        // to the key gen
+        // NOTE: looking for any cryptographic algorithm tracing to the keygen to handle
+        //  any odd cases we aren't awaare of where keygen can be used for other algorithm types
+        exists(DataFlow::Node src, DataFlow::Node sink |
+          EVP_PKEY_CTX_Ptr_Source_to_KeyGenOperationWithNoSize_Flow::flow(src, sink) and
+          isEVP_PKEY_CTX_Source(src, result) and
+          isKeyGen_EVP_PKEY_CTX_Sink(sink, this)
+          // TODO: what if there is no CTX source? then the keygen becomes an UNKNOWN sink
+        )
+    }
+
+    /**
+     * For this class, there is no known configuration source for any algorithm
+     */
+    override Expr getKeyConfigurationSource(CryptographicAlgorithm alg) { none() }
+  }
+}

cpp/ql/lib/experimental/cryptography/utils/OpenSSL/AlgorithmSink.qll

@@ -0,0 +1,296 @@
+/**
+ * Predicates/classes for identifying algorithm sinks.
+ * An Algorithm Sink is a function that takes an algorithm as an argument.
+ * In particular, any function that takes in an algorithm that until the call
+ * the algorithm is not definitely known to be an algorithm (e.g., an integer used as an identifier to fetch an algorithm)
+ */
+
+//TODO: enforce a hierarchy of AlgorithmSinkArgument, e.g., so I can get all Asymmetric SinkArguments that includes all the strictly RSA etc.
+import cpp
+import experimental.cryptography.utils.OpenSSL.LibraryFunction
+import experimental.cryptography.CryptoAlgorithmNames
+
+predicate isAlgorithmSink(AlgorithmSinkArgument arg, string algType) { arg.algType() = algType }
+
+abstract class AlgorithmSinkArgument extends Expr {
+  AlgorithmSinkArgument() {
+    exists(Call c | c.getAnArgument() = this and openSSLLibraryFunc(c.getTarget()))
+  }
+
+  /**
+   * Gets the function call in which the argument exists
+   */
+  Call getSinkCall() { result.getAnArgument() = this }
+
+  abstract string algType();
+}
+
+// https://www.openssl.org/docs/manmaster/man3/EVP_CIPHER_fetch.html
+predicate cipherAlgorithmSink(string funcName, int argInd) {
+  funcName in ["EVP_get_cipherbyname", "EVP_get_cipherbynid", "EVP_get_cipherbyobj"] and argInd = 0
+  or
+  funcName = "EVP_CIPHER_fetch" and argInd = 1
+}
+
+class CipherAlgorithmSink extends AlgorithmSinkArgument {
+  CipherAlgorithmSink() {
+    exists(Call c, string funcName, int argInd |
+      funcName = c.getTarget().getName() and this = c.getArgument(argInd)
+    |
+      cipherAlgorithmSink(funcName, argInd)
+    )
+  }
+
+  override string algType() { result = getSymmetricEncryptionType() }
+}
+
+// https://www.openssl.org/docs/manmaster/man3/EVP_MAC_fetch
+predicate macAlgorithmSink(string funcName, int argInd) {
+  (funcName = "EVP_MAC_fetch" and argInd = 1)
+}
+
+class MACAlgorithmSink extends AlgorithmSinkArgument {
+  MACAlgorithmSink() {
+    exists(Call c, string funcName, int argInd |
+      funcName = c.getTarget().getName() and this = c.getArgument(argInd)
+    |
+      macAlgorithmSink(funcName, argInd)
+    )
+  }
+
+  override string algType() { result = "TBD" }
+}
+
+// https://www.openssl.org/docs/manmaster/man3/EVP_MD_fetch
+predicate messageDigestAlgorithmSink(string funcName, int argInd) {
+  funcName in ["EVP_get_digestbyname", "EVP_get_digestbynid", "EVP_get_digestbyobj"] and argInd = 0
+  or
+  funcName = "EVP_MD_fetch" and argInd = 1
+}
+
+class MessageDigestAlgorithmSink extends AlgorithmSinkArgument {
+  MessageDigestAlgorithmSink() {
+    exists(Call c, string funcName, int argInd |
+      funcName = c.getTarget().getName() and this = c.getArgument(argInd)
+    |
+      messageDigestAlgorithmSink(funcName, argInd)
+    )
+  }
+
+  override string algType() { result = getHashType() }
+}
+
+// https://www.openssl.org/docs/manmaster/man3/EVP_KEYEXCH_fetch
+// https://www.openssl.org/docs/manmaster/man3/EVP_KEM_fetch
+predicate keyExchangeAlgorithmSink(string funcName, int argInd) {
+  funcName = "EVP_KEYEXCH_fetch" and argInd = 1
+  or
+  funcName = "EVP_KEM_fetch" and argInd = 1
+}
+
+class KeyExchangeAlgorithmSink extends AlgorithmSinkArgument {
+  KeyExchangeAlgorithmSink() {
+    exists(Call c, string funcName, int argInd |
+      funcName = c.getTarget().getName() and this = c.getArgument(argInd)
+    |
+      keyExchangeAlgorithmSink(funcName, argInd)
+    )
+  }
+
+  override string algType() { result = getKeyExchangeType() }
+}
+
+// https://www.openssl.org/docs/manmaster/man3/EVP_KEYMGMT_fetch
+predicate keyManagementAlgorithmSink(string funcName, int argInd) {
+  funcName = "EVP_KEYMGMT_fetch" and argInd = 1
+}
+
+class KeyManagementAlgorithmSink extends AlgorithmSinkArgument {
+  KeyManagementAlgorithmSink() {
+    exists(Call c, string funcName, int argInd |
+      funcName = c.getTarget().getName() and this = c.getArgument(argInd)
+    |
+      keyManagementAlgorithmSink(funcName, argInd)
+    )
+  }
+
+  override string algType() { result = "TBD" }
+}
+
+// https://www.openssl.org/docs/manmaster/man3/EVP_KDF
+predicate keyDerivationAlgorithmSink(string funcName, int argInd) {
+  funcName = "EVP_KDF_fetch" and argInd = 1
+}
+
+class KeyDerivationAlgorithmSink extends AlgorithmSinkArgument {
+  KeyDerivationAlgorithmSink() {
+    exists(Call c, string funcName, int argInd |
+      funcName = c.getTarget().getName() and this = c.getArgument(argInd)
+    |
+      keyDerivationAlgorithmSink(funcName, argInd)
+    )
+  }
+
+  override string algType() { result = getKeyDerivationType() }
+}
+
+// https://www.openssl.org/docs/manmaster/man3/EVP_ASYM_CIPHER_fetch
+// https://www.openssl.org/docs/manmaster/man3/EVP_PKEY_CTX_new_id
+// https://www.openssl.org/docs/manmaster/man3/EVP_PKEY_new_CMAC_key.html
+predicate asymmetricCipherAlgorithmSink(string funcName, int argInd) {
+  funcName = "EVP_ASYM_CIPHER_fetch" and argInd = 1
+  or
+  funcName = "EVP_PKEY_new_CMAC_key" and argInd = 3
+  // NOTE: other cases are handled by AsymmetricAlgorithmSink
+}
+
+class AsymmetricCipherAlgorithmSink extends AlgorithmSinkArgument {
+  AsymmetricCipherAlgorithmSink() {
+    exists(Call c, string funcName, int argInd |
+      funcName = c.getTarget().getName() and this = c.getArgument(argInd)
+    |
+      asymmetricCipherAlgorithmSink(funcName, argInd)
+    )
+  }
+
+  override string algType() { result = "ASYMMETRIC_ENCRYPTION" }
+}
+
+class AsymmetricCipherAlgorithmSink_EVP_PKEY_Q_keygen extends AlgorithmSinkArgument {
+  AsymmetricCipherAlgorithmSink_EVP_PKEY_Q_keygen() {
+    exists(Call c, string funcName |
+      funcName = c.getTarget().getName() and
+      this = c.getArgument(3)
+    |
+      funcName = "EVP_PKEY_Q_keygen" and
+      c.getArgument(3).getType().getUnderlyingType() instanceof IntegralType
+    )
+  }
+
+  override string algType() { result = "ASYMMETRIC_ENCRYPTION" }
+}
+
+// https://www.openssl.org/docs/manmaster/man3/EVP_RAND_fetch
+predicate randomAlgorithmSink(string funcName, int argInd) {
+  funcName = "EVP_RAND_fetch" and argInd = 1
+}
+
+class RandomAlgorithmSink extends AlgorithmSinkArgument {
+  RandomAlgorithmSink() {
+    exists(Call c, string funcName, int argInd |
+      funcName = c.getTarget().getName() and this = c.getArgument(argInd)
+    |
+      randomAlgorithmSink(funcName, argInd)
+    )
+  }
+
+  override string algType() { result = "TBD" }
+}
+
+// https://www.openssl.org/docs/manmaster/man3/EVP_SIGNATURE_fetch
+predicate signatureAlgorithmSink(string funcName, int argInd) {
+  funcName = "EVP_SIGNATURE_fetch" and argInd = 1
+}
+
+class SignatureAlgorithmSink extends AlgorithmSinkArgument {
+  SignatureAlgorithmSink() {
+    exists(Call c, string funcName, int argInd |
+      funcName = c.getTarget().getName() and this = c.getArgument(argInd)
+    |
+      signatureAlgorithmSink(funcName, argInd)
+    )
+  }
+
+  override string algType() { result = getSignatureType() }
+}
+
+// https://www.openssl.org/docs/manmaster/man3/EC_KEY_new_by_curve_name.html
+// https://www.openssl.org/docs/manmaster/man3/EVP_PKEY_CTX_set_ec_paramgen_curve_nid.html
+predicate ellipticCurveAlgorithmSink(string funcName, int argInd) {
+  funcName in ["EC_KEY_new_by_curve_name", "EVP_EC_gen"] and argInd = 0
+  or
+  funcName = "EC_KEY_new_by_curve_name_ex" and argInd = 2
+  or
+  funcName in ["EVP_PKEY_CTX_set_ec_paramgen_curve_nid"] and argInd = 1
+}
+
+class EllipticCurveAlgorithmSink extends AlgorithmSinkArgument {
+  EllipticCurveAlgorithmSink() {
+    exists(Call c, string funcName, int argInd |
+      funcName = c.getTarget().getName() and this = c.getArgument(argInd)
+    |
+      ellipticCurveAlgorithmSink(funcName, argInd)
+    )
+  }
+
+  override string algType() { result = getEllipticCurveType() }
+}
+
+/**
+ * Special cased to address the fact that arg index 3 (zero offset based) is the curve name.
+ * ASSUMPTION: if the arg ind 3 is a char* assume it is an elliptic curve
+ */
+class EllipticCurveAlgorithmSink_EVP_PKEY_Q_keygen extends AlgorithmSinkArgument {
+  EllipticCurveAlgorithmSink_EVP_PKEY_Q_keygen() {
+    exists(Call c, string funcName |
+      funcName = c.getTarget().getName() and
+      this = c.getArgument(3)
+    |
+      funcName = "EVP_PKEY_Q_keygen" and
+      c.getArgument(3).getType().getUnderlyingType() instanceof PointerType and
+      c.getArgument(3).getType().getUnderlyingType().stripType() instanceof CharType
+    )
+  }
+
+  override string algType() { result = getEllipticCurveType() }
+}
+
+// https://www.openssl.org/docs/manmaster/man3/EVP_PKEY_CTX_new_id.html
+// https://www.openssl.org/docs/man1.1.1/man3/EVP_PKEY_new_raw_private_key.html
+// https://www.openssl.org/docs/manmaster/man3/EVP_PKEY_new.html
+// https://www.openssl.org/docs/manmaster/man3/EVP_PKEY_CTX_ctrl.html
+// https://www.openssl.org/docs/manmaster/man3/EVP_PKEY_Q_keygen.html
+// https://www.openssl.org/docs/manmaster/man3/EVP_PKEY_CTX_ctrl.html
+predicate asymmetricAlgorithmSink(string funcName, int argInd) {
+  funcName = "EVP_PKEY_CTX_new_id" and argInd = 0
+  or
+  funcName = "EVP_PKEY_CTX_new_from_name" and argInd = 1
+  or
+  funcName in [
+      "EVP_PKEY_new_raw_private_key", "EVP_PKEY_new_raw_public_key", "EVP_PKEY_new_mac_key"
+    ] and
+  argInd = 0
+  or
+  funcName in ["EVP_PKEY_new_raw_private_key_ex", "EVP_PKEY_new_raw_public_key_ex"] and argInd = 1
+  or
+  // special casing this as arg index 3 must be specified depending on if RSA or ECC, and otherwise not specified for other algs
+  // funcName = "EVP_PKEY_Q_keygen" and argInd = 2
+  funcName in ["EVP_PKEY_CTX_ctrl", "EVP_PKEY_CTX_set_group_name"] and argInd = 1
+  // TODO consider void cases EVP_PKEY_new
+}
+
+class AsymmetricAlgorithmSink extends AlgorithmSinkArgument {
+  AsymmetricAlgorithmSink() {
+    exists(Call c, string funcName, int argInd |
+      funcName = c.getTarget().getName() and this = c.getArgument(argInd)
+    |
+      asymmetricAlgorithmSink(funcName, argInd)
+    )
+  }
+
+  override string algType() { result = getAsymmetricType() }
+}
+
+class AsymmetricAlgorithmSink_EVP_PKEY_Q_keygen extends AlgorithmSinkArgument {
+  AsymmetricAlgorithmSink_EVP_PKEY_Q_keygen() {
+    exists(Call c, string funcName |
+      funcName = c.getTarget().getName() and
+      this = c.getArgument(2)
+    |
+      funcName = "EVP_PKEY_Q_keygen" and
+      not exists(c.getArgument(3))
+    )
+  }
+
+  override string algType() { result = getAsymmetricType() }
+}

cpp/ql/lib/experimental/cryptography/utils/OpenSSL/CryptoFunction.qll

@@ -0,0 +1,121 @@
+import cpp
+import experimental.cryptography.utils.OpenSSL.LibraryFunction
+import experimental.cryptography.CryptoAlgorithmNames
+
+predicate inferredOpenSSLCryptoFunctionCall(Call c, string normalized, string algType) {
+  inferredOpenSSLCryptoFunction(c.getTarget(), normalized, algType)
+}
+
+predicate inferredOpenSSLCryptoFunction(Function f, string normalized, string algType) {
+  isPossibleOpenSSLFunction(f) and
+  normalizeFunctionName(f, algType) = normalized
+}
+
+predicate isOpenSSLCryptoFunction(Function f, string normalized, string algType) {
+  // NOTE: relying on inference as there are thousands of functions for crypto
+  //       enumerating them all and maintaining the list seems problematic.
+  //       For now, we will rely on dynamically inferring algorithms for function names.
+  //       This has been seen to be reasonably efficient and accurate.
+  inferredOpenSSLCryptoFunction(f, normalized, algType)
+}
+
+predicate isOpenSSLCryptoFunctionCall(Call c, string normalized, string algType) {
+  isOpenSSLCryptoFunction(c.getTarget(), normalized, algType)
+}
+
+private string basicNormalizeFunctionName(Function f, string algType) {
+  isPossibleOpenSSLFunction(f) and
+  isKnownAlgorithm(result, algType) and
+  exists(string normStr | normStr = f.getName().toUpperCase().regexpReplaceAll("[-_ ]|/", "") |
+    normStr.matches("%" + result + "%")
+  )
+}
+
+/**
+ * Converts a raw OpenSSL algorithm to a normalized algorithm name.
+ *
+ * If more than one match occurs for a given algorithm type, normalize attempts to find the "max"
+ * string (max in terms of string length) e.g., matching AES128 to AES128 and not simply AES.
+ *
+ * An unknown algorithm is only identified if there exists no known algorithm found for any algorithm type.
+ *
+ * `f` is the function name to normalize.
+ * `algType` is a string representing the classification of the algorithm (see `CryptoAlgorithmNames`)
+ */
+private string privateNormalizeFunctionName(Function f, string algType) {
+  isPossibleOpenSSLFunction(f) and
+  result = basicNormalizeFunctionName(f, algType) and
+  not exists(string res2 |
+    result != res2 and
+    res2 = basicNormalizeFunctionName(f, algType) and
+    res2.length() > result.length()
+  )
+}
+
+/**
+ * Normalizes a function name to a known algorithm name, similar to `normalizeName`.
+ * A function is not, however, allowed to be UNKNOWN. The function either
+ * normalizes to a known algorithm name, or the predicate does not hold (no result).
+ *
+ * The predicate attempts to restrict normalization to what looks like an openssl
+ * library by looking for functions only in an openssl path (see `isPossibleOpenSSLFunction`).
+ * This may give false postive functions if a directory erronously appears to be openssl;
+ * however, we take the stance that if a function
+ * exists strongly mapping to a known function name in a directory such as these,
+ * regardless of whether its actually a part of openSSL or not, we will analyze it as though it were.
+ */
+private string normalizeFunctionName(Function f, string algType) {
+  algType != "UNKNOWN" and
+  isPossibleOpenSSLFunction(f) and
+  result = privateNormalizeFunctionName(f, algType) and
+  // Addressing false positives
+  // For algorithm names less than or equal to 4, we must see the algorithm name
+  // in the original function as upper case (it can't be split between tokens)
+  // One exception found is DES_xcbc_encrypt, this is DESX
+  (
+    (result.length() <= 4 and result != "DESX")
+    implies
+    f.getName().toUpperCase().matches("%" + result + "%")
+  ) and
+  (
+    (result.length() <= 4 and result = "DESX")
+    implies
+    (f.getName().toUpperCase().matches("%DESX%") or f.getName().toUpperCase().matches("%DES_X%"))
+  ) and
+  // (result.length() <= 3 implies (not f.getName().toUpperCase().regexpMatch(".*" + result + "[a-zA-Z0-9].*|.*[a-zA-Z0-9]" + result + ".*")))
+  // and
+  // DES specific false positives
+  (
+    result.matches("DES")
+    implies
+    not f.getName().toUpperCase().regexpMatch(".*DES[a-zA-Z0-9].*|.*[a-zA-Z0-9]DES.*")
+  ) and
+  // ((result.matches("%DES%")) implies not exists(string s | s in ["DESCRIBE", "DESTROY", "DESCRIPTION", "DESCRIPTOR", "NODES"] |
+  //     f.getName().toUpperCase().matches("%" + s + "%"))) and
+  // SEED specific false positives
+  (
+    result.matches("SEED")
+    implies
+    not exists(string s |
+      s in [
+          "SEED_SRC_GENERATE", "RAND", "NEW_SEED", "GEN_SEED", "SEED_GEN", "SET_SEED", "GET_SEED",
+          "GET0_SEED", "RESEED", "SEEDING"
+        ]
+    |
+      f.getName().toUpperCase().matches("%" + s + "%")
+    )
+  ) and
+  // ARIA specific false positives
+  (result.matches("ARIA") implies not f.getName().toUpperCase().matches("%VARIANT%")) and
+  // CTR false positives
+  (result.matches("CTR") implies not f.getName().toUpperCase().matches("%CTRL%")) and
+  // ES false positives (e.g., ES256 from AES256)
+  (result.matches("ES%") implies not f.getName().toUpperCase().matches("%AES%")) and
+  // RSA false positives
+  (result.matches("RSA") implies not f.getName().toUpperCase().matches("%UNIVERSAL%")) and
+  //rsaz functions deemed to be too low level, and can be ignored
+  not f.getLocation().getFile().getBaseName().matches("rsaz_exp.c") and
+  // General False positives
+  // Functions that 'get' do not set an algorithm, and therefore are considered ignorable
+  not f.getName().toLowerCase().matches("%get%")
+}

cpp/ql/lib/experimental/cryptography/utils/OpenSSL/DataBuilders.qll

@@ -0,0 +1,153 @@
+/**
+ * This file contains predicates create to build up initial data sets for OpenSSL
+ * predicates. E.g., These predicates were used to assist in associating all
+ * openSSL functions with their known crypto algorithms.
+ */
+
+import cpp
+import experimental.cryptography.CryptoAlgorithmNames
+import experimental.cryptography.utils.OpenSSL.CryptoFunction
+
+private string basicNormalizeFunctionName(Function f, string algType) {
+  isKnownAlgorithm(result, algType) and
+  exists(string normStr | normStr = f.getName().toUpperCase().regexpReplaceAll("[-_ ]|/", "") |
+    normStr.matches("%" + result + "%")
+  )
+}
+
+/**
+ * Converts a raw OpenSSL algorithm to a normalized algorithm name.
+ *
+ * If more than one match occurs for a given algorithm type, normalize attempts to find the "max"
+ * string (max in terms of string length) e.g., matching AES128 to AES128 and not simply AES.
+ *
+ * An unknown algorithm is only identified if there exists no known algorithm found for any algorithm type.
+ *
+ * `f` is the function name to normalize.
+ * `algType` is a string representing the classification of the algorithm (see `CryptoAlgorithmNames`)
+ */
+private string privateNormalizeFunctionName(Function f, string algType) {
+  result = basicNormalizeFunctionName(f, algType) and
+  not exists(string res2 |
+    result != res2 and
+    res2 = basicNormalizeFunctionName(f, algType) and
+    res2.length() > result.length()
+  ) and
+  // Addressing bad normalization case-by-case
+  // CASE: ES256 being identified when the algorithm is AES256
+  (
+    result.matches("ES256")
+    implies
+    not exists(string res2 | res2 = basicNormalizeFunctionName(f, _) and res2.matches("AES%"))
+  )
+}
+
+/**
+ * Normalizes a function name to a known algorithm name, similar to `normalizeName`.
+ * A function is not, however, allowed to be UNKNOWN. The function either
+ * normalizes to a known algorithm name, or the predicate does not hold (no result).
+ *
+ * The predicate attempts to restrict normalization to what looks like an openssl
+ * library by looking for functions only in an openssl path (see `isPossibleOpenSSLFunction`).
+ * This may give false postive functions if a directory erronously appears to be openssl;
+ * however, we take the stance that if a function
+ * exists strongly mapping to a known function name in a directory such as these,
+ * regardless of whether its actually a part of openSSL or not, we will analyze it as though it were.
+ */
+string normalizeFunctionName(Function f, string algType) {
+  algType != "UNKNOWN" and
+  result = privateNormalizeFunctionName(f, algType) and
+  openSSLLibraryFunc(f) and
+  // Addressing false positives
+  // For algorithm names less than or equal to 4, we must see the algorithm name
+  // in the original function as upper case (it can't be split between tokens)
+  // One exception found is DES_xcbc_encrypt, this is DESX
+  (
+    (result.length() <= 4 and result != "DESX")
+    implies
+    f.getName().toUpperCase().matches("%" + result + "%")
+  ) and
+  (
+    (result.length() <= 4 and result = "DESX")
+    implies
+    (f.getName().toUpperCase().matches("%DESX%") or f.getName().toUpperCase().matches("%DES_X%"))
+  ) and
+  // (result.length() <= 3 implies (not f.getName().toUpperCase().regexpMatch(".*" + result + "[a-zA-Z0-9].*|.*[a-zA-Z0-9]" + result + ".*")))
+  // and
+  // DES specific false positives
+  (
+    result.matches("DES")
+    implies
+    not f.getName().toUpperCase().regexpMatch(".*DES[a-zA-Z0-9].*|.*[a-zA-Z0-9]DES.*")
+  ) and
+  // ((result.matches("%DES%")) implies not exists(string s | s in ["DESCRIBE", "DESTROY", "DESCRIPTION", "DESCRIPTOR", "NODES"] |
+  //     f.getName().toUpperCase().matches("%" + s + "%"))) and
+  // SEED specific false positives
+  (
+    result.matches("%SEED%")
+    implies
+    not not exists(string s |
+      s in ["NEW_SEED", "GEN_SEED", "SET_SEED", "GET_SEED", "GET0_SEED", "RESEED", "SEEDING"]
+    |
+      f.getName().toUpperCase().matches("%" + s + "%")
+    )
+  ) and
+  // ARIA specific false positives
+  (result.matches("%ARIA%") implies not f.getName().toUpperCase().matches("%VARIANT%"))
+}
+
+/**
+ * Predicate to support name normalization.
+ * Converts the raw name upper-case with no hyphen, slash, underscore, hash, or space.
+ * Looks for substrings that are known algorithms, and normalizes the name.
+ * If the algorithm cannot be determined or is in the ignorable list (`isIgnorableOpenSSLAlgorithm`)
+ * this predicate will not resolve a name.
+ *
+ * Rationale for private: For normalization, we want to get the longest string for a normalized name match
+ *       for a given algorithm type. I found this easier to express if the public normalizeName
+ *       checks that the name is the longest, and that UNKNOWN is reserved if there exists no
+ *       result from this predicate that is known.
+ */
+bindingset[name]
+string privateNormalizeName(string name, string algType) {
+  //not isIgnorableOpenSSLAlgorithm(name, _, _) and
+  // targetOpenSSLAlgorithm(name, _) and
+  isKnownAlgorithm(result, algType) and
+  exists(string normStr | normStr = name.toUpperCase().regexpReplaceAll("[-_ ]|/", "") |
+    normStr.matches("%" + result + "%")
+  )
+}
+
+/**
+ * Converts a raw OpenSSL algorithm to a normalized algorithm name.
+ *
+ * If more than one match occurs for a given algorithm type, normalize attempts to find the "max"
+ * string (max in terms of string length) e.g., matching AES128 to AES128 and not simply AES.
+ *
+ * An unknown algorithm is only identified if there exists no known algorithm found for any algorithm type.
+ *
+ * `name` is the name to normalize.
+ * `algType` is a string representing the classification of the algorithm (see `CryptoAlgorithmNames`)
+ */
+bindingset[name]
+string normalizeName(string name, string algType) {
+  (
+    if exists(privateNormalizeName(name, _))
+    then result = privateNormalizeName(name, algType)
+    else (
+      result = unknownAlgorithm() and algType = "UNKNOWN"
+    )
+  ) and
+  not exists(string res2 |
+    result != res2 and
+    res2 = privateNormalizeName(name, algType) and
+    res2.length() > result.length()
+  ) and
+  // Addressing bad normalization case-by-case
+  // CASE: ES256 being identified when the algorithm is AES256
+  (
+    result.matches("ES256")
+    implies
+    not exists(string res2 | res2 = privateNormalizeName(name, _) and res2.matches("AES%"))
+  )
+}

cpp/ql/lib/experimental/cryptography/utils/OpenSSL/PassthroughFunction.qll

@@ -0,0 +1,59 @@
+import cpp
+import experimental.cryptography.utils.OpenSSL.LibraryFunction
+import semmle.code.cpp.ir.dataflow.DataFlow
+
+// TODO: possible use of extensible predicates here
+// NOTE: -1 for outInd represents the return value
+predicate knownPassthroughFunction(Function f, int inInd, int outInd) {
+  // Trace through functions
+  // See https://www.openssl.org/docs/man1.1.1/man3/OBJ_obj2txt
+  //     https://www.openssl.org/docs/man3.0/man3/EVP_CIPHER_get0_name
+  openSSLLibraryFunc(f) and
+  (
+    f.getName() in [
+        "OBJ_nid2obj", "OBJ_nid2ln", "OBJ_nid2sn", "OBJ_obj2nid", "OBJ_ln2nid", "OBJ_sn2nid",
+        "OBJ_txt2nid", "OBJ_txt2obj", "OBJ_dup", "EVP_CIPHER_get0_name"
+      ] and
+    inInd = 0 and
+    outInd = -1
+    or
+    f.getName() in ["OBJ_obj2txt", "i2t_ASN1_OBJECT"] and
+    inInd = 2 and
+    outInd = 0
+    or
+    // Dup/copy pattern occurs in more places,
+    //see: https://www.openssl.org/docs/manmaster/man3/EC_KEY_copy.html and https://www.openssl.org/docs/manmaster/man3/EVP_PKEY_CTX_dup.html
+    f.getName().matches("%_dup") and inInd = 0 and outInd = -1
+    or
+    f.getName().matches("%_copy") and inInd = 0 and outInd = -1
+  )
+}
+
+/**
+ * `c` is a call to a function that preserves the algorithm but changes its form.
+ * `onExpr` is the input argument passing through to, `outExpr` is the next expression in a dataflow step associated with `c`
+ */
+predicate knownPassthoughCall(Call c, Expr inExpr, Expr outExpr) {
+  exists(int inInd, int outInd |
+    knownPassthroughFunction(c.getTarget(), inInd, outInd) and
+    inExpr = c.getArgument(inInd) and
+    if outInd = -1 then outExpr = c else outExpr = c.getArgument(outInd)
+  )
+}
+
+/*
+ * Explicitly add flow through openssl functions that preserve the algorithm but alter the form (e.g., from NID to string)
+ */
+
+predicate knownPassThroughStep(DataFlow::Node node1, DataFlow::Node node2) {
+  exists(Expr cur, Expr next |
+    (cur = node1.asExpr() or cur = node1.asIndirectArgument()) and
+    (
+      next = node2.asExpr() or
+      next = node2.asIndirectArgument() or
+      next = node2.asDefiningArgument()
+    )
+  |
+    exists(Call c | knownPassthoughCall(c, cur, next))
+  )
+}

cpp/ql/lib/qlpack.yml

@@ -1,5 +1,5 @@
 name: codeql/cpp-all
-version: 0.9.2-dev
+version: 0.10.2-dev
 groups: cpp
 dbscheme: semmlecode.cpp.dbscheme
 extractor: cpp

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

@@ -158,9 +158,7 @@ class NameQualifyingElement extends Element, @namequalifyingelement {
 /**
  * A special name-qualifying element. For example: `__super`.
  */
-library class SpecialNameQualifyingElement extends NameQualifyingElement,
-  @specialnamequalifyingelement
-{
+class SpecialNameQualifyingElement extends NameQualifyingElement, @specialnamequalifyingelement {
   /** Gets the name of this special qualifying element. */
   override string getName() { specialnamequalifyingelements(underlyingElement(this), result) }
 

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

@@ -819,6 +819,30 @@ private predicate floatingPointTypeMapping(
   or
   // _Complex _Float16
   kind = 53 and base = 2 and domain = TComplexDomain() and realKind = 52 and extended = false
+  or
+  // __fp16
+  kind = 54 and base = 2 and domain = TRealDomain() and realKind = 54 and extended = false
+  or
+  // __bf16
+  kind = 55 and base = 2 and domain = TRealDomain() and realKind = 55 and extended = false
+  or
+  // std::float16_t
+  kind = 56 and base = 2 and domain = TRealDomain() and realKind = 56 and extended = false
+  or
+  // _Complex _Float32
+  kind = 57 and base = 2 and domain = TComplexDomain() and realKind = 45 and extended = false
+  or
+  // _Complex _Float32x
+  kind = 58 and base = 2 and domain = TComplexDomain() and realKind = 46 and extended = true
+  or
+  // _Complex _Float64
+  kind = 59 and base = 2 and domain = TComplexDomain() and realKind = 47 and extended = false
+  or
+  // _Complex _Float64x
+  kind = 60 and base = 2 and domain = TComplexDomain() and realKind = 48 and extended = true
+  or
+  // _Complex _Float128
+  kind = 61 and base = 2 and domain = TComplexDomain() and realKind = 49 and extended = false
 }
 
 /**

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

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

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

@@ -27,9 +27,6 @@ predicate canValueFlow(Expr fromExpr, Expr toExpr) {
   fromExpr = toExpr.(ConditionalExpr).getElse()
 }
 
-/** DEPRECATED: Alias for AnalyzedString */
-deprecated class AnalysedString = AnalyzedString;
-
 /**
  * An analyzed null terminated string.
  */

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

@@ -78,7 +78,7 @@ predicate parameterUsePair(Parameter p, VariableAccess va) {
 /**
  * Utility class: A definition or use of a stack variable.
  */
-library class DefOrUse extends ControlFlowNodeBase {
+class DefOrUse extends ControlFlowNodeBase {
   DefOrUse() {
     // Uninstantiated templates are purely syntax, and only on instantiation
     // will they be complete with information about types, conversions, call
@@ -140,7 +140,7 @@ library class DefOrUse extends ControlFlowNodeBase {
 }
 
 /** A definition of a stack variable. */
-library class Def extends DefOrUse {
+class Def extends DefOrUse {
   Def() { definition(_, this) }
 
   override SemanticStackVariable getVariable(boolean isDef) {
@@ -155,7 +155,7 @@ private predicate parameterIsOverwritten(Function f, Parameter p) {
 }
 
 /** A definition of a parameter. */
-library class ParameterDef extends DefOrUse {
+class ParameterDef extends DefOrUse {
   ParameterDef() {
     // Optimization: parameters that are not overwritten do not require
     // reachability analysis
@@ -169,7 +169,7 @@ library class ParameterDef extends DefOrUse {
 }
 
 /** A use of a stack variable. */
-library class Use extends DefOrUse {
+class Use extends DefOrUse {
   Use() { useOfVar(_, this) }
 
   override SemanticStackVariable getVariable(boolean isDef) {

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

@@ -10,7 +10,7 @@ import SSAUtils
  * The SSA logic comes in two versions: the standard SSA and range-analysis RangeSSA.
  * This class provides the standard SSA logic.
  */
-library class StandardSsa extends SsaHelper {
+class StandardSsa extends SsaHelper {
   StandardSsa() { this = 0 }
 }
 

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

@@ -114,7 +114,7 @@ private predicate live_at_exit_of_bb(StackVariable v, BasicBlock b) {
 
 /** Common SSA logic for standard SSA and range-analysis SSA. */
 cached
-library class SsaHelper extends int {
+class SsaHelper extends int {
   /* 0 = StandardSSA, 1 = RangeSSA */
   cached
   SsaHelper() { this in [0 .. 1] }

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

@@ -366,12 +366,12 @@ class CompileTimeConstantInt extends Expr {
   int getIntValue() { result = val }
 }
 
-library class CompileTimeVariableExpr extends Expr {
+class CompileTimeVariableExpr extends Expr {
   CompileTimeVariableExpr() { not this instanceof CompileTimeConstantInt }
 }
 
 /** A helper class for evaluation of expressions. */
-library class ExprEvaluator extends int {
+class ExprEvaluator extends int {
   /*
    * 0 = ConditionEvaluator,
    * 1 = SwitchEvaluator,
@@ -956,7 +956,7 @@ private predicate returnStmt(Function f, Expr value) {
 }
 
 /** A helper class for evaluation of conditions. */
-library class ConditionEvaluator extends ExprEvaluator {
+class ConditionEvaluator extends ExprEvaluator {
   ConditionEvaluator() { this = 0 }
 
   override predicate interesting(Expr e) {
@@ -967,7 +967,7 @@ library class ConditionEvaluator extends ExprEvaluator {
 }
 
 /** A helper class for evaluation of switch expressions. */
-library class SwitchEvaluator extends ExprEvaluator {
+class SwitchEvaluator extends ExprEvaluator {
   SwitchEvaluator() { this = 1 }
 
   override predicate interesting(Expr e) { e = getASwitchExpr(_, _) }
@@ -976,7 +976,7 @@ library class SwitchEvaluator extends ExprEvaluator {
 private int getSwitchValue(Expr e) { exists(SwitchEvaluator x | result = x.getValue(e)) }
 
 /** A helper class for evaluation of loop entry conditions. */
-library class LoopEntryConditionEvaluator extends ExprEvaluator {
+class LoopEntryConditionEvaluator extends ExprEvaluator {
   LoopEntryConditionEvaluator() { this in [2 .. 3] }
 
   abstract override predicate interesting(Expr e);
@@ -1149,7 +1149,7 @@ library class LoopEntryConditionEvaluator extends ExprEvaluator {
 }
 
 /** A helper class for evaluation of while-loop entry conditions. */
-library class WhileLoopEntryConditionEvaluator extends LoopEntryConditionEvaluator {
+class WhileLoopEntryConditionEvaluator extends LoopEntryConditionEvaluator {
   WhileLoopEntryConditionEvaluator() { this = 2 }
 
   override predicate interesting(Expr e) { exists(WhileStmt while | e = while.getCondition()) }
@@ -1162,7 +1162,7 @@ library class WhileLoopEntryConditionEvaluator extends LoopEntryConditionEvaluat
 }
 
 /** A helper class for evaluation of for-loop entry conditions. */
-library class ForLoopEntryConditionEvaluator extends LoopEntryConditionEvaluator {
+class ForLoopEntryConditionEvaluator extends LoopEntryConditionEvaluator {
   ForLoopEntryConditionEvaluator() { this = 3 }
 
   override predicate interesting(Expr e) { exists(ForStmt for | e = for.getCondition()) }

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

@@ -91,21 +91,6 @@ abstract class Configuration extends string {
   /** Holds if data flow out of `node` is prohibited. */
   predicate isBarrierOut(Node node) { none() }
 
-  /**
-   * DEPRECATED: Use `isBarrier` and `BarrierGuard` module instead.
-   *
-   * Holds if data flow through nodes guarded by `guard` is prohibited.
-   */
-  deprecated predicate isBarrierGuard(BarrierGuard guard) { none() }
-
-  /**
-   * DEPRECATED: Use `isBarrier` and `BarrierGuard` module instead.
-   *
-   * Holds if data flow through nodes guarded by `guard` is prohibited when
-   * the flow state is `state`
-   */
-  deprecated predicate isBarrierGuard(BarrierGuard guard, FlowState state) { none() }
-
   /**
    * Holds if data may flow from `node1` to `node2` in addition to the normal data-flow steps.
    */
@@ -225,29 +210,6 @@ abstract private class ConfigurationRecursionPrevention extends Configuration {
   }
 }
 
-/** A bridge class to access the deprecated `isBarrierGuard`. */
-private class BarrierGuardGuardedNodeBridge extends Unit {
-  abstract predicate guardedNode(Node n, Configuration config);
-
-  abstract predicate guardedNode(Node n, FlowState state, Configuration config);
-}
-
-private class BarrierGuardGuardedNode extends BarrierGuardGuardedNodeBridge {
-  deprecated override predicate guardedNode(Node n, Configuration config) {
-    exists(BarrierGuard g |
-      config.isBarrierGuard(g) and
-      n = g.getAGuardedNode()
-    )
-  }
-
-  deprecated override predicate guardedNode(Node n, FlowState state, Configuration config) {
-    exists(BarrierGuard g |
-      config.isBarrierGuard(g, state) and
-      n = g.getAGuardedNode()
-    )
-  }
-}
-
 private FlowState relevantState(Configuration config) {
   config.isSource(_, result) or
   config.isSink(_, result) or
@@ -288,9 +250,7 @@ private module Config implements FullStateConfigSig {
 
   predicate isBarrier(Node node, FlowState state) {
     getConfig(state).isBarrier(node, getState(state)) or
-    getConfig(state).isBarrier(node) or
-    any(BarrierGuardGuardedNodeBridge b).guardedNode(node, getState(state), getConfig(state)) or
-    any(BarrierGuardGuardedNodeBridge b).guardedNode(node, getConfig(state))
+    getConfig(state).isBarrier(node)
   }
 
   predicate isBarrierIn(Node node) { any(Configuration config).isBarrierIn(node) }

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

@@ -91,21 +91,6 @@ abstract class Configuration extends string {
   /** Holds if data flow out of `node` is prohibited. */
   predicate isBarrierOut(Node node) { none() }
 
-  /**
-   * DEPRECATED: Use `isBarrier` and `BarrierGuard` module instead.
-   *
-   * Holds if data flow through nodes guarded by `guard` is prohibited.
-   */
-  deprecated predicate isBarrierGuard(BarrierGuard guard) { none() }
-
-  /**
-   * DEPRECATED: Use `isBarrier` and `BarrierGuard` module instead.
-   *
-   * Holds if data flow through nodes guarded by `guard` is prohibited when
-   * the flow state is `state`
-   */
-  deprecated predicate isBarrierGuard(BarrierGuard guard, FlowState state) { none() }
-
   /**
    * Holds if data may flow from `node1` to `node2` in addition to the normal data-flow steps.
    */
@@ -225,29 +210,6 @@ abstract private class ConfigurationRecursionPrevention extends Configuration {
   }
 }
 
-/** A bridge class to access the deprecated `isBarrierGuard`. */
-private class BarrierGuardGuardedNodeBridge extends Unit {
-  abstract predicate guardedNode(Node n, Configuration config);
-
-  abstract predicate guardedNode(Node n, FlowState state, Configuration config);
-}
-
-private class BarrierGuardGuardedNode extends BarrierGuardGuardedNodeBridge {
-  deprecated override predicate guardedNode(Node n, Configuration config) {
-    exists(BarrierGuard g |
-      config.isBarrierGuard(g) and
-      n = g.getAGuardedNode()
-    )
-  }
-
-  deprecated override predicate guardedNode(Node n, FlowState state, Configuration config) {
-    exists(BarrierGuard g |
-      config.isBarrierGuard(g, state) and
-      n = g.getAGuardedNode()
-    )
-  }
-}
-
 private FlowState relevantState(Configuration config) {
   config.isSource(_, result) or
   config.isSink(_, result) or
@@ -288,9 +250,7 @@ private module Config implements FullStateConfigSig {
 
   predicate isBarrier(Node node, FlowState state) {
     getConfig(state).isBarrier(node, getState(state)) or
-    getConfig(state).isBarrier(node) or
-    any(BarrierGuardGuardedNodeBridge b).guardedNode(node, getState(state), getConfig(state)) or
-    any(BarrierGuardGuardedNodeBridge b).guardedNode(node, getConfig(state))
+    getConfig(state).isBarrier(node)
   }
 
   predicate isBarrierIn(Node node) { any(Configuration config).isBarrierIn(node) }

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

@@ -91,21 +91,6 @@ abstract class Configuration extends string {
   /** Holds if data flow out of `node` is prohibited. */
   predicate isBarrierOut(Node node) { none() }
 
-  /**
-   * DEPRECATED: Use `isBarrier` and `BarrierGuard` module instead.
-   *
-   * Holds if data flow through nodes guarded by `guard` is prohibited.
-   */
-  deprecated predicate isBarrierGuard(BarrierGuard guard) { none() }
-
-  /**
-   * DEPRECATED: Use `isBarrier` and `BarrierGuard` module instead.
-   *
-   * Holds if data flow through nodes guarded by `guard` is prohibited when
-   * the flow state is `state`
-   */
-  deprecated predicate isBarrierGuard(BarrierGuard guard, FlowState state) { none() }
-
   /**
    * Holds if data may flow from `node1` to `node2` in addition to the normal data-flow steps.
    */
@@ -225,29 +210,6 @@ abstract private class ConfigurationRecursionPrevention extends Configuration {
   }
 }
 
-/** A bridge class to access the deprecated `isBarrierGuard`. */
-private class BarrierGuardGuardedNodeBridge extends Unit {
-  abstract predicate guardedNode(Node n, Configuration config);
-
-  abstract predicate guardedNode(Node n, FlowState state, Configuration config);
-}
-
-private class BarrierGuardGuardedNode extends BarrierGuardGuardedNodeBridge {
-  deprecated override predicate guardedNode(Node n, Configuration config) {
-    exists(BarrierGuard g |
-      config.isBarrierGuard(g) and
-      n = g.getAGuardedNode()
-    )
-  }
-
-  deprecated override predicate guardedNode(Node n, FlowState state, Configuration config) {
-    exists(BarrierGuard g |
-      config.isBarrierGuard(g, state) and
-      n = g.getAGuardedNode()
-    )
-  }
-}
-
 private FlowState relevantState(Configuration config) {
   config.isSource(_, result) or
   config.isSink(_, result) or
@@ -288,9 +250,7 @@ private module Config implements FullStateConfigSig {
 
   predicate isBarrier(Node node, FlowState state) {
     getConfig(state).isBarrier(node, getState(state)) or
-    getConfig(state).isBarrier(node) or
-    any(BarrierGuardGuardedNodeBridge b).guardedNode(node, getState(state), getConfig(state)) or
-    any(BarrierGuardGuardedNodeBridge b).guardedNode(node, getConfig(state))
+    getConfig(state).isBarrier(node)
   }
 
   predicate isBarrierIn(Node node) { any(Configuration config).isBarrierIn(node) }

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

@@ -91,21 +91,6 @@ abstract class Configuration extends string {
   /** Holds if data flow out of `node` is prohibited. */
   predicate isBarrierOut(Node node) { none() }
 
-  /**
-   * DEPRECATED: Use `isBarrier` and `BarrierGuard` module instead.
-   *
-   * Holds if data flow through nodes guarded by `guard` is prohibited.
-   */
-  deprecated predicate isBarrierGuard(BarrierGuard guard) { none() }
-
-  /**
-   * DEPRECATED: Use `isBarrier` and `BarrierGuard` module instead.
-   *
-   * Holds if data flow through nodes guarded by `guard` is prohibited when
-   * the flow state is `state`
-   */
-  deprecated predicate isBarrierGuard(BarrierGuard guard, FlowState state) { none() }
-
   /**
    * Holds if data may flow from `node1` to `node2` in addition to the normal data-flow steps.
    */
@@ -225,29 +210,6 @@ abstract private class ConfigurationRecursionPrevention extends Configuration {
   }
 }
 
-/** A bridge class to access the deprecated `isBarrierGuard`. */
-private class BarrierGuardGuardedNodeBridge extends Unit {
-  abstract predicate guardedNode(Node n, Configuration config);
-
-  abstract predicate guardedNode(Node n, FlowState state, Configuration config);
-}
-
-private class BarrierGuardGuardedNode extends BarrierGuardGuardedNodeBridge {
-  deprecated override predicate guardedNode(Node n, Configuration config) {
-    exists(BarrierGuard g |
-      config.isBarrierGuard(g) and
-      n = g.getAGuardedNode()
-    )
-  }
-
-  deprecated override predicate guardedNode(Node n, FlowState state, Configuration config) {
-    exists(BarrierGuard g |
-      config.isBarrierGuard(g, state) and
-      n = g.getAGuardedNode()
-    )
-  }
-}
-
 private FlowState relevantState(Configuration config) {
   config.isSource(_, result) or
   config.isSink(_, result) or
@@ -288,9 +250,7 @@ private module Config implements FullStateConfigSig {
 
   predicate isBarrier(Node node, FlowState state) {
     getConfig(state).isBarrier(node, getState(state)) or
-    getConfig(state).isBarrier(node) or
-    any(BarrierGuardGuardedNodeBridge b).guardedNode(node, getState(state), getConfig(state)) or
-    any(BarrierGuardGuardedNodeBridge b).guardedNode(node, getConfig(state))
+    getConfig(state).isBarrier(node)
   }
 
   predicate isBarrierIn(Node node) { any(Configuration config).isBarrierIn(node) }

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

@@ -91,21 +91,6 @@ abstract class Configuration extends string {
   /** Holds if data flow out of `node` is prohibited. */
   predicate isBarrierOut(Node node) { none() }
 
-  /**
-   * DEPRECATED: Use `isBarrier` and `BarrierGuard` module instead.
-   *
-   * Holds if data flow through nodes guarded by `guard` is prohibited.
-   */
-  deprecated predicate isBarrierGuard(BarrierGuard guard) { none() }
-
-  /**
-   * DEPRECATED: Use `isBarrier` and `BarrierGuard` module instead.
-   *
-   * Holds if data flow through nodes guarded by `guard` is prohibited when
-   * the flow state is `state`
-   */
-  deprecated predicate isBarrierGuard(BarrierGuard guard, FlowState state) { none() }
-
   /**
    * Holds if data may flow from `node1` to `node2` in addition to the normal data-flow steps.
    */
@@ -225,29 +210,6 @@ abstract private class ConfigurationRecursionPrevention extends Configuration {
   }
 }
 
-/** A bridge class to access the deprecated `isBarrierGuard`. */
-private class BarrierGuardGuardedNodeBridge extends Unit {
-  abstract predicate guardedNode(Node n, Configuration config);
-
-  abstract predicate guardedNode(Node n, FlowState state, Configuration config);
-}
-
-private class BarrierGuardGuardedNode extends BarrierGuardGuardedNodeBridge {
-  deprecated override predicate guardedNode(Node n, Configuration config) {
-    exists(BarrierGuard g |
-      config.isBarrierGuard(g) and
-      n = g.getAGuardedNode()
-    )
-  }
-
-  deprecated override predicate guardedNode(Node n, FlowState state, Configuration config) {
-    exists(BarrierGuard g |
-      config.isBarrierGuard(g, state) and
-      n = g.getAGuardedNode()
-    )
-  }
-}
-
 private FlowState relevantState(Configuration config) {
   config.isSource(_, result) or
   config.isSink(_, result) or
@@ -288,9 +250,7 @@ private module Config implements FullStateConfigSig {
 
   predicate isBarrier(Node node, FlowState state) {
     getConfig(state).isBarrier(node, getState(state)) or
-    getConfig(state).isBarrier(node) or
-    any(BarrierGuardGuardedNodeBridge b).guardedNode(node, getState(state), getConfig(state)) or
-    any(BarrierGuardGuardedNodeBridge b).guardedNode(node, getConfig(state))
+    getConfig(state).isBarrier(node)
   }
 
   predicate isBarrierIn(Node node) { any(Configuration config).isBarrierIn(node) }

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

@@ -874,28 +874,3 @@ module BarrierGuard<guardChecksSig/3 guardChecks> {
     )
   }
 }
-
-/**
- * DEPRECATED: Use `BarrierGuard` module instead.
- *
- * A guard that validates some expression.
- *
- * To use this in a configuration, extend the class and provide a
- * characteristic predicate precisely specifying the guard, and override
- * `checks` to specify what is being validated and in which branch.
- *
- * It is important that all extending classes in scope are disjoint.
- */
-deprecated class BarrierGuard extends GuardCondition {
-  /** Override this predicate to hold if this guard validates `e` upon evaluating to `b`. */
-  abstract predicate checks(Expr e, boolean b);
-
-  /** Gets a node guarded by this guard. */
-  final ExprNode getAGuardedNode() {
-    exists(SsaDefinition def, Variable v, boolean branch |
-      result.getExpr() = def.getAUse(v) and
-      this.checks(def.getAUse(v), branch) and
-      this.controls(result.getExpr().getBasicBlock(), branch)
-    )
-  }
-}

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

@@ -116,33 +116,6 @@ abstract class Configuration extends DataFlow::Configuration {
 
   final override predicate isBarrierOut(DataFlow::Node node) { this.isSanitizerOut(node) }
 
-  /**
-   * DEPRECATED: Use `isSanitizer` and `BarrierGuard` module instead.
-   *
-   * Holds if taint propagation through nodes guarded by `guard` is prohibited.
-   */
-  deprecated predicate isSanitizerGuard(DataFlow::BarrierGuard guard) { none() }
-
-  deprecated final override predicate isBarrierGuard(DataFlow::BarrierGuard guard) {
-    this.isSanitizerGuard(guard)
-  }
-
-  /**
-   * DEPRECATED: Use `isSanitizer` and `BarrierGuard` module instead.
-   *
-   * Holds if taint propagation through nodes guarded by `guard` is prohibited
-   * when the flow state is `state`.
-   */
-  deprecated predicate isSanitizerGuard(DataFlow::BarrierGuard guard, DataFlow::FlowState state) {
-    none()
-  }
-
-  deprecated final override predicate isBarrierGuard(
-    DataFlow::BarrierGuard guard, DataFlow::FlowState state
-  ) {
-    this.isSanitizerGuard(guard, state)
-  }
-
   /**
    * Holds if taint may propagate from `node1` to `node2` in addition to the normal data-flow and taint steps.
    */

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

@@ -116,33 +116,6 @@ abstract class Configuration extends DataFlow::Configuration {
 
   final override predicate isBarrierOut(DataFlow::Node node) { this.isSanitizerOut(node) }
 
-  /**
-   * DEPRECATED: Use `isSanitizer` and `BarrierGuard` module instead.
-   *
-   * Holds if taint propagation through nodes guarded by `guard` is prohibited.
-   */
-  deprecated predicate isSanitizerGuard(DataFlow::BarrierGuard guard) { none() }
-
-  deprecated final override predicate isBarrierGuard(DataFlow::BarrierGuard guard) {
-    this.isSanitizerGuard(guard)
-  }
-
-  /**
-   * DEPRECATED: Use `isSanitizer` and `BarrierGuard` module instead.
-   *
-   * Holds if taint propagation through nodes guarded by `guard` is prohibited
-   * when the flow state is `state`.
-   */
-  deprecated predicate isSanitizerGuard(DataFlow::BarrierGuard guard, DataFlow::FlowState state) {
-    none()
-  }
-
-  deprecated final override predicate isBarrierGuard(
-    DataFlow::BarrierGuard guard, DataFlow::FlowState state
-  ) {
-    this.isSanitizerGuard(guard, state)
-  }
-
   /**
    * Holds if taint may propagate from `node1` to `node2` in addition to the normal data-flow and taint steps.
    */

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

@@ -7,9 +7,12 @@ private import DataFlowImplCommon as DataFlowImplCommon
 
 /**
  * Gets a function that might be called by `call`.
+ *
+ * This predicate does not take additional call targets
+ * from `AdditionalCallTarget` into account.
  */
 cached
-DataFlowCallable viableCallable(DataFlowCall call) {
+DataFlowCallable defaultViableCallable(DataFlowCall call) {
   DataFlowImplCommon::forceCachingInSameStage() and
   result = call.getStaticCallTarget()
   or
@@ -29,6 +32,17 @@ DataFlowCallable viableCallable(DataFlowCall call) {
   result = call.(VirtualDispatch::DataSensitiveCall).resolve()
 }
 
+/**
+ * Gets a function that might be called by `call`.
+ */
+cached
+DataFlowCallable viableCallable(DataFlowCall call) {
+  result = defaultViableCallable(call)
+  or
+  // Additional call targets
+  result = any(AdditionalCallTarget additional).viableTarget(call.getUnconvertedResultExpression())
+}
+
 /**
  * Provides virtual dispatch support compatible with the original
  * implementation of `semmle.code.cpp.security.TaintTracking`.

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

@@ -91,21 +91,6 @@ abstract class Configuration extends string {
   /** Holds if data flow out of `node` is prohibited. */
   predicate isBarrierOut(Node node) { none() }
 
-  /**
-   * DEPRECATED: Use `isBarrier` and `BarrierGuard` module instead.
-   *
-   * Holds if data flow through nodes guarded by `guard` is prohibited.
-   */
-  deprecated predicate isBarrierGuard(BarrierGuard guard) { none() }
-
-  /**
-   * DEPRECATED: Use `isBarrier` and `BarrierGuard` module instead.
-   *
-   * Holds if data flow through nodes guarded by `guard` is prohibited when
-   * the flow state is `state`
-   */
-  deprecated predicate isBarrierGuard(BarrierGuard guard, FlowState state) { none() }
-
   /**
    * Holds if data may flow from `node1` to `node2` in addition to the normal data-flow steps.
    */
@@ -225,29 +210,6 @@ abstract private class ConfigurationRecursionPrevention extends Configuration {
   }
 }
 
-/** A bridge class to access the deprecated `isBarrierGuard`. */
-private class BarrierGuardGuardedNodeBridge extends Unit {
-  abstract predicate guardedNode(Node n, Configuration config);
-
-  abstract predicate guardedNode(Node n, FlowState state, Configuration config);
-}
-
-private class BarrierGuardGuardedNode extends BarrierGuardGuardedNodeBridge {
-  deprecated override predicate guardedNode(Node n, Configuration config) {
-    exists(BarrierGuard g |
-      config.isBarrierGuard(g) and
-      n = g.getAGuardedNode()
-    )
-  }
-
-  deprecated override predicate guardedNode(Node n, FlowState state, Configuration config) {
-    exists(BarrierGuard g |
-      config.isBarrierGuard(g, state) and
-      n = g.getAGuardedNode()
-    )
-  }
-}
-
 private FlowState relevantState(Configuration config) {
   config.isSource(_, result) or
   config.isSink(_, result) or
@@ -288,9 +250,7 @@ private module Config implements FullStateConfigSig {
 
   predicate isBarrier(Node node, FlowState state) {
     getConfig(state).isBarrier(node, getState(state)) or
-    getConfig(state).isBarrier(node) or
-    any(BarrierGuardGuardedNodeBridge b).guardedNode(node, getState(state), getConfig(state)) or
-    any(BarrierGuardGuardedNodeBridge b).guardedNode(node, getConfig(state))
+    getConfig(state).isBarrier(node)
   }
 
   predicate isBarrierIn(Node node) { any(Configuration config).isBarrierIn(node) }

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

@@ -91,21 +91,6 @@ abstract class Configuration extends string {
   /** Holds if data flow out of `node` is prohibited. */
   predicate isBarrierOut(Node node) { none() }
 
-  /**
-   * DEPRECATED: Use `isBarrier` and `BarrierGuard` module instead.
-   *
-   * Holds if data flow through nodes guarded by `guard` is prohibited.
-   */
-  deprecated predicate isBarrierGuard(BarrierGuard guard) { none() }
-
-  /**
-   * DEPRECATED: Use `isBarrier` and `BarrierGuard` module instead.
-   *
-   * Holds if data flow through nodes guarded by `guard` is prohibited when
-   * the flow state is `state`
-   */
-  deprecated predicate isBarrierGuard(BarrierGuard guard, FlowState state) { none() }
-
   /**
    * Holds if data may flow from `node1` to `node2` in addition to the normal data-flow steps.
    */
@@ -225,29 +210,6 @@ abstract private class ConfigurationRecursionPrevention extends Configuration {
   }
 }
 
-/** A bridge class to access the deprecated `isBarrierGuard`. */
-private class BarrierGuardGuardedNodeBridge extends Unit {
-  abstract predicate guardedNode(Node n, Configuration config);
-
-  abstract predicate guardedNode(Node n, FlowState state, Configuration config);
-}
-
-private class BarrierGuardGuardedNode extends BarrierGuardGuardedNodeBridge {
-  deprecated override predicate guardedNode(Node n, Configuration config) {
-    exists(BarrierGuard g |
-      config.isBarrierGuard(g) and
-      n = g.getAGuardedNode()
-    )
-  }
-
-  deprecated override predicate guardedNode(Node n, FlowState state, Configuration config) {
-    exists(BarrierGuard g |
-      config.isBarrierGuard(g, state) and
-      n = g.getAGuardedNode()
-    )
-  }
-}
-
 private FlowState relevantState(Configuration config) {
   config.isSource(_, result) or
   config.isSink(_, result) or
@@ -288,9 +250,7 @@ private module Config implements FullStateConfigSig {
 
   predicate isBarrier(Node node, FlowState state) {
     getConfig(state).isBarrier(node, getState(state)) or
-    getConfig(state).isBarrier(node) or
-    any(BarrierGuardGuardedNodeBridge b).guardedNode(node, getState(state), getConfig(state)) or
-    any(BarrierGuardGuardedNodeBridge b).guardedNode(node, getConfig(state))
+    getConfig(state).isBarrier(node)
   }
 
   predicate isBarrierIn(Node node) { any(Configuration config).isBarrierIn(node) }

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

@@ -91,21 +91,6 @@ abstract class Configuration extends string {
   /** Holds if data flow out of `node` is prohibited. */
   predicate isBarrierOut(Node node) { none() }
 
-  /**
-   * DEPRECATED: Use `isBarrier` and `BarrierGuard` module instead.
-   *
-   * Holds if data flow through nodes guarded by `guard` is prohibited.
-   */
-  deprecated predicate isBarrierGuard(BarrierGuard guard) { none() }
-
-  /**
-   * DEPRECATED: Use `isBarrier` and `BarrierGuard` module instead.
-   *
-   * Holds if data flow through nodes guarded by `guard` is prohibited when
-   * the flow state is `state`
-   */
-  deprecated predicate isBarrierGuard(BarrierGuard guard, FlowState state) { none() }
-
   /**
    * Holds if data may flow from `node1` to `node2` in addition to the normal data-flow steps.
    */
@@ -225,29 +210,6 @@ abstract private class ConfigurationRecursionPrevention extends Configuration {
   }
 }
 
-/** A bridge class to access the deprecated `isBarrierGuard`. */
-private class BarrierGuardGuardedNodeBridge extends Unit {
-  abstract predicate guardedNode(Node n, Configuration config);
-
-  abstract predicate guardedNode(Node n, FlowState state, Configuration config);
-}
-
-private class BarrierGuardGuardedNode extends BarrierGuardGuardedNodeBridge {
-  deprecated override predicate guardedNode(Node n, Configuration config) {
-    exists(BarrierGuard g |
-      config.isBarrierGuard(g) and
-      n = g.getAGuardedNode()
-    )
-  }
-
-  deprecated override predicate guardedNode(Node n, FlowState state, Configuration config) {
-    exists(BarrierGuard g |
-      config.isBarrierGuard(g, state) and
-      n = g.getAGuardedNode()
-    )
-  }
-}
-
 private FlowState relevantState(Configuration config) {
   config.isSource(_, result) or
   config.isSink(_, result) or
@@ -288,9 +250,7 @@ private module Config implements FullStateConfigSig {
 
   predicate isBarrier(Node node, FlowState state) {
     getConfig(state).isBarrier(node, getState(state)) or
-    getConfig(state).isBarrier(node) or
-    any(BarrierGuardGuardedNodeBridge b).guardedNode(node, getState(state), getConfig(state)) or
-    any(BarrierGuardGuardedNodeBridge b).guardedNode(node, getConfig(state))
+    getConfig(state).isBarrier(node)
   }
 
   predicate isBarrierIn(Node node) { any(Configuration config).isBarrierIn(node) }

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

@@ -91,21 +91,6 @@ abstract class Configuration extends string {
   /** Holds if data flow out of `node` is prohibited. */
   predicate isBarrierOut(Node node) { none() }
 
-  /**
-   * DEPRECATED: Use `isBarrier` and `BarrierGuard` module instead.
-   *
-   * Holds if data flow through nodes guarded by `guard` is prohibited.
-   */
-  deprecated predicate isBarrierGuard(BarrierGuard guard) { none() }
-
-  /**
-   * DEPRECATED: Use `isBarrier` and `BarrierGuard` module instead.
-   *
-   * Holds if data flow through nodes guarded by `guard` is prohibited when
-   * the flow state is `state`
-   */
-  deprecated predicate isBarrierGuard(BarrierGuard guard, FlowState state) { none() }
-
   /**
    * Holds if data may flow from `node1` to `node2` in addition to the normal data-flow steps.
    */
@@ -225,29 +210,6 @@ abstract private class ConfigurationRecursionPrevention extends Configuration {
   }
 }
 
-/** A bridge class to access the deprecated `isBarrierGuard`. */
-private class BarrierGuardGuardedNodeBridge extends Unit {
-  abstract predicate guardedNode(Node n, Configuration config);
-
-  abstract predicate guardedNode(Node n, FlowState state, Configuration config);
-}
-
-private class BarrierGuardGuardedNode extends BarrierGuardGuardedNodeBridge {
-  deprecated override predicate guardedNode(Node n, Configuration config) {
-    exists(BarrierGuard g |
-      config.isBarrierGuard(g) and
-      n = g.getAGuardedNode()
-    )
-  }
-
-  deprecated override predicate guardedNode(Node n, FlowState state, Configuration config) {
-    exists(BarrierGuard g |
-      config.isBarrierGuard(g, state) and
-      n = g.getAGuardedNode()
-    )
-  }
-}
-
 private FlowState relevantState(Configuration config) {
   config.isSource(_, result) or
   config.isSink(_, result) or
@@ -288,9 +250,7 @@ private module Config implements FullStateConfigSig {
 
   predicate isBarrier(Node node, FlowState state) {
     getConfig(state).isBarrier(node, getState(state)) or
-    getConfig(state).isBarrier(node) or
-    any(BarrierGuardGuardedNodeBridge b).guardedNode(node, getState(state), getConfig(state)) or
-    any(BarrierGuardGuardedNodeBridge b).guardedNode(node, getConfig(state))
+    getConfig(state).isBarrier(node)
   }
 
   predicate isBarrierIn(Node node) { any(Configuration config).isBarrierIn(node) }

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

@@ -555,7 +555,7 @@ predicate instructionForFullyConvertedCall(Instruction instr, CallInstruction ca
 }
 
 /** Holds if `node` represents the output node for `call`. */
-private predicate simpleOutNode(Node node, CallInstruction call) {
+predicate simpleOutNode(Node node, CallInstruction call) {
   operandForFullyConvertedCall(node.asOperand(), call)
   or
   instructionForFullyConvertedCall(node.asInstruction(), call)

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

@@ -14,6 +14,7 @@ private import DataFlowPrivate
 private import ModelUtil
 private import SsaInternals as Ssa
 private import DataFlowImplCommon as DataFlowImplCommon
+private import codeql.util.Unit
 
 /**
  * The IR dataflow graph consists of the following nodes:
@@ -1696,16 +1697,7 @@ private module Cached {
     // Reverse flow: data that flows from the definition node back into the indirection returned
     // by a function. This allows data to flow 'in' through references returned by a modeled
     // function such as `operator[]`.
-    exists(Operand address, int indirectionIndex |
-      nodeHasOperand(nodeTo.(IndirectReturnOutNode), address, indirectionIndex)
-    |
-      exists(StoreInstruction store |
-        nodeHasInstruction(nodeFrom, store, indirectionIndex - 1) and
-        store.getDestinationAddressOperand() = address
-      )
-      or
-      Ssa::outNodeHasAddressAndIndex(nodeFrom, address, indirectionIndex)
-    )
+    reverseFlow(nodeFrom, nodeTo)
   }
 
   private predicate simpleInstructionLocalFlowStep(Operand opFrom, Instruction iTo) {
@@ -1736,6 +1728,39 @@ private module Cached {
       )
     )
   }
+
+  private predicate reverseFlow(Node nodeFrom, Node nodeTo) {
+    reverseFlowOperand(nodeFrom, nodeTo)
+    or
+    reverseFlowInstruction(nodeFrom, nodeTo)
+  }
+
+  private predicate reverseFlowOperand(Node nodeFrom, IndirectReturnOutNode nodeTo) {
+    exists(Operand address, int indirectionIndex |
+      nodeHasOperand(nodeTo, address, indirectionIndex)
+    |
+      exists(StoreInstruction store |
+        nodeHasInstruction(nodeFrom, store, indirectionIndex - 1) and
+        store.getDestinationAddressOperand() = address
+      )
+      or
+      // We also want a write coming out of an `OutNode` to flow `nodeTo`.
+      // This is different from `reverseFlowInstruction` since `nodeFrom` can never
+      // be an `OutNode` when it's defined by an instruction.
+      Ssa::outNodeHasAddressAndIndex(nodeFrom, address, indirectionIndex)
+    )
+  }
+
+  private predicate reverseFlowInstruction(Node nodeFrom, IndirectReturnOutNode nodeTo) {
+    exists(Instruction address, int indirectionIndex |
+      nodeHasInstruction(nodeTo, address, indirectionIndex)
+    |
+      exists(StoreInstruction store |
+        nodeHasInstruction(nodeFrom, store, indirectionIndex - 1) and
+        store.getDestinationAddress() = address
+      )
+    )
+  }
 }
 
 import Cached
@@ -2215,33 +2240,41 @@ module InstructionBarrierGuard<instructionGuardChecksSig/3 instructionGuardCheck
 }
 
 /**
- * DEPRECATED: Use `BarrierGuard` module instead.
+ * A unit class for adding additional call steps.
  *
- * A guard that validates some instruction.
+ * Extend this class to add additional call steps to the data flow graph.
  *
- * To use this in a configuration, extend the class and provide a
- * characteristic predicate precisely specifying the guard, and override
- * `checks` to specify what is being validated and in which branch.
+ * For example, if the following subclass is added:
+ * ```ql
+ * class MyAdditionalCallTarget extends DataFlow::AdditionalCallTarget {
+ *   override Function viableTarget(Call call) {
+ *     call.getTarget().hasName("f") and
+ *     result.hasName("g")
+ *   }
+ * }
+ * ```
+ * then flow from `source()` to `x` in `sink(x)` is reported in the following example:
+ * ```cpp
+ * void sink(int);
+ * int source();
+ * void f(int);
  *
- * It is important that all extending classes in scope are disjoint.
+ * void g(int x) {
+ *   sink(x);
+ * }
+ *
+ * void test() {
+ *   int x = source();
+ *   f(x);
+ * }
+ * ```
+ *
+ * Note: To prevent reevaluation of cached dataflow-related predicates any
+ * subclass of `AdditionalCallTarget` must be imported in all dataflow queries.
  */
-deprecated class BarrierGuard extends IRGuardCondition {
-  /** Override this predicate to hold if this guard validates `instr` upon evaluating to `b`. */
-  predicate checksInstr(Instruction instr, boolean b) { none() }
-
-  /** Override this predicate to hold if this guard validates `expr` upon evaluating to `b`. */
-  predicate checks(Expr e, boolean b) { none() }
-
-  /** Gets a node guarded by this guard. */
-  final Node getAGuardedNode() {
-    exists(ValueNumber value, boolean edge |
-      (
-        this.checksInstr(value.getAnInstruction(), edge)
-        or
-        this.checks(value.getAnInstruction().getConvertedResultExpression(), edge)
-      ) and
-      result.asInstruction() = value.getAnInstruction() and
-      this.controls(result.asInstruction().getBlock(), edge)
-    )
-  }
+class AdditionalCallTarget extends Unit {
+  /**
+   * Gets a viable target for `call`.
+   */
+  abstract DataFlowCallable viableTarget(Call call);
 }

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

@@ -6,6 +6,7 @@
 private import semmle.code.cpp.ir.IR
 private import semmle.code.cpp.ir.dataflow.DataFlow
 private import DataFlowUtil
+private import DataFlowPrivate
 private import SsaInternals as Ssa
 
 /**
@@ -35,7 +36,7 @@ DataFlow::Node callInput(CallInstruction call, FunctionInput input) {
  */
 Node callOutput(CallInstruction call, FunctionOutput output) {
   // The return value
-  result.asInstruction() = call and
+  simpleOutNode(result, call) and
   output.isReturnValue()
   or
   // The side effect of a call on the value pointed to by an argument or qualifier
@@ -82,7 +83,7 @@ Node callOutput(CallInstruction call, FunctionOutput output, int d) {
     // If there isn't an indirect out node for the call with indirection `d` then
     // we conflate this with the underlying `CallInstruction`.
     not exists(getIndirectReturnOutNode(call, d)) and
-    n.asInstruction() = result.asInstruction()
+    n = result
     or
     // The side effect of a call on the value pointed to by an argument or qualifier
     exists(Operand operand, int indirectionIndex |

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

@@ -374,6 +374,8 @@ module ProductFlow {
 
       predicate isBarrier(DataFlow::Node node, FlowState state) { Config::isBarrier1(node, state) }
 
+      predicate isBarrier(DataFlow::Node node) { Config::isBarrier1(node) }
+
       predicate isBarrierOut(DataFlow::Node node) { Config::isBarrierOut1(node) }
 
       predicate isAdditionalFlowStep(
@@ -408,6 +410,8 @@ module ProductFlow {
 
       predicate isBarrier(DataFlow::Node node, FlowState state) { Config::isBarrier2(node, state) }
 
+      predicate isBarrier(DataFlow::Node node) { Config::isBarrier2(node) }
+
       predicate isBarrierOut(DataFlow::Node node) { Config::isBarrierOut2(node) }
 
       predicate isAdditionalFlowStep(

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

@@ -645,12 +645,24 @@ private predicate adjustForPointerArith(PostUpdateNode pun, UseOrPhi use) {
   )
 }
 
+/**
+ * Holds if `nodeFrom` flows to `nodeTo` because there is `def-use` or
+ * `use-use` flow from `defOrUse` to `use`.
+ *
+ * `uncertain` is `true` if the `defOrUse` is an uncertain definition.
+ */
+private predicate localSsaFlow(
+  SsaDefOrUse defOrUse, Node nodeFrom, UseOrPhi use, Node nodeTo, boolean uncertain
+) {
+  nodeToDefOrUse(nodeFrom, defOrUse, uncertain) and
+  adjacentDefRead(defOrUse, use) and
+  useToNode(use, nodeTo) and
+  nodeFrom != nodeTo
+}
+
 private predicate ssaFlowImpl(SsaDefOrUse defOrUse, Node nodeFrom, Node nodeTo, boolean uncertain) {
   exists(UseOrPhi use |
-    nodeToDefOrUse(nodeFrom, defOrUse, uncertain) and
-    adjacentDefRead(defOrUse, use) and
-    useToNode(use, nodeTo) and
-    nodeFrom != nodeTo
+    localSsaFlow(defOrUse, nodeFrom, use, nodeTo, uncertain)
     or
     // Initial global variable value to a first use
     nodeFrom.(InitialGlobalValue).getGlobalDef() = defOrUse and
@@ -728,15 +740,62 @@ private predicate isArgumentOfCallable(DataFlowCall call, Node n) {
   )
 }
 
-/** Holds if there is def-use or use-use flow from `pun` to `nodeTo`. */
-predicate postUpdateFlow(PostUpdateNode pun, Node nodeTo) {
-  exists(UseOrPhi use, Node preUpdate |
+/**
+ * Holds if there is use-use flow from `pun`'s pre-update node to `n`.
+ */
+private predicate postUpdateNodeToFirstUse(PostUpdateNode pun, Node n) {
+  exists(UseOrPhi use |
     adjustForPointerArith(pun, use) and
-    useToNode(use, nodeTo) and
+    useToNode(use, n)
+  )
+}
+
+private predicate stepUntilNotInCall(DataFlowCall call, Node n1, Node n2) {
+  isArgumentOfCallable(call, n1) and
+  exists(Node mid | localSsaFlow(_, n1, _, mid, _) |
+    isArgumentOfCallable(call, mid) and
+    stepUntilNotInCall(call, mid, n2)
+    or
+    not isArgumentOfCallable(call, mid) and
+    mid = n2
+  )
+}
+
+bindingset[n1, n2]
+pragma[inline_late]
+private predicate isArgumentOfSameCall(DataFlowCall call, Node n1, Node n2) {
+  isArgumentOfCallable(call, n1) and isArgumentOfCallable(call, n2)
+}
+
+/**
+ * Holds if there is def-use or use-use flow from `pun` to `nodeTo`.
+ *
+ * Note: This is more complex than it sounds. Consider a call such as:
+ * ```cpp
+ * write_first_argument(x, x);
+ * sink(x);
+ * ```
+ * Assume flow comes out of the first argument to `write_first_argument`. We
+ * don't want flow to go to the `x` that's also an argument to
+ * `write_first_argument` (because we just flowed out of that function, and we
+ * don't want to flow back into it again).
+ *
+ * We do, however, want flow from the output argument to `x` on the next line, and
+ * similarly we want flow from the second argument of `write_first_argument` to `x`
+ * on the next line.
+ */
+predicate postUpdateFlow(PostUpdateNode pun, Node nodeTo) {
+  exists(Node preUpdate, Node mid |
     preUpdate = pun.getPreUpdateNode() and
-    not exists(DataFlowCall call |
-      isArgumentOfCallable(call, preUpdate) and isArgumentOfCallable(call, nodeTo)
+    postUpdateNodeToFirstUse(pun, mid)
+  |
+    exists(DataFlowCall call |
+      isArgumentOfSameCall(call, preUpdate, mid) and
+      stepUntilNotInCall(call, mid, nodeTo)
     )
+    or
+    not isArgumentOfSameCall(_, preUpdate, mid) and
+    nodeTo = mid
   )
 }
 

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

@@ -815,7 +815,7 @@ private module Cached {
   ) {
     indirectionIndex = [1 .. countIndirectionsForCppType(getResultLanguageType(instr))] and
     exists(Instruction load, Operand address |
-      address.getDef() = instr and
+      address = unique( | | getAUse(instr)) and
       isDereference(load, address, false) and
       instrRepr = load and
       indirectionIndexRepr = indirectionIndex - 1

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

@@ -116,33 +116,6 @@ abstract class Configuration extends DataFlow::Configuration {
 
   final override predicate isBarrierOut(DataFlow::Node node) { this.isSanitizerOut(node) }
 
-  /**
-   * DEPRECATED: Use `isSanitizer` and `BarrierGuard` module instead.
-   *
-   * Holds if taint propagation through nodes guarded by `guard` is prohibited.
-   */
-  deprecated predicate isSanitizerGuard(DataFlow::BarrierGuard guard) { none() }
-
-  deprecated final override predicate isBarrierGuard(DataFlow::BarrierGuard guard) {
-    this.isSanitizerGuard(guard)
-  }
-
-  /**
-   * DEPRECATED: Use `isSanitizer` and `BarrierGuard` module instead.
-   *
-   * Holds if taint propagation through nodes guarded by `guard` is prohibited
-   * when the flow state is `state`.
-   */
-  deprecated predicate isSanitizerGuard(DataFlow::BarrierGuard guard, DataFlow::FlowState state) {
-    none()
-  }
-
-  deprecated final override predicate isBarrierGuard(
-    DataFlow::BarrierGuard guard, DataFlow::FlowState state
-  ) {
-    this.isSanitizerGuard(guard, state)
-  }
-
   /**
    * Holds if taint may propagate from `node1` to `node2` in addition to the normal data-flow and taint steps.
    */

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

@@ -116,33 +116,6 @@ abstract class Configuration extends DataFlow::Configuration {
 
   final override predicate isBarrierOut(DataFlow::Node node) { this.isSanitizerOut(node) }
 
-  /**
-   * DEPRECATED: Use `isSanitizer` and `BarrierGuard` module instead.
-   *
-   * Holds if taint propagation through nodes guarded by `guard` is prohibited.
-   */
-  deprecated predicate isSanitizerGuard(DataFlow::BarrierGuard guard) { none() }
-
-  deprecated final override predicate isBarrierGuard(DataFlow::BarrierGuard guard) {
-    this.isSanitizerGuard(guard)
-  }
-
-  /**
-   * DEPRECATED: Use `isSanitizer` and `BarrierGuard` module instead.
-   *
-   * Holds if taint propagation through nodes guarded by `guard` is prohibited
-   * when the flow state is `state`.
-   */
-  deprecated predicate isSanitizerGuard(DataFlow::BarrierGuard guard, DataFlow::FlowState state) {
-    none()
-  }
-
-  deprecated final override predicate isBarrierGuard(
-    DataFlow::BarrierGuard guard, DataFlow::FlowState state
-  ) {
-    this.isSanitizerGuard(guard, state)
-  }
-
   /**
    * Holds if taint may propagate from `node1` to `node2` in addition to the normal data-flow and taint steps.
    */

cpp/ql/lib/semmle/code/cpp/ir/dataflow/internal/tainttracking3/TaintTrackingImpl.qll

@@ -116,33 +116,6 @@ abstract class Configuration extends DataFlow::Configuration {
 
   final override predicate isBarrierOut(DataFlow::Node node) { this.isSanitizerOut(node) }
 
-  /**
-   * DEPRECATED: Use `isSanitizer` and `BarrierGuard` module instead.
-   *
-   * Holds if taint propagation through nodes guarded by `guard` is prohibited.
-   */
-  deprecated predicate isSanitizerGuard(DataFlow::BarrierGuard guard) { none() }
-
-  deprecated final override predicate isBarrierGuard(DataFlow::BarrierGuard guard) {
-    this.isSanitizerGuard(guard)
-  }
-
-  /**
-   * DEPRECATED: Use `isSanitizer` and `BarrierGuard` module instead.
-   *
-   * Holds if taint propagation through nodes guarded by `guard` is prohibited
-   * when the flow state is `state`.
-   */
-  deprecated predicate isSanitizerGuard(DataFlow::BarrierGuard guard, DataFlow::FlowState state) {
-    none()
-  }
-
-  deprecated final override predicate isBarrierGuard(
-    DataFlow::BarrierGuard guard, DataFlow::FlowState state
-  ) {
-    this.isSanitizerGuard(guard, state)
-  }
-
   /**
    * Holds if taint may propagate from `node1` to `node2` in addition to the normal data-flow and taint steps.
    */

cpp/ql/lib/semmle/code/cpp/ir/implementation/aliased_ssa/IRBlock.qll

@@ -8,6 +8,22 @@ private import internal.IRBlockImports as Imports
 import Imports::EdgeKind
 private import Cached
 
+/**
+ * Holds if `block` is a block in `func` and `sortOverride`, `sortKey1`, and `sortKey2` are the
+ * sort keys of the block (derived from its first instruction)
+ */
+pragma[nomagic]
+private predicate blockSortKeys(
+  IRFunction func, IRBlockBase block, int sortOverride, int sortKey1, int sortKey2
+) {
+  block.getEnclosingIRFunction() = func and
+  block.getFirstInstruction().hasSortKeys(sortKey1, sortKey2) and
+  // Ensure that the block containing `EnterFunction` always comes first.
+  if block.getFirstInstruction() instanceof EnterFunctionInstruction
+  then sortOverride = 0
+  else sortOverride = 1
+}
+
 /**
  * A basic block in the IR. A basic block consists of a sequence of `Instructions` with the only
  * incoming edges at the beginning of the sequence and the only outgoing edges at the end of the
@@ -37,17 +53,14 @@ class IRBlockBase extends TIRBlock {
     exists(IRConfiguration::IRConfiguration config |
       config.shouldEvaluateDebugStringsForFunction(this.getEnclosingFunction())
     ) and
-    this =
-      rank[result + 1](IRBlock funcBlock, int sortOverride, int sortKey1, int sortKey2 |
-        funcBlock.getEnclosingFunction() = this.getEnclosingFunction() and
-        funcBlock.getFirstInstruction().hasSortKeys(sortKey1, sortKey2) and
-        // Ensure that the block containing `EnterFunction` always comes first.
-        if funcBlock.getFirstInstruction() instanceof EnterFunctionInstruction
-        then sortOverride = 0
-        else sortOverride = 1
-      |
-        funcBlock order by sortOverride, sortKey1, sortKey2
-      )
+    exists(IRFunction func |
+      this =
+        rank[result + 1](IRBlock funcBlock, int sortOverride, int sortKey1, int sortKey2 |
+          blockSortKeys(func, funcBlock, sortOverride, sortKey1, sortKey2)
+        |
+          funcBlock order by sortOverride, sortKey1, sortKey2
+        )
+    )
   }
 
   /**

cpp/ql/lib/semmle/code/cpp/ir/implementation/aliased_ssa/Instruction.qll

@@ -116,14 +116,14 @@ class Instruction extends Construction::TStageInstruction {
 
   private int getLineRank() {
     this.shouldGenerateDumpStrings() and
-    this =
-      rank[result](Instruction instr |
-        instr =
-          getAnInstructionAtLine(this.getEnclosingIRFunction(), this.getLocation().getFile(),
-            this.getLocation().getStartLine())
-      |
-        instr order by instr.getBlock().getDisplayIndex(), instr.getDisplayIndexInBlock()
-      )
+    exists(IRFunction enclosing, Language::File file, int line |
+      this =
+        rank[result](Instruction instr |
+          instr = getAnInstructionAtLine(enclosing, file, line)
+        |
+          instr order by instr.getBlock().getDisplayIndex(), instr.getDisplayIndexInBlock()
+        )
+    )
   }
 
   /**

cpp/ql/lib/semmle/code/cpp/ir/implementation/raw/IRBlock.qll

@@ -8,6 +8,22 @@ private import internal.IRBlockImports as Imports
 import Imports::EdgeKind
 private import Cached
 
+/**
+ * Holds if `block` is a block in `func` and `sortOverride`, `sortKey1`, and `sortKey2` are the
+ * sort keys of the block (derived from its first instruction)
+ */
+pragma[nomagic]
+private predicate blockSortKeys(
+  IRFunction func, IRBlockBase block, int sortOverride, int sortKey1, int sortKey2
+) {
+  block.getEnclosingIRFunction() = func and
+  block.getFirstInstruction().hasSortKeys(sortKey1, sortKey2) and
+  // Ensure that the block containing `EnterFunction` always comes first.
+  if block.getFirstInstruction() instanceof EnterFunctionInstruction
+  then sortOverride = 0
+  else sortOverride = 1
+}
+
 /**
  * A basic block in the IR. A basic block consists of a sequence of `Instructions` with the only
  * incoming edges at the beginning of the sequence and the only outgoing edges at the end of the
@@ -37,17 +53,14 @@ class IRBlockBase extends TIRBlock {
     exists(IRConfiguration::IRConfiguration config |
       config.shouldEvaluateDebugStringsForFunction(this.getEnclosingFunction())
     ) and
-    this =
-      rank[result + 1](IRBlock funcBlock, int sortOverride, int sortKey1, int sortKey2 |
-        funcBlock.getEnclosingFunction() = this.getEnclosingFunction() and
-        funcBlock.getFirstInstruction().hasSortKeys(sortKey1, sortKey2) and
-        // Ensure that the block containing `EnterFunction` always comes first.
-        if funcBlock.getFirstInstruction() instanceof EnterFunctionInstruction
-        then sortOverride = 0
-        else sortOverride = 1
-      |
-        funcBlock order by sortOverride, sortKey1, sortKey2
-      )
+    exists(IRFunction func |
+      this =
+        rank[result + 1](IRBlock funcBlock, int sortOverride, int sortKey1, int sortKey2 |
+          blockSortKeys(func, funcBlock, sortOverride, sortKey1, sortKey2)
+        |
+          funcBlock order by sortOverride, sortKey1, sortKey2
+        )
+    )
   }
 
   /**

cpp/ql/lib/semmle/code/cpp/ir/implementation/raw/Instruction.qll

@@ -116,14 +116,14 @@ class Instruction extends Construction::TStageInstruction {
 
   private int getLineRank() {
     this.shouldGenerateDumpStrings() and
-    this =
-      rank[result](Instruction instr |
-        instr =
-          getAnInstructionAtLine(this.getEnclosingIRFunction(), this.getLocation().getFile(),
-            this.getLocation().getStartLine())
-      |
-        instr order by instr.getBlock().getDisplayIndex(), instr.getDisplayIndexInBlock()
-      )
+    exists(IRFunction enclosing, Language::File file, int line |
+      this =
+        rank[result](Instruction instr |
+          instr = getAnInstructionAtLine(enclosing, file, line)
+        |
+          instr order by instr.getBlock().getDisplayIndex(), instr.getDisplayIndexInBlock()
+        )
+    )
   }
 
   /**

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

@@ -423,7 +423,12 @@ private module CachedForDebugging {
   cached
   predicate instructionHasSortKeys(Instruction instruction, int key1, int key2) {
     key1 = getInstructionTranslatedElement(instruction).getId() and
-    getInstructionTag(instruction) =
+    getInstructionTag(instruction) = tagByRank(key2)
+  }
+
+  pragma[nomagic]
+  private InstructionTag tagByRank(int key2) {
+    result =
       rank[key2](InstructionTag tag, string tagId |
         tagId = getInstructionTagId(tag)
       |

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

@@ -361,6 +361,12 @@ predicate ignoreLoad(Expr expr) {
     or
     expr instanceof FunctionAccess
     or
+    // The load is duplicated from the operand.
+    expr instanceof ParenthesisExpr
+    or
+    // The load is duplicated from the right operand.
+    expr instanceof CommaExpr
+    or
     expr.(PointerDereferenceExpr).getOperand().getFullyConverted().getType().getUnspecifiedType()
       instanceof FunctionPointerType
     or

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

@@ -648,19 +648,7 @@ abstract class TranslatedCrementOperation extends TranslatedNonConstantExpr {
 class TranslatedPrefixCrementOperation extends TranslatedCrementOperation {
   override PrefixCrementOperation expr;
 
-  override Instruction getResult() {
-    if expr.isPRValueCategory()
-    then
-      // If this is C, then the result of a prefix crement is a prvalue for the
-      // new value assigned to the operand. If this is C++, then the result is
-      // an lvalue, but that lvalue is being loaded as part of this expression.
-      // EDG doesn't mark this as a load.
-      result = this.getInstruction(CrementOpTag())
-    else
-      // This is C++, where the result is an lvalue for the operand, and that
-      // lvalue is not being loaded as part of this expression.
-      result = this.getUnloadedOperand().getResult()
-  }
+  override Instruction getResult() { result = this.getUnloadedOperand().getResult() }
 }
 
 class TranslatedPostfixCrementOperation extends TranslatedCrementOperation {
@@ -1503,19 +1491,7 @@ class TranslatedAssignExpr extends TranslatedNonConstantExpr {
     result = this.getRightOperand().getFirstInstruction()
   }
 
-  final override Instruction getResult() {
-    if expr.isPRValueCategory()
-    then
-      // If this is C, then the result of an assignment is a prvalue for the new
-      // value assigned to the left operand. If this is C++, then the result is
-      // an lvalue, but that lvalue is being loaded as part of this expression.
-      // EDG doesn't mark this as a load.
-      result = this.getRightOperand().getResult()
-    else
-      // This is C++, where the result is an lvalue for the left operand,
-      // and that lvalue is not being loaded as part of this expression.
-      result = this.getLeftOperand().getResult()
-  }
+  final override Instruction getResult() { result = this.getLeftOperand().getResult() }
 
   final TranslatedExpr getLeftOperand() {
     result = getTranslatedExpr(expr.getLValue().getFullyConverted())
@@ -1641,19 +1617,7 @@ class TranslatedAssignOperation extends TranslatedNonConstantExpr {
     result = this.getRightOperand().getFirstInstruction()
   }
 
-  final override Instruction getResult() {
-    if expr.isPRValueCategory()
-    then
-      // If this is C, then the result of an assignment is a prvalue for the new
-      // value assigned to the left operand. If this is C++, then the result is
-      // an lvalue, but that lvalue is being loaded as part of this expression.
-      // EDG doesn't mark this as a load.
-      result = this.getStoredValue()
-    else
-      // This is C++, where the result is an lvalue for the left operand,
-      // and that lvalue is not being loaded as part of this expression.
-      result = this.getUnloadedLeftOperand().getResult()
-  }
+  final override Instruction getResult() { result = this.getUnloadedLeftOperand().getResult() }
 
   final TranslatedExpr getUnloadedLeftOperand() {
     result = this.getLoadedLeftOperand().getOperand()
@@ -2191,8 +2155,15 @@ abstract class TranslatedConditionalExpr extends TranslatedNonConstantExpr {
         not this.elseIsVoid() and tag = ConditionValueFalseStoreTag()
       ) and
       opcode instanceof Opcode::Store and
-      resultType = this.getResultType()
+      (
+        not expr.hasLValueToRValueConversion() and
+        resultType = this.getResultType()
+        or
+        expr.hasLValueToRValueConversion() and
+        resultType = getTypeForPRValue(expr.getType())
+      )
       or
+      not expr.hasLValueToRValueConversion() and
       tag = ConditionValueResultLoadTag() and
       opcode instanceof Opcode::Load and
       resultType = this.getResultType()
@@ -2222,8 +2193,15 @@ abstract class TranslatedConditionalExpr extends TranslatedNonConstantExpr {
       )
       or
       tag = ConditionValueResultTempAddressTag() and
-      result = this.getInstruction(ConditionValueResultLoadTag())
+      (
+        not expr.hasLValueToRValueConversion() and
+        result = this.getInstruction(ConditionValueResultLoadTag())
+        or
+        expr.hasLValueToRValueConversion() and
+        result = this.getParent().getChildSuccessor(this)
+      )
       or
+      not expr.hasLValueToRValueConversion() and
       tag = ConditionValueResultLoadTag() and
       result = this.getParent().getChildSuccessor(this)
     )
@@ -2252,18 +2230,23 @@ abstract class TranslatedConditionalExpr extends TranslatedNonConstantExpr {
         result = this.getElse().getResult()
       )
       or
+      not expr.hasLValueToRValueConversion() and
       tag = ConditionValueResultLoadTag() and
-      (
-        operandTag instanceof AddressOperandTag and
-        result = this.getInstruction(ConditionValueResultTempAddressTag())
-      )
+      operandTag instanceof AddressOperandTag and
+      result = this.getInstruction(ConditionValueResultTempAddressTag())
     )
   }
 
   final override predicate hasTempVariable(TempVariableTag tag, CppType type) {
     not this.resultIsVoid() and
     tag = ConditionValueTempVar() and
-    type = this.getResultType()
+    (
+      not expr.hasLValueToRValueConversion() and
+      type = this.getResultType()
+      or
+      expr.hasLValueToRValueConversion() and
+      type = getTypeForPRValue(expr.getType())
+    )
   }
 
   final override IRVariable getInstructionVariable(InstructionTag tag) {
@@ -2278,7 +2261,13 @@ abstract class TranslatedConditionalExpr extends TranslatedNonConstantExpr {
 
   final override Instruction getResult() {
     not this.resultIsVoid() and
-    result = this.getInstruction(ConditionValueResultLoadTag())
+    (
+      expr.hasLValueToRValueConversion() and
+      result = this.getInstruction(ConditionValueResultTempAddressTag())
+      or
+      not expr.hasLValueToRValueConversion() and
+      result = this.getInstruction(ConditionValueResultLoadTag())
+    )
   }
 
   override Instruction getChildSuccessor(TranslatedElement child) {
@@ -3237,11 +3226,9 @@ predicate exprNeedsCopyIfNotLoaded(Expr expr) {
   (
     expr instanceof AssignExpr
     or
-    expr instanceof AssignOperation and
-    not expr.isPRValueCategory() // is C++
+    expr instanceof AssignOperation
     or
-    expr instanceof PrefixCrementOperation and
-    not expr.isPRValueCategory() // is C++
+    expr instanceof PrefixCrementOperation
     or
     // Because the load is on the `e` in `e++`.
     expr instanceof PostfixCrementOperation

cpp/ql/lib/semmle/code/cpp/ir/implementation/unaliased_ssa/IRBlock.qll

@@ -8,6 +8,22 @@ private import internal.IRBlockImports as Imports
 import Imports::EdgeKind
 private import Cached
 
+/**
+ * Holds if `block` is a block in `func` and `sortOverride`, `sortKey1`, and `sortKey2` are the
+ * sort keys of the block (derived from its first instruction)
+ */
+pragma[nomagic]
+private predicate blockSortKeys(
+  IRFunction func, IRBlockBase block, int sortOverride, int sortKey1, int sortKey2
+) {
+  block.getEnclosingIRFunction() = func and
+  block.getFirstInstruction().hasSortKeys(sortKey1, sortKey2) and
+  // Ensure that the block containing `EnterFunction` always comes first.
+  if block.getFirstInstruction() instanceof EnterFunctionInstruction
+  then sortOverride = 0
+  else sortOverride = 1
+}
+
 /**
  * A basic block in the IR. A basic block consists of a sequence of `Instructions` with the only
  * incoming edges at the beginning of the sequence and the only outgoing edges at the end of the
@@ -37,17 +53,14 @@ class IRBlockBase extends TIRBlock {
     exists(IRConfiguration::IRConfiguration config |
       config.shouldEvaluateDebugStringsForFunction(this.getEnclosingFunction())
     ) and
-    this =
-      rank[result + 1](IRBlock funcBlock, int sortOverride, int sortKey1, int sortKey2 |
-        funcBlock.getEnclosingFunction() = this.getEnclosingFunction() and
-        funcBlock.getFirstInstruction().hasSortKeys(sortKey1, sortKey2) and
-        // Ensure that the block containing `EnterFunction` always comes first.
-        if funcBlock.getFirstInstruction() instanceof EnterFunctionInstruction
-        then sortOverride = 0
-        else sortOverride = 1
-      |
-        funcBlock order by sortOverride, sortKey1, sortKey2
-      )
+    exists(IRFunction func |
+      this =
+        rank[result + 1](IRBlock funcBlock, int sortOverride, int sortKey1, int sortKey2 |
+          blockSortKeys(func, funcBlock, sortOverride, sortKey1, sortKey2)
+        |
+          funcBlock order by sortOverride, sortKey1, sortKey2
+        )
+    )
   }
 
   /**

cpp/ql/lib/semmle/code/cpp/ir/implementation/unaliased_ssa/Instruction.qll

@@ -116,14 +116,14 @@ class Instruction extends Construction::TStageInstruction {
 
   private int getLineRank() {
     this.shouldGenerateDumpStrings() and
-    this =
-      rank[result](Instruction instr |
-        instr =
-          getAnInstructionAtLine(this.getEnclosingIRFunction(), this.getLocation().getFile(),
-            this.getLocation().getStartLine())
-      |
-        instr order by instr.getBlock().getDisplayIndex(), instr.getDisplayIndexInBlock()
-      )
+    exists(IRFunction enclosing, Language::File file, int line |
+      this =
+        rank[result](Instruction instr |
+          instr = getAnInstructionAtLine(enclosing, file, line)
+        |
+          instr order by instr.getBlock().getDisplayIndex(), instr.getDisplayIndexInBlock()
+        )
+    )
   }
 
   /**

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

@@ -29,7 +29,7 @@ private import RangeAnalysisUtils
  * The SSA logic comes in two versions: the standard SSA and range-analysis RangeSSA.
  * This class provides the range-analysis SSA logic.
  */
-library class RangeSsa extends SsaHelper {
+class RangeSsa extends SsaHelper {
   RangeSsa() { this = 1 }
 
   /**

cpp/ql/lib/semmle/code/cpp/security/InvalidPointerDereference/AllocationToInvalidPointer.qll

@@ -60,17 +60,31 @@ private import semmle.code.cpp.rangeanalysis.new.RangeAnalysisUtil
 
 private VariableAccess getAVariableAccess(Expr e) { e.getAChild*() = result }
 
+/**
+ * Gets a (sub)expression that may be the result of evaluating `size`.
+ *
+ * For example, `getASizeCandidate(a ? b : c)` gives `a ? b : c`, `b` and `c`.
+ */
+bindingset[size]
+pragma[inline_late]
+private Expr getASizeCandidate(Expr size) {
+  result = size
+  or
+  result = [size.(ConditionalExpr).getThen(), size.(ConditionalExpr).getElse()]
+}
+
 /**
  * Holds if the `(n, state)` pair represents the source of flow for the size
  * expression associated with `alloc`.
  */
 predicate hasSize(HeuristicAllocationExpr alloc, DataFlow::Node n, int state) {
-  exists(VariableAccess va, Expr size, int delta |
+  exists(VariableAccess va, Expr size, int delta, Expr s |
     size = alloc.getSizeExpr() and
+    s = getASizeCandidate(size) and
     // Get the unique variable in a size expression like `x` in `malloc(x + 1)`.
-    va = unique( | | getAVariableAccess(size)) and
+    va = unique( | | getAVariableAccess(s)) and
     // Compute `delta` as the constant difference between `x` and `x + 1`.
-    bounded1(any(Instruction instr | instr.getUnconvertedResultExpression() = size),
+    bounded1(any(Instruction instr | instr.getUnconvertedResultExpression() = s),
       any(LoadInstruction load | load.getUnconvertedResultExpression() = va), delta) and
     n.asExpr() = va and
     state = delta
@@ -284,10 +298,33 @@ private module Config implements ProductFlow::StateConfigSig {
     pointerAddInstructionHasBounds0(_, allocSink, sizeSink, sizeAddend)
   }
 
+  private import semmle.code.cpp.ir.dataflow.internal.DataFlowPrivate
+
   predicate isBarrier2(DataFlow::Node node, FlowState2 state) {
     node = SizeBarrier::getABarrierNode(state)
   }
 
+  predicate isBarrier2(DataFlow::Node node) {
+    // Block flow from `*p` to `*(p + n)` when `n` is not `0`. This removes
+    // false positives
+    // when tracking the size of the allocation as an element of an array such
+    // as:
+    // ```
+    // size_t* p = new size_t[n];
+    // ...
+    // p[0] = n;
+    // int i = p[1];
+    // p[i] = ...
+    // ```
+    // In the above case, this barrier blocks flow from the indirect node
+    // for `p` to `p[1]`.
+    exists(Operand operand, PointerAddInstruction add |
+      node.(IndirectOperand).hasOperandAndIndirectionIndex(operand, _) and
+      add.getLeftOperand() = operand and
+      add.getRight().(ConstantInstruction).getValue() != "0"
+    )
+  }
+
   predicate isBarrierIn1(DataFlow::Node node) { isSourcePair(node, _, _, _) }
 
   predicate isBarrierOut2(DataFlow::Node node) {

cpp/ql/lib/semmlecode.cpp.dbscheme

@@ -612,6 +612,14 @@ case @builtintype.kind of
 | 51 = @char8_t
 | 52 = @float16               // _Float16
 | 53 = @complex_float16       // _Complex _Float16
+| 54 = @fp16                  // __fp16
+| 55 = @std_bfloat16          // __bf16
+| 56 = @std_float16           // std::float16_t
+| 57 = @complex_std_float32   // _Complex _Float32
+| 58 = @complex_float32x      // _Complex _Float32x
+| 59 = @complex_std_float64   // _Complex _Float64
+| 60 = @complex_float64x      // _Complex _Float64x
+| 61 = @complex_std_float128  // _Complex _Float128
 ;
 
 builtintypes(

cpp/ql/lib/upgrades/dbe9c8eb5fc6f54b7ae08c7317d0795b24961564/upgrade.properties

@@ -0,0 +1,2 @@
+description: Introduce new floating-point types from C23 and C++23
+compatibility: full

cpp/ql/src/CHANGELOG.md

@@ -1,3 +1,40 @@
+## 0.8.1
+
+### New Queries
+
+* The query `cpp/redundant-null-check-simple` has been promoted to Code Scanning. The query finds cases where a pointer is compared to null after it has already been dereferenced. Such comparisons likely indicate a bug at the place where the pointer is dereferenced, or where the pointer is compared to null.
+
+  Note: This query was incorrectly noted as being promoted to Code Scanning in CodeQL version 2.14.6.
+
+## 0.8.0
+
+### Query Metadata Changes
+
+* The `cpp/double-free` query has been further improved to reduce false positives and its precision has been increased from `medium` to `high`.
+* The `cpp/use-after-free` query has been further improved to reduce false positives and its precision has been increased from `medium` to `high`.
+
+### Minor Analysis Improvements
+
+* The queries `cpp/double-free` and `cpp/use-after-free` find fewer false positives
+  in cases where a non-returning function is called.
+* The number of duplicated dataflow paths reported by queries has been significantly reduced.
+
+## 0.7.5
+
+No user-facing changes.
+
+## 0.7.4
+
+### New Queries
+
+* Added a new query, `cpp/invalid-pointer-deref`, to detect out-of-bounds pointer reads and writes.
+
+### Minor Analysis Improvements
+
+* The "Comparison where assignment was intended" query (`cpp/compare-where-assign-meant`) no longer reports comparisons that appear in macro expansions.
+* Some queries that had repeated results corresponding to different levels of indirection for `argv` now only have a single result.
+* The `cpp/non-constant-format` query no longer considers an assignment on the right-hand side of another assignment to be a source of non-constant format strings. As a result, the query may now produce fewer results.
+
 ## 0.7.3
 
 No user-facing changes.

cpp/ql/src/Critical/DoubleFree.ql

@@ -2,7 +2,7 @@
  * @name Potential double free
  * @description Freeing a resource more than once can lead to undefined behavior and cause memory corruption.
  * @kind path-problem
- * @precision medium
+ * @precision high
  * @id cpp/double-free
  * @problem.severity warning
  * @security-severity 9.3

cpp/ql/src/Critical/UseAfterFree.ql

@@ -2,7 +2,7 @@
  * @name Potential use after free
  * @description An allocated memory block is used after it has been freed. Behavior in such cases is undefined and can cause memory corruption.
  * @kind path-problem
- * @precision medium
+ * @precision high
  * @id cpp/use-after-free
  * @problem.severity warning
  * @security-severity 9.3
@@ -29,8 +29,7 @@ private predicate externalCallNeverDereferences(FormattingFunctionCall call, int
   )
 }
 
-predicate isUse0(DataFlow::Node n, Expr e) {
-  e = n.asExpr() and
+predicate isUse0(Expr e) {
   not isFree(_, e, _) and
   (
     e = any(PointerDereferenceExpr pde).getOperand()
@@ -43,7 +42,7 @@ predicate isUse0(DataFlow::Node n, Expr e) {
     or
     // Assume any function without a body will dereference the pointer
     exists(int i, Call call, Function f |
-      n.asExpr() = call.getArgument(i) and
+      e = call.getArgument(i) and
       f = call.getTarget() and
       not f.hasEntryPoint() and
       // Exclude known functions we know won't dereference the pointer.
@@ -57,7 +56,7 @@ module ParameterSinks {
   import semmle.code.cpp.ir.ValueNumbering
 
   predicate flowsToUse(DataFlow::Node n) {
-    isUse0(n, _)
+    isUse0(n.asExpr())
     or
     exists(DataFlow::Node succ |
       flowsToUse(succ) and
@@ -90,7 +89,7 @@ module ParameterSinks {
   ) {
     pragma[only_bind_out](source.asParameter()) = pragma[only_bind_out](init.getParameter()) and
     paramToUse(source, sink) and
-    isUse0(sink, _)
+    isUse0(sink.asExpr())
   }
 
   private InitializeParameterInstruction getAnAlwaysDereferencedParameter0() {
@@ -139,7 +138,7 @@ module IsUse {
   private import semmle.code.cpp.ir.dataflow.internal.DataFlowImplCommon
 
   predicate isUse(DataFlow::Node n, Expr e) {
-    isUse0(n, e)
+    isUse0(e) and n.asExpr() = e
     or
     exists(CallInstruction call, InitializeParameterInstruction init |
       n.asOperand().getDef().getUnconvertedResultExpression() = e and

cpp/ql/src/Likely Bugs/RedundantNullCheckSimple.ql

@@ -5,10 +5,12 @@
  *              it should be moved before the dereference.
  * @kind path-problem
  * @problem.severity error
+ * @security-severity 7.5
  * @precision high
  * @id cpp/redundant-null-check-simple
  * @tags reliability
  *       correctness
+ *       security
  *       external/cwe/cwe-476
  */
 

cpp/ql/src/Microsoft/SAL.qll

@@ -161,7 +161,7 @@ private predicate annotatesAtPosition(SalPosition pos, DeclarationEntry d, File
  * A SAL element, that is, a SAL annotation or a declaration entry
  * that may have SAL annotations.
  */
-library class SalElement extends Element {
+class SalElement extends Element {
   SalElement() {
     containsSalAnnotation(this.(DeclarationEntry).getFile()) or
     this instanceof SalAnnotation

cpp/ql/src/Security/CWE/CWE-079/CgiXss.ql

@@ -13,15 +13,13 @@
 
 import cpp
 import semmle.code.cpp.commons.Environment
-import semmle.code.cpp.ir.dataflow.internal.DefaultTaintTrackingImpl
-import TaintedWithPath
+import semmle.code.cpp.ir.dataflow.TaintTracking
+import semmle.code.cpp.ir.IR
+import Flow::PathGraph
 
 /** A call that prints its arguments to `stdout`. */
 class PrintStdoutCall extends FunctionCall {
-  PrintStdoutCall() {
-    this.getTarget().hasGlobalOrStdName("puts") or
-    this.getTarget().hasGlobalOrStdName("printf")
-  }
+  PrintStdoutCall() { this.getTarget().hasGlobalOrStdName(["puts", "printf"]) }
 }
 
 /** A read of the QUERY_STRING environment variable */
@@ -29,19 +27,25 @@ class QueryString extends EnvironmentRead {
   QueryString() { this.getEnvironmentVariable() = "QUERY_STRING" }
 }
 
-class Configuration extends TaintTrackingConfiguration {
-  override predicate isSource(Expr source) { source instanceof QueryString }
+module Config implements DataFlow::ConfigSig {
+  predicate isSource(DataFlow::Node node) { node.asIndirectExpr() instanceof QueryString }
 
-  override predicate isSink(Element tainted) {
-    exists(PrintStdoutCall call | call.getAnArgument() = tainted)
+  predicate isSink(DataFlow::Node node) {
+    exists(PrintStdoutCall call | call.getAnArgument() = [node.asIndirectExpr(), node.asExpr()])
   }
 
-  override predicate isBarrier(Expr e) {
-    super.isBarrier(e) or e.getUnspecifiedType() instanceof IntegralType
+  predicate isBarrier(DataFlow::Node node) {
+    isSink(node) and node.asExpr().getUnspecifiedType() instanceof ArithmeticType
+    or
+    node.asInstruction().(StoreInstruction).getResultType() instanceof ArithmeticType
   }
 }
 
-from QueryString query, Element printedArg, PathNode sourceNode, PathNode sinkNode
-where taintedWithPath(query, printedArg, sourceNode, sinkNode)
-select printedArg, sourceNode, sinkNode, "Cross-site scripting vulnerability due to $@.", query,
-  "this query data"
+module Flow = TaintTracking::Global<Config>;
+
+from QueryString query, Flow::PathNode sourceNode, Flow::PathNode sinkNode
+where
+  Flow::flowPath(sourceNode, sinkNode) and
+  query = sourceNode.getNode().asIndirectExpr()
+select sinkNode.getNode(), sourceNode, sinkNode, "Cross-site scripting vulnerability due to $@.",
+  query, "this query data"

cpp/ql/src/Security/CWE/CWE-119/OverrunWriteProductFlow.ql

@@ -82,36 +82,20 @@ module ValidState {
    * library will perform, and visit all the places where the size argument is modified.
    * 2. Once that dataflow traversal is done, we accumulate the offsets added at each places
    * where the offset is modified (see `validStateImpl`).
-   *
-   * Because we want to guarantee that each place where we modify the offset has a `PathNode`
-   * we "flip" a boolean flow state in each `isAdditionalFlowStep`. This ensures that the node
-   * has a corresponding `PathNode`.
    */
-  private module ValidStateConfig implements DataFlow::StateConfigSig {
-    class FlowState = boolean;
+  private module ValidStateConfig implements DataFlow::ConfigSig {
+    predicate isSource(DataFlow::Node source) { hasSize(_, source, _) }
 
-    predicate isSource(DataFlow::Node source, FlowState state) {
-      hasSize(_, source, _) and
-      state = false
-    }
+    predicate isSink(DataFlow::Node sink) { isSinkPairImpl(_, _, sink, _, _) }
 
-    predicate isSink(DataFlow::Node sink, FlowState state) {
-      isSinkPairImpl(_, _, sink, _, _) and
-      state = [false, true]
-    }
-
-    predicate isAdditionalFlowStep(
-      DataFlow::Node node1, FlowState state1, DataFlow::Node node2, FlowState state2
-    ) {
-      isAdditionalFlowStep2(node1, node2, _) and
-      state1 = [false, true] and
-      state2 = state1.booleanNot()
+    predicate isAdditionalFlowStep(DataFlow::Node node1, DataFlow::Node node2) {
+      isAdditionalFlowStep2(node1, node2, _)
     }
 
     predicate includeHiddenNodes() { any() }
   }
 
-  private import DataFlow::GlobalWithState<ValidStateConfig>
+  private import DataFlow::Global<ValidStateConfig>
 
   private predicate inLoop(PathNode n) { n.getASuccessor+() = n }
 

cpp/ql/src/change-notes/2023-08-21-invalid-pointer-deref.md

@@ -1,4 +0,0 @@
----
-category: newQuery
----
-* Added a new query, `cpp/invalid-pointer-deref`, to detect out-of-bounds pointer reads and writes.

cpp/ql/src/change-notes/2023-08-24-no-taint-argv-indirections.md

@@ -1,4 +0,0 @@
----
-category: minorAnalysis
----
-* Some queries that had repeated results corresponding to different levels of indirection for `argv` now only have a single result.

cpp/ql/src/change-notes/2023-08-24-remove-non-constant-assign-sources.md

@@ -1,4 +0,0 @@
----
-category: minorAnalysis
----
-* The `cpp/non-constant-format` query no longer considers an assignment on the right-hand side of another assignment to be a source of non-constant format strings. As a result, the query may now produce fewer results.

cpp/ql/src/change-notes/2023-08-25-compare-where-assign-meant.md

@@ -1,4 +0,0 @@
----
-category: minorAnalysis
----
-* The "Comparison where assignment was intended" query (`cpp/compare-where-assign-meant`) no longer reports comparisons that appear in macro expansions.

cpp/ql/src/change-notes/2023-09-06-deduplicated-results.md

@@ -1,4 +0,0 @@
----
-category: minorAnalysis
----
-* The number of duplicated dataflow paths reported by queries has been significantly reduced.

cpp/ql/src/change-notes/2023-09-08-unreachble-edges.md

@@ -1,5 +0,0 @@
----
-category: minorAnalysis
----
-* The queries `cpp/double-free` and `cpp/use-after-free` find fewer false positives
-  in cases where a non-returning function is called.

cpp/ql/src/change-notes/released/0.7.4.md

@@ -0,0 +1,11 @@
+## 0.7.4
+
+### New Queries
+
+* Added a new query, `cpp/invalid-pointer-deref`, to detect out-of-bounds pointer reads and writes.
+
+### Minor Analysis Improvements
+
+* The "Comparison where assignment was intended" query (`cpp/compare-where-assign-meant`) no longer reports comparisons that appear in macro expansions.
+* Some queries that had repeated results corresponding to different levels of indirection for `argv` now only have a single result.
+* The `cpp/non-constant-format` query no longer considers an assignment on the right-hand side of another assignment to be a source of non-constant format strings. As a result, the query may now produce fewer results.

cpp/ql/src/change-notes/released/0.7.5.md

@@ -0,0 +1,3 @@
+## 0.7.5
+
+No user-facing changes.

cpp/ql/src/change-notes/released/0.8.0.md

@@ -0,0 +1,12 @@
+## 0.8.0
+
+### Query Metadata Changes
+
+* The `cpp/double-free` query has been further improved to reduce false positives and its precision has been increased from `medium` to `high`.
+* The `cpp/use-after-free` query has been further improved to reduce false positives and its precision has been increased from `medium` to `high`.
+
+### Minor Analysis Improvements
+
+* The queries `cpp/double-free` and `cpp/use-after-free` find fewer false positives
+  in cases where a non-returning function is called.
+* The number of duplicated dataflow paths reported by queries has been significantly reduced.

cpp/ql/src/change-notes/released/0.8.1.md

@@ -0,0 +1,7 @@
+## 0.8.1
+
+### New Queries
+
+* The query `cpp/redundant-null-check-simple` has been promoted to Code Scanning. The query finds cases where a pointer is compared to null after it has already been dereferenced. Such comparisons likely indicate a bug at the place where the pointer is dereferenced, or where the pointer is compared to null.
+
+  Note: This query was incorrectly noted as being promoted to Code Scanning in CodeQL version 2.14.6.