Compare commits

..

4 Commits

Author SHA1 Message Date
Nick Rolfe
4aa4364200 Merge remote-tracking branch 'origin/dependabot/cargo/ruby/extractor/clap-4.0' into nickrolfe/clap4 2022-09-29 09:31:15 +01:00
Nick Rolfe
88fb187923 Merge remote-tracking branch 'origin/dependabot/cargo/ruby/generator/clap-4.0' into nickrolfe/clap4 2022-09-29 09:30:53 +01:00
dependabot[bot]
753042d883 Update clap requirement from 3.0 to 4.0 in /ruby/extractor
Updates the requirements on [clap](https://github.com/clap-rs/clap) to permit the latest version.
- [Release notes](https://github.com/clap-rs/clap/releases)
- [Changelog](https://github.com/clap-rs/clap/blob/master/CHANGELOG.md)
- [Commits](https://github.com/clap-rs/clap/compare/clap_complete-v3.0.0...v4.0.2)

---
updated-dependencies:
- dependency-name: clap
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>
2022-09-29 03:19:36 +00:00
dependabot[bot]
5299cd9c69 Update clap requirement from 3.0 to 4.0 in /ruby/generator
Updates the requirements on [clap](https://github.com/clap-rs/clap) to permit the latest version.
- [Release notes](https://github.com/clap-rs/clap/releases)
- [Changelog](https://github.com/clap-rs/clap/blob/master/CHANGELOG.md)
- [Commits](https://github.com/clap-rs/clap/compare/clap_complete-v3.0.0...v4.0.2)

---
updated-dependencies:
- dependency-name: clap
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>
2022-09-29 03:19:22 +00:00
6831 changed files with 299705 additions and 460680 deletions

View File

@@ -0,0 +1,24 @@
---
name: LGTM.com - false positive
about: Tell us about an alert that shouldn't be reported
title: LGTM.com - false positive
labels: false-positive
assignees: ''
---
**Description of the false positive**
<!-- Please explain briefly why you think it shouldn't be included. -->
**URL to the alert on the project page on LGTM.com**
<!--
1. Open the project on LGTM.com.
For example, https://lgtm.com/projects/g/pallets/click/.
2. Switch to the `Alerts` tab. For example, https://lgtm.com/projects/g/pallets/click/alerts/.
3. Scroll to the alert that you would like to report.
4. Click on the right most icon `View this alert within the complete file`.
5. A new browser tab opens. Copy and paste the page URL here.
For example, https://lgtm.com/projects/g/pallets/click/snapshot/719fb7d8322b0767cdd1e5903ba3eb3233ba8dd5/files/click/_winconsole.py#xa08d213ab3289f87:1.
-->

View File

@@ -10,5 +10,5 @@ assignees: ''
**Description of the issue**
<!-- Please explain briefly what is the problem.
If it is about a GitHub project, please include its URL. -->
If it is about an LGTM project, please include its URL.-->

View File

@@ -1,36 +0,0 @@
---
name: CodeQL false positive
about: Report CodeQL alerts that you think should not have been detected (not applicable, not exploitable, etc.)
title: False positive
labels: false-positive
assignees: ''
---
**Description of the false positive**
<!-- Please explain briefly why you think it shouldn't be included. -->
**Code samples or links to source code**
<!--
For open source code: file links with line numbers on GitHub, for example:
https://github.com/github/codeql/blob/dc440aaee6695deb0d9676b87e06ea984e1b4ae5/javascript/ql/test/query-tests/Security/CWE-078/CommandInjection/exec-sh2.js#L10
For closed source code: (redacted) code samples that illustrate the problem, for example:
```
function execSh(command, options) {
return cp.spawn(getShell(), ["-c", command], options) // <- command line injection
};
```
-->
**URL to the alert on GitHub code scanning (optional)**
<!--
1. Open the project on GitHub.com.
2. Switch to the `Security` tab.
3. Browse to the alert that you would like to report.
4. Copy and paste the page URL here.
-->

View File

@@ -1,54 +0,0 @@
name: Cache query compilation
description: Caches CodeQL compilation caches - should be run both on PRs and pushes to main.
inputs:
key:
description: 'The cache key to use - should be unique to the workflow'
required: true
outputs:
cache-dir:
description: "The directory where the cache was stored"
value: ${{ steps.fill-compilation-dir.outputs.compdir }}
runs:
using: composite
steps:
# calculate the merge-base with main, in a way that works both on PRs and pushes to main.
- name: Calculate merge-base
shell: bash
if: ${{ github.event_name == 'pull_request' }}
env:
BASE_BRANCH: ${{ github.base_ref }}
run: |
MERGE_BASE=$(git cat-file commit $GITHUB_SHA | grep '^parent ' | head -1 | cut -f 2 -d " ")
echo "merge_base=$MERGE_BASE" >> $GITHUB_ENV
- name: Restore cache (PR)
if: ${{ github.event_name == 'pull_request' }}
uses: actions/cache/restore@v3
with:
path: '**/.cache'
key: codeql-compile-${{ inputs.key }}-pr-${{ github.sha }}
restore-keys: |
codeql-compile-${{ inputs.key }}-${{ github.base_ref }}-${{ env.merge_base }}
codeql-compile-${{ inputs.key }}-${{ github.base_ref }}-
codeql-compile-${{ inputs.key }}-main-
- name: Fill cache (only branch push)
if: ${{ github.event_name != 'pull_request' }}
uses: actions/cache@v3
with:
path: '**/.cache'
key: codeql-compile-${{ inputs.key }}-${{ github.ref_name }}-${{ github.sha }} # just fill on main
restore-keys: | # restore the latest cache if the exact cache is unavailable, to speed up compilation.
codeql-compile-${{ inputs.key }}-${{ github.ref_name }}-
codeql-compile-${{ inputs.key }}-main-
- name: Fill compilation cache directory
id: fill-compilation-dir
shell: bash
run: |
# Move all the existing cache into another folder, so we only preserve the cache for the current queries.
node $GITHUB_WORKSPACE/.github/actions/cache-query-compilation/move-caches.js ${COMBINED_CACHE_DIR}
echo "compdir=${COMBINED_CACHE_DIR}" >> $GITHUB_OUTPUT
env:
COMBINED_CACHE_DIR: ${{ runner.temp }}/compilation-dir

View File

@@ -1,75 +0,0 @@
// # Move all the existing cache into another folder, so we only preserve the cache for the current queries.
// mkdir -p ${COMBINED_CACHE_DIR}
// rm -f **/.cache/{lock,size} # -f to avoid errors if the cache is empty.
// # copy the contents of the .cache folders into the combined cache folder.
// cp -r **/.cache/* ${COMBINED_CACHE_DIR}/ || : # ignore missing files
// # clean up the .cache folders
// rm -rf **/.cache/*
const fs = require("fs");
const path = require("path");
// the first argv is the cache folder to create.
const COMBINED_CACHE_DIR = process.argv[2];
function* walkCaches(dir) {
const files = fs.readdirSync(dir, { withFileTypes: true });
for (const file of files) {
if (file.isDirectory()) {
const filePath = path.join(dir, file.name);
yield* walkCaches(filePath);
if (file.name === ".cache") {
yield filePath;
}
}
}
}
async function copyDir(src, dest) {
for await (const file of await fs.promises.readdir(src, { withFileTypes: true })) {
const srcPath = path.join(src, file.name);
const destPath = path.join(dest, file.name);
if (file.isDirectory()) {
if (!fs.existsSync(destPath)) {
fs.mkdirSync(destPath);
}
await copyDir(srcPath, destPath);
} else {
await fs.promises.copyFile(srcPath, destPath);
}
}
}
async function main() {
const cacheDirs = [...walkCaches(".")];
for (const dir of cacheDirs) {
console.log(`Found .cache dir at ${dir}`);
}
// mkdir -p ${COMBINED_CACHE_DIR}
fs.mkdirSync(COMBINED_CACHE_DIR, { recursive: true });
// rm -f **/.cache/{lock,size} # -f to avoid errors if the cache is empty.
await Promise.all(
cacheDirs.map((cacheDir) =>
(async function () {
await fs.promises.rm(path.join(cacheDir, "lock"), { force: true });
await fs.promises.rm(path.join(cacheDir, "size"), { force: true });
})()
)
);
// # copy the contents of the .cache folders into the combined cache folder.
// cp -r **/.cache/* ${COMBINED_CACHE_DIR}/ || : # ignore missing files
await Promise.all(
cacheDirs.map((cacheDir) => copyDir(cacheDir, COMBINED_CACHE_DIR))
);
// # clean up the .cache folders
// rm -rf **/.cache/*
await Promise.all(
cacheDirs.map((cacheDir) => fs.promises.rm(cacheDir, { recursive: true }))
);
}
main();

View File

@@ -1,24 +1,14 @@
name: Fetch CodeQL
description: Fetches the latest version of CodeQL
inputs:
channel:
description: 'The CodeQL channel to use'
required: false
default: 'nightly'
runs:
using: composite
steps:
- name: Fetch CodeQL
shell: bash
env:
GITHUB_TOKEN: ${{ github.token }}
CHANNEL: ${{ inputs.channel }}
run: |
gh extension install github/gh-codeql
gh codeql set-channel "$CHANNEL"
gh codeql set-channel nightly
gh codeql version
printf "CODEQL_FETCHED_CODEQL_PATH=" >> "${GITHUB_ENV}"
gh codeql version --format=json | jq -r .unpackedLocation >> "${GITHUB_ENV}"
gh codeql version --format=json | jq -r .unpackedLocation >> "${GITHUB_PATH}"
env:
GITHUB_TOKEN: ${{ github.token }}

View File

@@ -1,26 +0,0 @@
name: Find Latest CodeQL Bundle
description: Finds the URL of the latest released version of the CodeQL bundle.
outputs:
url:
description: The download URL of the latest CodeQL bundle release
value: ${{ steps.find-latest.outputs.url }}
runs:
using: composite
steps:
- name: Find Latest Release
id: find-latest
shell: pwsh
run: |
$Latest = gh release list --repo github/codeql-action --exclude-drafts --limit 1000 |
ForEach-Object { $C = $_ -split "`t"; return @{ type = $C[1]; tag = $C[2]; } } |
Where-Object { $_.type -eq 'Latest' }
$Tag = $Latest.tag
if ($Tag -eq '') {
throw 'Failed to find latest bundle release.'
}
Write-Output "Latest bundle tag is '${Tag}'."
"url=https://github.com/github/codeql-action/releases/download/${Tag}/codeql-bundle-linux64.tar.gz" >> $env:GITHUB_OUTPUT
env:
GITHUB_TOKEN: ${{ github.token }}

View File

@@ -1,32 +0,0 @@
name: OS Version
description: Get OS version.
outputs:
version:
description: "OS version"
value: ${{ steps.version.outputs.version }}
runs:
using: composite
steps:
- if: runner.os == 'Linux'
shell: bash
run: |
. /etc/os-release
echo "VERSION=${NAME} ${VERSION}" >> $GITHUB_ENV
- if: runner.os == 'Windows'
shell: powershell
run: |
$objects = systeminfo.exe /FO CSV | ConvertFrom-Csv
"VERSION=$($objects.'OS Name') $($objects.'OS Version')" >> $env:GITHUB_ENV
- if: runner.os == 'macOS'
shell: bash
run: |
echo "VERSION=$(sw_vers -productName) $(sw_vers -productVersion)" >> $GITHUB_ENV
- name: Emit OS version
id: version
shell: bash
run: |
echo "$VERSION"
echo "version=${VERSION}" >> $GITHUB_OUTPUT

11
.github/labeler.yml vendored
View File

@@ -43,14 +43,3 @@ documentation:
"QL-for-QL":
- ql/**/*
- .github/workflows/ql-for-ql*
# Since these are all shared files that need to be synced, just pick _one_ copy of each.
"DataFlow Library":
- "java/ql/lib/semmle/code/java/dataflow/internal/DataFlowImpl.qll"
- "java/ql/lib/semmle/code/java/dataflow/internal/DataFlowImplCommon.qll"
- "java/ql/lib/semmle/code/java/dataflow/internal/tainttracking1/TaintTrackingImpl.qll"
- "java/ql/lib/semmle/code/java/dataflow/internal/DataFlowImplConsistency.qll"
- "java/ql/lib/semmle/code/java/dataflow/internal/FlowSummaryImpl.qll"
"ATM":
- javascript/ql/experimental/adaptivethreatmodeling/**/*

View File

@@ -1,102 +0,0 @@
name: "ATM - Check query suite"
env:
QUERY_PACK: javascript/ql/experimental/adaptivethreatmodeling/src
QUERY_SUITE: codeql-suites/javascript-atm-code-scanning.qls
on:
pull_request:
paths:
- ".github/workflows/atm-check-query-suite.yml"
- "javascript/ql/experimental/adaptivethreatmodeling/**"
workflow_dispatch:
jobs:
atm-check-query-suite:
runs-on: ubuntu-latest-xl
steps:
- uses: actions/checkout@v3
- name: Setup CodeQL
uses: ./.github/actions/fetch-codeql
with:
channel: release
- name: Cache compilation cache
id: query-cache
uses: ./.github/actions/cache-query-compilation
with:
key: atm-suite
- name: Install ATM model
run: |
set -exu
# Install dependencies of ATM query pack, i.e. the ATM model
codeql pack install "${QUERY_PACK}"
# Retrieve model checksum
model_checksum=$(codeql resolve extensions "${QUERY_PACK}/${QUERY_SUITE}" | jq -r '.models[0].checksum')
# Trust the model so that we can use it in the ATM boosted queries
mkdir -p "$HOME/.config/codeql"
echo "--insecurely-execute-ml-model-checksums ${model_checksum}" >> "$HOME/.config/codeql/config"
- name: Create test DB
run: |
DB_PATH="${RUNNER_TEMP}/db"
echo "DB_PATH=${DB_PATH}" >> "${GITHUB_ENV}"
codeql database create "${DB_PATH}" --source-root config/atm --language javascript
- name: Run ATM query suite
run: |
SARIF_PATH="${RUNNER_TEMP}/sarif.json"
echo "SARIF_PATH=${SARIF_PATH}" >> "${GITHUB_ENV}"
codeql database analyze \
--threads=0 \
--ram 50000 \
--format sarif-latest \
--output "${SARIF_PATH}" \
--sarif-group-rules-by-pack \
-vv \
--compilation-cache "${{ steps.query-cache.outputs.cache-dir }}" \
-- \
"${DB_PATH}" \
"${QUERY_PACK}/${QUERY_SUITE}"
- name: Upload SARIF
uses: actions/upload-artifact@v3
with:
name: javascript-ml-powered-queries.sarif
path: "${{ env.SARIF_PATH }}"
retention-days: 5
- name: Check results
run: |
# We should run at least the ML-powered queries in `expected_rules`.
expected_rules="js/ml-powered/nosql-injection js/ml-powered/path-injection js/ml-powered/sql-injection js/ml-powered/xss"
for rule in ${expected_rules}; do
found_rule=$(jq --arg rule "${rule}" '[.runs[0].tool.extensions[].rules | select(. != null) |
flatten | .[].id] | any(. == $rule)' "${SARIF_PATH}")
if [[ "${found_rule}" != "true" ]]; then
echo "Expected SARIF output to contain rule '${rule}', but found no such rule."
exit 1
else
echo "Found rule '${rule}'."
fi
done
# We should have at least one alert from an ML-powered query.
num_alerts=$(jq '[.runs[0].results[] |
select(.properties.score != null and (.rule.id | startswith("js/ml-powered/")))] | length' \
"${SARIF_PATH}")
if [[ "${num_alerts}" -eq 0 ]]; then
echo "Expected to find at least one alert from an ML-powered query but found ${num_alerts}."
exit 1
else
echo "Found ${num_alerts} alerts from ML-powered queries.";
fi

View File

@@ -1,12 +0,0 @@
name: ATM Model Integration Tests
on:
workflow_dispatch:
jobs:
hello-world:
runs-on: ubuntu-latest
steps:
- name: foo
run: echo "Hello world"

View File

@@ -26,9 +26,3 @@ jobs:
run: |
gh api 'repos/${{github.repository}}/pulls/${{github.event.number}}/files' --paginate --jq 'any(.[].filename ; test("/change-notes/.*[.]md$"))' |
grep true -c
- 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' or 'YYYY-MM-DD-{title}.md', where '{title}' is arbitrary text.
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$"))' |
grep true -c

View File

@@ -1,21 +0,0 @@
name: Check query IDs
on:
pull_request:
paths:
- "**/src/**/*.ql"
- misc/scripts/check-query-ids.py
- .github/workflows/check-query-ids.yml
branches:
- main
- "rc/*"
workflow_dispatch:
jobs:
check:
name: Check query IDs
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
- name: Check for duplicate query IDs
run: python3 misc/scripts/check-query-ids.py

View File

@@ -12,7 +12,7 @@ jobs:
runs-on: ubuntu-latest
steps:
- uses: actions/stale@v7
- uses: actions/stale@v6
with:
repo-token: ${{ secrets.GITHUB_TOKEN }}
stale-issue-message: 'This issue is stale because it has been open 14 days with no activity. Comment or remove the `Stale` label in order to avoid having this issue closed in 7 days.'

View File

@@ -1,37 +0,0 @@
name: "Compile all queries using the latest stable CodeQL CLI"
on:
push:
branches: # makes sure the cache gets populated - running on the branches people tend to merge into.
- main
- "rc/*"
- "codeql-cli-*"
pull_request:
jobs:
compile-queries:
runs-on: ubuntu-latest-xl
steps:
- uses: actions/checkout@v3
- name: Setup CodeQL
uses: ./.github/actions/fetch-codeql
with:
channel: 'release'
- name: Cache compilation cache
id: query-cache
uses: ./.github/actions/cache-query-compilation
with:
key: all-queries
- name: check formatting
run: find */ql -type f \( -name "*.qll" -o -name "*.ql" \) -print0 | xargs -0 codeql query format --check-only
- name: compile queries - check-only
# run with --check-only if running in a PR (github.sha != main)
if : ${{ github.event_name == 'pull_request' }}
shell: bash
run: codeql query compile -j0 */ql/{src,examples} --keep-going --warnings=error --check-only --compilation-cache "${{ steps.query-cache.outputs.cache-dir }}"
- name: compile queries - full
# do full compile if running on main - this populates the cache
if : ${{ github.event_name != 'pull_request' }}
shell: bash
run: codeql query compile -j0 */ql/{src,examples} --keep-going --warnings=error --compilation-cache "${{ steps.query-cache.outputs.cache-dir }}"

View File

@@ -1,86 +0,0 @@
name: "C#: Run QL Tests"
on:
push:
paths:
- "csharp/**"
- "shared/**"
- .github/actions/fetch-codeql/action.yml
- codeql-workspace.yml
branches:
- main
- "rc/*"
pull_request:
paths:
- "csharp/**"
- "shared/**"
- .github/workflows/csharp-qltest.yml
- .github/actions/fetch-codeql/action.yml
- codeql-workspace.yml
branches:
- main
- "rc/*"
defaults:
run:
working-directory: csharp
jobs:
qlupgrade:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
- uses: ./.github/actions/fetch-codeql
- name: Check DB upgrade scripts
run: |
echo >empty.trap
codeql dataset import -S ql/lib/upgrades/initial/semmlecode.csharp.dbscheme testdb empty.trap
codeql dataset upgrade testdb --additional-packs ql/lib
diff -q testdb/semmlecode.csharp.dbscheme ql/lib/semmlecode.csharp.dbscheme
- name: Check DB downgrade scripts
run: |
echo >empty.trap
rm -rf testdb; codeql dataset import -S ql/lib/semmlecode.csharp.dbscheme testdb empty.trap
codeql resolve upgrades --format=lines --allow-downgrades --additional-packs downgrades \
--dbscheme=ql/lib/semmlecode.csharp.dbscheme --target-dbscheme=downgrades/initial/semmlecode.csharp.dbscheme |
xargs codeql execute upgrades testdb
diff -q testdb/semmlecode.csharp.dbscheme downgrades/initial/semmlecode.csharp.dbscheme
qltest:
runs-on: ubuntu-latest-xl
strategy:
fail-fast: false
matrix:
slice: ["1/2", "2/2"]
steps:
- uses: actions/checkout@v3
- uses: ./.github/actions/fetch-codeql
- uses: ./csharp/actions/create-extractor-pack
- name: Cache compilation cache
id: query-cache
uses: ./.github/actions/cache-query-compilation
with:
key: csharp-qltest-${{ matrix.slice }}
- name: Run QL tests
run: |
CODEQL_PATH=$(gh codeql version --format=json | jq -r .unpackedLocation)
# The legacy ASP extractor is not in this repo, so take the one from the nightly build
mv "$CODEQL_PATH/csharp/tools/extractor-asp.jar" "${{ github.workspace }}/csharp/extractor-pack/tools"
# Safe guard against using the bundled extractor
rm -rf "$CODEQL_PATH/csharp"
codeql test run --threads=0 --ram 50000 --slice ${{ matrix.slice }} --search-path "${{ github.workspace }}/csharp/extractor-pack" --check-databases --check-undefined-labels --check-repeated-labels --check-redefined-labels --consistency-queries ql/consistency-queries ql/test --compilation-cache "${{ steps.query-cache.outputs.cache-dir }}"
env:
GITHUB_TOKEN: ${{ github.token }}
unit-tests:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
- name: Setup dotnet
uses: actions/setup-dotnet@v3
with:
dotnet-version: 6.0.202
- name: Extractor unit tests
run: |
dotnet test -p:RuntimeFrameworkVersion=6.0.4 "${{ github.workspace }}/csharp/extractor/Semmle.Util.Tests"
dotnet test -p:RuntimeFrameworkVersion=6.0.4 "${{ github.workspace }}/csharp/extractor/Semmle.Extraction.Tests"
dotnet test -p:RuntimeFrameworkVersion=6.0.4 "${{ github.workspace }}/csharp/autobuilder/Semmle.Autobuild.CSharp.Tests"
dotnet test -p:RuntimeFrameworkVersion=6.0.4 "${{ github.workspace }}/cpp/autobuilder/Semmle.Autobuild.Cpp.Tests"

View File

@@ -1,80 +0,0 @@
name: "Go: Run Tests - Other OS"
on:
pull_request:
paths:
- "go/**"
- "!go/ql/**" # don't run other-os if only ql/ files changed
- .github/workflows/go-tests-other-os.yml
- .github/actions/**
- codeql-workspace.yml
jobs:
test-mac:
name: Test MacOS
runs-on: macos-latest
steps:
- name: Set up Go 1.19
uses: actions/setup-go@v3
with:
go-version: 1.19
id: go
- name: Check out code
uses: actions/checkout@v2
- name: Set up CodeQL CLI
uses: ./.github/actions/fetch-codeql
- name: Enable problem matchers in repository
shell: bash
run: 'find .github/problem-matchers -name \*.json -exec echo "::add-matcher::{}" \;'
- name: Build
run: |
cd go
make
- name: Cache compilation cache
id: query-cache
uses: ./.github/actions/cache-query-compilation
with:
key: go-qltest
- name: Test
run: |
cd go
make test cache="${{ steps.query-cache.outputs.cache-dir }}"
test-win:
name: Test Windows
runs-on: windows-latest-xl
steps:
- name: Set up Go 1.19
uses: actions/setup-go@v3
with:
go-version: 1.19
id: go
- name: Check out code
uses: actions/checkout@v2
- name: Set up CodeQL CLI
uses: ./.github/actions/fetch-codeql
- name: Enable problem matchers in repository
shell: bash
run: 'find .github/problem-matchers -name \*.json -exec echo "::add-matcher::{}" \;'
- name: Build
run: |
cd go
make
- name: Cache compilation cache
id: query-cache
uses: ./.github/actions/cache-query-compilation
with:
key: go-qltest
- name: Test
run: |
cd go
make test cache="${{ steps.query-cache.outputs.cache-dir }}"

View File

@@ -1,24 +1,15 @@
name: "Go: Run Tests"
on:
push:
paths:
- "go/**"
- .github/workflows/go-tests.yml
- .github/actions/**
- codeql-workspace.yml
branches:
- main
- "rc/*"
pull_request:
paths:
- "go/**"
- .github/workflows/go-tests.yml
- .github/actions/**
- .github/actions/fetch-codeql/action.yml
- codeql-workspace.yml
jobs:
test-linux:
name: Test Linux (Ubuntu)
runs-on: ubuntu-latest-xl
runs-on: ubuntu-latest
steps:
- name: Set up Go 1.19
uses: actions/setup-go@v3
@@ -41,7 +32,7 @@ jobs:
cd go
make
- name: Check that all Go code is autoformatted
- name: Check that all QL and Go code is autoformatted
run: |
cd go
make check-formatting
@@ -57,13 +48,67 @@ jobs:
name: qhelp-markdown
path: go/qhelp-out/**/*.md
- name: Cache compilation cache
id: query-cache
uses: ./.github/actions/cache-query-compilation
- name: Test
run: |
cd go
make test
test-mac:
name: Test MacOS
runs-on: macos-latest
steps:
- name: Set up Go 1.19
uses: actions/setup-go@v3
with:
key: go-qltest
go-version: 1.19
id: go
- name: Check out code
uses: actions/checkout@v2
- name: Set up CodeQL CLI
uses: ./.github/actions/fetch-codeql
- name: Enable problem matchers in repository
shell: bash
run: 'find .github/problem-matchers -name \*.json -exec echo "::add-matcher::{}" \;'
- name: Build
run: |
cd go
make
- name: Test
run: |
cd go
make test cache="${{ steps.query-cache.outputs.cache-dir }}"
make test
test-win:
name: Test Windows
runs-on: windows-2019
steps:
- name: Set up Go 1.19
uses: actions/setup-go@v3
with:
go-version: 1.19
id: go
- name: Check out code
uses: actions/checkout@v2
- name: Set up CodeQL CLI
uses: ./.github/actions/fetch-codeql
- name: Enable problem matchers in repository
shell: bash
run: 'find .github/problem-matchers -name \*.json -exec echo "::add-matcher::{}" \;'
- name: Build
run: |
cd go
make
- name: Test
run: |
cd go
make test

View File

@@ -23,9 +23,22 @@ defaults:
working-directory: javascript/ql/experimental/adaptivethreatmodeling
jobs:
qltest:
name: Test QL
runs-on: ubuntu-latest-xl
qlformat:
name: Check QL formatting
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
- uses: ./.github/actions/fetch-codeql
- name: Check QL formatting
run: |
find . "(" -name "*.ql" -or -name "*.qll" ")" -print0 | \
xargs -0 codeql query format --check-only
qlcompile:
name: Check QL compilation
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
@@ -33,33 +46,36 @@ jobs:
- name: Install pack dependencies
run: |
for pack in modelbuilding src test; do
for pack in modelbuilding src; do
codeql pack install --mode verify -- "${pack}"
done
- name: Cache compilation cache
id: query-cache
uses: ./.github/actions/cache-query-compilation
with:
key: js-ml-test
- name: Check QL compilation
run: |
codeql query compile \
--check-only \
--ram 50000 \
--ram 5120 \
--additional-packs "${{ github.workspace }}" \
--threads=0 \
--compilation-cache "${{ steps.query-cache.outputs.cache-dir }}" \
-- \
lib modelbuilding src
qltest:
name: Run QL tests
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
- uses: ./.github/actions/fetch-codeql
- name: Install pack dependencies
run: codeql pack install -- test
- name: Run QL tests
run: |
codeql test run \
--threads=0 \
--ram 50000 \
--ram 5120 \
--additional-packs "${{ github.workspace }}" \
--compilation-cache "${{ steps.query-cache.outputs.cache-dir }}" \
-- \
test
test

View File

@@ -11,7 +11,7 @@ on:
branches:
- main
paths:
- "java/ql/src/utils/modelgenerator/**/*.*"
- "java/ql/src/utils/model-generator/**/*.*"
- ".github/workflows/mad_modelDiff.yml"
permissions:
@@ -40,12 +40,12 @@ jobs:
- name: Download database
env:
SLUG: ${{ matrix.slug }}
GH_TOKEN: ${{ github.token }}
run: |
set -x
mkdir lib-dbs
SHORTNAME=${SLUG//[^a-zA-Z0-9_]/}
gh api -H "Accept: application/zip" "/repos/${SLUG}/code-scanning/codeql/databases/java" > "$SHORTNAME.zip"
projectId=`curl -s https://lgtm.com/api/v1.0/projects/g/${SLUG} | jq .id`
curl -L "https://lgtm.com/api/v1.0/snapshots/$projectId/java" -o "$SHORTNAME.zip"
unzip -q -d "${SHORTNAME}-db" "${SHORTNAME}.zip"
mkdir "lib-dbs/$SHORTNAME/"
mv "${SHORTNAME}-db/"$(ls -1 "${SHORTNAME}"-db)/* "lib-dbs/${SHORTNAME}/"
@@ -61,8 +61,8 @@ jobs:
DATABASE=$2
cd codeql-$QL_VARIANT
SHORTNAME=`basename $DATABASE`
python java/ql/src/utils/modelgenerator/GenerateFlowModel.py --with-summaries --with-sinks $DATABASE ${SHORTNAME}.temp.model.yml
mv java/ql/lib/ext/generated/${SHORTNAME}.temp.model.yml $MODELS/${SHORTNAME}Generated_${QL_VARIANT}.model.yml
python java/ql/src/utils/model-generator/GenerateFlowModel.py --with-summaries --with-sinks $DATABASE $MODELS/${SHORTNAME}.qll
mv $MODELS/${SHORTNAME}.qll $MODELS/${SHORTNAME}Generated_${QL_VARIANT}.qll
cd ..
}
@@ -85,21 +85,19 @@ jobs:
set -x
MODELS=`pwd`/tmp-models
ls -1 tmp-models/
for m in $MODELS/*_main.model.yml ; do
for m in $MODELS/*_main.qll ; do
t="${m/main/"pr"}"
basename=`basename $m`
name="diff_${basename/_main.model.yml/""}"
name="diff_${basename/_main.qll/""}"
(diff -w -u $m $t | diff2html -i stdin -F $MODELS/$name.html) || true
done
- uses: actions/upload-artifact@v3
with:
name: models
path: tmp-models/*.model.yml
path: tmp-models/*.qll
retention-days: 20
- uses: actions/upload-artifact@v3
with:
name: diffs
path: tmp-models/*.html
# An html file is only produced if the generated models differ.
if-no-files-found: ignore
retention-days: 20

View File

@@ -50,10 +50,10 @@ jobs:
SLUG: ${{ matrix.slug }}
run: |
SHORTNAME=${SLUG//[^a-zA-Z0-9_]/}
java/ql/src/utils/modelgenerator/RegenerateModels.py "${SLUG}" dbs/${SHORTNAME}
java/ql/src/utils/model-generator/RegenerateModels.py "${SLUG}" dbs/${SHORTNAME}
- name: Stage changes
run: |
find java -name "*.model.yml" -print0 | xargs -0 git add
find java -name "*.qll" -print0 | xargs -0 git add
git status
git diff --cached > models.patch
- uses: actions/upload-artifact@v3

View File

@@ -27,7 +27,7 @@ on:
- main
- "rc/*"
paths:
- "**/*.qhelp"
- "ruby/**/*.qhelp"
jobs:
qhelp:
@@ -52,7 +52,7 @@ jobs:
id: changes
run: |
(git diff -z --name-only --diff-filter=ACMRT HEAD~1 HEAD | grep -z '.qhelp$' | grep -z -v '.inc.qhelp';
git diff -z --name-only --diff-filter=ACMRT HEAD~1 HEAD | grep -z '.inc.qhelp$' | xargs --null -rn1 basename -z | xargs --null -rn1 git grep -z -l) |
git diff -z --name-only --diff-filter=ACMRT HEAD~1 HEAD | grep -z '.inc.qhelp$' | xargs --null -rn1 basename | xargs --null -rn1 git grep -z -l) |
grep -z '.qhelp$' | grep -z -v '^-' | sort -z -u > "${RUNNER_TEMP}/paths.txt"
- name: QHelp preview

View File

@@ -22,30 +22,24 @@ jobs:
steps:
### Build the queries ###
- uses: actions/checkout@v3
- name: Find latest bundle
id: find-latest-bundle
uses: ./.github/actions/find-latest-bundle
- name: Find codeql
id: find-codeql
uses: github/codeql-action/init@45955cb1830b640e2c1603ad72ad542a49d47b96
uses: github/codeql-action/init@71a8b35ff4c80fcfcd05bc1cd932fe3c08f943ca
with:
languages: javascript # does not matter
tools: ${{ steps.find-latest-bundle.outputs.url }}
- name: Get CodeQL version
id: get-codeql-version
run: |
echo "version=$("${CODEQL}" --version | head -n 1 | rev | cut -d " " -f 1 | rev)" >> $GITHUB_OUTPUT
echo "::set-output name=version::$("${CODEQL}" --version | head -n 1 | rev | cut -d " " -f 1 | rev)"
shell: bash
env:
CODEQL: ${{ steps.find-codeql.outputs.codeql-path }}
- uses: ./.github/actions/os-version
id: os_version
- name: Cache entire pack
id: cache-pack
uses: actions/cache@v3
with:
path: ${{ runner.temp }}/pack
key: ${{ runner.os }}-${{ steps.os_version.outputs.version }}-pack-${{ hashFiles('ql/**/Cargo.lock') }}-${{ hashFiles('ql/**/*.rs') }}-${{ hashFiles('ql/**/*.ql*') }}-${{ hashFiles('ql/**/qlpack.yml') }}-${{ hashFiles('ql/ql/src/ql.dbscheme*') }}-${{ steps.get-codeql-version.outputs.version }}--${{ hashFiles('.github/workflows/ql-for-ql-build.yml') }}
key: ${{ runner.os }}-pack-${{ hashFiles('ql/**/Cargo.lock') }}-${{ hashFiles('ql/**/*.rs') }}-${{ hashFiles('ql/**/*.ql*') }}-${{ hashFiles('ql/**/qlpack.yml') }}-${{ hashFiles('ql/ql/src/ql.dbscheme*') }}-${{ steps.get-codeql-version.outputs.version }}--${{ hashFiles('.github/workflows/ql-for-ql-build.yml') }}
- name: Cache queries
if: steps.cache-pack.outputs.cache-hit != 'true'
id: cache-queries
@@ -79,7 +73,7 @@ jobs:
ql/target/release/ql-autobuilder.exe
ql/target/release/ql-extractor
ql/target/release/ql-extractor.exe
key: ${{ runner.os }}-${{ steps.os_version.outputs.version }}-extractor-${{ hashFiles('ql/**/Cargo.lock') }}-${{ hashFiles('ql/**/*.rs') }}
key: ${{ runner.os }}-extractor-${{ hashFiles('ql/**/Cargo.lock') }}-${{ hashFiles('ql/**/*.rs') }}
- name: Cache cargo
if: steps.cache-extractor.outputs.cache-hit != 'true' && steps.cache-pack.outputs.cache-hit != 'true'
uses: actions/cache@v3
@@ -88,7 +82,7 @@ jobs:
~/.cargo/registry
~/.cargo/git
ql/target
key: ${{ runner.os }}-${{ steps.os_version.outputs.version }}-rust-cargo-${{ hashFiles('ql/**/Cargo.lock') }}
key: ${{ runner.os }}-rust-cargo-${{ hashFiles('ql/**/Cargo.lock') }}
- name: Check formatting
if: steps.cache-extractor.outputs.cache-hit != 'true' && steps.cache-pack.outputs.cache-hit != 'true'
run: cd ql; cargo fmt --all -- --check
@@ -139,20 +133,19 @@ jobs:
env:
CONF: ./ql-for-ql-config.yml
- name: Initialize CodeQL
uses: github/codeql-action/init@45955cb1830b640e2c1603ad72ad542a49d47b96
uses: github/codeql-action/init@71a8b35ff4c80fcfcd05bc1cd932fe3c08f943ca
with:
languages: ql
db-location: ${{ runner.temp }}/db
config-file: ./ql-for-ql-config.yml
tools: ${{ steps.find-latest-bundle.outputs.url }}
- name: Move pack queries
- name: Move pack cache
run: |
cp -r ${PACK}/queries ql/ql/src
cp -r ${PACK}/.cache ql/ql/src/.cache
env:
PACK: ${{ runner.temp }}/pack
- name: Perform CodeQL Analysis
uses: github/codeql-action/analyze@45955cb1830b640e2c1603ad72ad542a49d47b96
uses: github/codeql-action/analyze@71a8b35ff4c80fcfcd05bc1cd932fe3c08f943ca
with:
category: "ql-for-ql"
- name: Copy sarif file to CWD
@@ -174,4 +167,4 @@ jobs:
with:
name: ql-for-ql-langs
path: split-sarif
retention-days: 1
retention-days: 1

View File

@@ -25,18 +25,16 @@ jobs:
- name: Find codeql
id: find-codeql
uses: github/codeql-action/init@45955cb1830b640e2c1603ad72ad542a49d47b96
uses: github/codeql-action/init@71a8b35ff4c80fcfcd05bc1cd932fe3c08f943ca
with:
languages: javascript # does not matter
- uses: ./.github/actions/os-version
id: os_version
- uses: actions/cache@v3
with:
path: |
~/.cargo/registry
~/.cargo/git
ql/target
key: ${{ runner.os }}-${{ steps.os_version.outputs.version }}-qltest-cargo-${{ hashFiles('ql/**/Cargo.lock') }}
key: ${{ runner.os }}-qltest-cargo-${{ hashFiles('ql/**/Cargo.lock') }}
- name: Build Extractor
run: cd ql; env "PATH=$PATH:`dirname ${CODEQL}`" ./scripts/create-extractor-pack.sh
env:

View File

@@ -22,84 +22,33 @@ jobs:
- uses: actions/checkout@v3
- name: Find codeql
id: find-codeql
uses: github/codeql-action/init@45955cb1830b640e2c1603ad72ad542a49d47b96
uses: github/codeql-action/init@71a8b35ff4c80fcfcd05bc1cd932fe3c08f943ca
with:
languages: javascript # does not matter
- uses: ./.github/actions/os-version
id: os_version
- uses: actions/cache@v3
with:
path: |
~/.cargo/registry
~/.cargo/git
ql/target
key: ${{ runner.os }}-${{ steps.os_version.outputs.version }}-qltest-cargo-${{ hashFiles('ql/rust-toolchain.toml', 'ql/**/Cargo.lock') }}
key: ${{ runner.os }}-qltest-cargo-${{ hashFiles('ql/**/Cargo.lock') }}
- name: Build extractor
run: |
cd ql;
codeqlpath=$(dirname ${{ steps.find-codeql.outputs.codeql-path }});
env "PATH=$PATH:$codeqlpath" ./scripts/create-extractor-pack.sh
- name: Cache compilation cache
id: query-cache
uses: ./.github/actions/cache-query-compilation
with:
key: ql-for-ql-tests
- name: Run QL tests
run: |
"${CODEQL}" test run --check-databases --check-unused-labels --check-repeated-labels --check-redefined-labels --check-use-before-definition --search-path "${{ github.workspace }}/ql/extractor-pack" --consistency-queries ql/ql/consistency-queries --compilation-cache "${{ steps.query-cache.outputs.cache-dir }}" ql/ql/test
"${CODEQL}" test run --check-databases --check-unused-labels --check-repeated-labels --check-redefined-labels --check-use-before-definition --search-path "${{ github.workspace }}/ql/extractor-pack" --consistency-queries ql/ql/consistency-queries ql/ql/test
env:
CODEQL: ${{ steps.find-codeql.outputs.codeql-path }}
other-os:
strategy:
matrix:
os: [macos-latest, windows-latest]
needs: [qltest]
runs-on: ${{ matrix.os }}
steps:
- uses: actions/checkout@v3
- name: Install GNU tar
if: runner.os == 'macOS'
- name: Check QL formatting
run: |
brew install gnu-tar
echo "/usr/local/opt/gnu-tar/libexec/gnubin" >> $GITHUB_PATH
- name: Find codeql
id: find-codeql
uses: github/codeql-action/init@77a8d2d10c0b403a8b4aadbd223dc489ecd22683
with:
languages: javascript # does not matter
- uses: ./.github/actions/os-version
id: os_version
- uses: actions/cache@v3
with:
path: |
~/.cargo/registry
~/.cargo/git
ql/target
key: ${{ runner.os }}-${{ steps.os_version.outputs.version }}-qltest-cargo-${{ hashFiles('ql/rust-toolchain.toml', 'ql/**/Cargo.lock') }}
- name: Build extractor
if: runner.os != 'Windows'
run: |
cd ql;
codeqlpath=$(dirname ${{ steps.find-codeql.outputs.codeql-path }});
env "PATH=$PATH:$codeqlpath" ./scripts/create-extractor-pack.sh
- name: Build extractor (Windows)
if: runner.os == 'Windows'
shell: pwsh
run: |
cd ql;
$Env:PATH += ";$(dirname ${{ steps.find-codeql.outputs.codeql-path }})"
pwsh ./scripts/create-extractor-pack.ps1
- name: Run a single QL tests - Unix
if: runner.os != 'Windows'
run: |
"${CODEQL}" test run --check-databases --search-path "${{ github.workspace }}/ql/extractor-pack" ql/ql/test/queries/style/DeadCode/DeadCode.qlref
find ql/ql/src "(" -name "*.ql" -or -name "*.qll" ")" -print0 | xargs -0 "${CODEQL}" query format --check-only
env:
CODEQL: ${{ steps.find-codeql.outputs.codeql-path }}
- name: Run a single QL tests - Windows
if: runner.os == 'Windows'
shell: pwsh
- name: Check QL compilation
run: |
$Env:PATH += ";$(dirname ${{ steps.find-codeql.outputs.codeql-path }})"
codeql test run --check-databases --search-path "${{ github.workspace }}/ql/extractor-pack" ql/ql/test/queries/style/DeadCode/DeadCode.qlref
"${CODEQL}" query compile --check-only --threads=4 --warnings=error --search-path "${{ github.workspace }}/ql/extractor-pack" "ql/ql/src" "ql/ql/examples"
env:
CODEQL: ${{ steps.find-codeql.outputs.codeql-path }}

View File

@@ -48,41 +48,23 @@ jobs:
run: |
brew install gnu-tar
echo "/usr/local/opt/gnu-tar/libexec/gnubin" >> $GITHUB_PATH
- uses: ./.github/actions/os-version
id: os_version
- name: Cache entire extractor
uses: actions/cache@v3
id: cache-extractor
with:
path: |
ruby/target/release/ruby-autobuilder
ruby/target/release/ruby-autobuilder.exe
ruby/target/release/ruby-extractor
ruby/target/release/ruby-extractor.exe
ruby/ql/lib/codeql/ruby/ast/internal/TreeSitter.qll
key: ${{ runner.os }}-${{ steps.os_version.outputs.version }}-ruby-extractor-${{ hashFiles('ruby/rust-toolchain.toml', 'ruby/**/Cargo.lock') }}--${{ hashFiles('ruby/**/*.rs') }}
- uses: actions/cache@v3
if: steps.cache-extractor.outputs.cache-hit != 'true'
with:
path: |
~/.cargo/registry
~/.cargo/git
ruby/target
key: ${{ runner.os }}-${{ steps.os_version.outputs.version }}-ruby-rust-cargo-${{ hashFiles('ruby/rust-toolchain.toml', 'ruby/**/Cargo.lock') }}
key: ${{ runner.os }}-ruby-rust-cargo-${{ hashFiles('ruby/rust-toolchain.toml', 'ruby/**/Cargo.lock') }}
- name: Check formatting
if: steps.cache-extractor.outputs.cache-hit != 'true'
run: cargo fmt --all -- --check
- name: Build
if: steps.cache-extractor.outputs.cache-hit != 'true'
run: cargo build --verbose
- name: Run tests
if: steps.cache-extractor.outputs.cache-hit != 'true'
run: cargo test --verbose
- name: Release build
if: steps.cache-extractor.outputs.cache-hit != 'true'
run: cargo build --release
- name: Generate dbscheme
if: ${{ matrix.os == 'ubuntu-latest' && steps.cache-extractor.outputs.cache-hit != 'true'}}
if: ${{ matrix.os == 'ubuntu-latest' }}
run: target/release/ruby-generator --dbscheme ql/lib/ruby.dbscheme --library ql/lib/codeql/ruby/ast/internal/TreeSitter.qll
- uses: actions/upload-artifact@v3
if: ${{ matrix.os == 'ubuntu-latest' }}
@@ -104,25 +86,19 @@ jobs:
ruby/target/release/ruby-extractor.exe
retention-days: 1
compile-queries:
runs-on: ubuntu-latest-xl
runs-on: ubuntu-latest
env:
CODEQL_THREADS: 4 # TODO: remove this once it's set by the CLI
steps:
- uses: actions/checkout@v3
- name: Fetch CodeQL
uses: ./.github/actions/fetch-codeql
- name: Cache compilation cache
id: query-cache
uses: ./.github/actions/cache-query-compilation
with:
key: ruby-build
- name: Build Query Pack
run: |
rm -rf target/packs
codeql pack create ../misc/suite-helpers --output target/packs
codeql pack create ../shared/regex --output target/packs
codeql pack create ../shared/ssa --output target/packs
codeql pack create ../shared/tutorial --output target/packs
codeql pack create ql/lib --output target/packs
codeql pack create -j0 ql/src --output target/packs --compilation-cache "${{ steps.query-cache.outputs.cache-dir }}"
codeql pack install ql/src
codeql pack create ql/src --output target/packs
PACK_FOLDER=$(readlink -f target/packs/codeql/ruby-queries/*)
codeql generate query-help --format=sarifv2.1.0 --output="${PACK_FOLDER}/rules.sarif" ql/src
(cd ql/src; find queries \( -name '*.qhelp' -o -name '*.rb' -o -name '*.erb' \) -exec bash -c 'mkdir -p "'"${PACK_FOLDER}"'/$(dirname "{}")"' \; -exec cp "{}" "${PACK_FOLDER}/{}" \;)
@@ -205,6 +181,11 @@ jobs:
- name: Fetch CodeQL
uses: ./.github/actions/fetch-codeql
- uses: actions/checkout@v3
with:
repository: Shopify/example-ruby-app
ref: 67a0decc5eb550f3a9228eda53925c3afd40dfe9
- name: Download Ruby bundle
uses: actions/download-artifact@v3
with:
@@ -213,15 +194,26 @@ jobs:
- name: Unzip Ruby bundle
shell: bash
run: unzip -q -d "${{ runner.temp }}/ruby-bundle" "${{ runner.temp }}/codeql-ruby-bundle.zip"
- name: Prepare test files
shell: bash
run: |
echo "import codeql.ruby.AST select count(File f)" > "test.ql"
echo "| 4 |" > "test.expected"
echo 'name: sample-tests
version: 0.0.0
dependencies:
codeql/ruby-all: 0.0.1
extractor: ruby
tests: .
' > qlpack.yml
- name: Run QL test
shell: bash
run: |
codeql test run --search-path "${{ runner.temp }}/ruby-bundle" --additional-packs "${{ runner.temp }}/ruby-bundle" ruby/ql/test/library-tests/ast/constants/
codeql test run --search-path "${{ runner.temp }}/ruby-bundle" --additional-packs "${{ runner.temp }}/ruby-bundle" .
- name: Create database
shell: bash
run: |
codeql database create --search-path "${{ runner.temp }}/ruby-bundle" --language ruby --source-root ruby/ql/test/library-tests/ast/constants/ ../database
codeql database create --search-path "${{ runner.temp }}/ruby-bundle" --language ruby --source-root . ../database
- name: Analyze database
shell: bash
run: |

View File

@@ -4,7 +4,7 @@ on:
push:
paths:
- "ruby/**"
- .github/workflows/ruby-build.yml
- .github/workflows/ruby-qltest.yml
- .github/actions/fetch-codeql/action.yml
- codeql-workspace.yml
branches:
@@ -28,6 +28,23 @@ defaults:
working-directory: ruby
jobs:
qlformat:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
- uses: ./.github/actions/fetch-codeql
- name: Check QL formatting
run: find ql "(" -name "*.ql" -or -name "*.qll" ")" -print0 | xargs -0 codeql query format --check-only
qlcompile:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
- uses: ./.github/actions/fetch-codeql
- name: Check QL compilation
run: |
codeql query compile --check-only --threads=0 --ram 5000 --warnings=error "ql/src" "ql/examples"
env:
GITHUB_TOKEN: ${{ github.token }}
qlupgrade:
runs-on: ubuntu-latest
steps:
@@ -48,20 +65,17 @@ jobs:
xargs codeql execute upgrades testdb
diff -q testdb/ruby.dbscheme downgrades/initial/ruby.dbscheme
qltest:
runs-on: ubuntu-latest-xl
runs-on: ubuntu-latest
strategy:
fail-fast: false
matrix:
slice: ["1/2", "2/2"]
steps:
- uses: actions/checkout@v3
- uses: ./.github/actions/fetch-codeql
- uses: ./ruby/actions/create-extractor-pack
- name: Cache compilation cache
id: query-cache
uses: ./.github/actions/cache-query-compilation
with:
key: ruby-qltest
- name: Run QL tests
run: |
codeql test run --threads=0 --ram 50000 --search-path "${{ github.workspace }}/ruby/extractor-pack" --check-databases --check-undefined-labels --check-unused-labels --check-repeated-labels --check-redefined-labels --check-use-before-definition --consistency-queries ql/consistency-queries ql/test --compilation-cache "${{ steps.query-cache.outputs.cache-dir }}"
codeql test run --threads=0 --ram 5000 --slice ${{ matrix.slice }} --search-path "${{ github.workspace }}/ruby/extractor-pack" --check-databases --check-unused-labels --check-repeated-labels --check-redefined-labels --check-use-before-definition --consistency-queries ql/consistency-queries ql/test
env:
GITHUB_TOKEN: ${{ github.token }}

39
.github/workflows/swift-codegen.yml vendored Normal file
View File

@@ -0,0 +1,39 @@
name: "Swift: Check code generation"
on:
pull_request:
paths:
- "swift/**"
- "misc/bazel/**"
- "*.bazel*"
- .github/workflows/swift-codegen.yml
- .github/actions/fetch-codeql/action.yml
branches:
- main
jobs:
codegen:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
- uses: ./.github/actions/fetch-codeql
- uses: bazelbuild/setup-bazelisk@v2
- uses: actions/setup-python@v3
- uses: pre-commit/action@v3.0.0
name: Check that python code is properly formatted
with:
extra_args: autopep8 --all-files
- name: Run unit tests
run: |
bazel test //swift/codegen/test --test_output=errors
- uses: pre-commit/action@v3.0.0
name: Check that QL generated code was checked in
with:
extra_args: swift-codegen --all-files
- name: Generate C++ files
run: |
bazel run //swift/codegen:codegen -- --generate=trap,cpp --cpp-output=$PWD/swift-generated-cpp-files
- uses: actions/upload-artifact@v3
with:
name: swift-generated-cpp-files
path: swift-generated-cpp-files/**

View File

@@ -0,0 +1,45 @@
name: "Swift: Run Integration Tests"
on:
pull_request:
paths:
- "swift/**"
- "misc/bazel/**"
- "*.bazel*"
- .github/workflows/swift-integration-tests.yml
- .github/actions/fetch-codeql/action.yml
- codeql-workspace.yml
branches:
- main
defaults:
run:
working-directory: swift
jobs:
integration-tests:
runs-on: ${{ matrix.os }}
strategy:
fail-fast: false
matrix:
os:
- ubuntu-20.04
# - macos-latest TODO
steps:
- uses: actions/checkout@v3
- uses: ./.github/actions/fetch-codeql
- uses: bazelbuild/setup-bazelisk@v2
- uses: actions/setup-python@v3
- name: Build Swift extractor
run: |
bazel run //swift:create-extractor-pack
- name: Get Swift version
id: get_swift_version
run: |
VERSION=$(bazel run //swift/extractor -- --version | sed -ne 's/.*version \(\S*\).*/\1/p')
echo "::set-output name=version::$VERSION"
- uses: swift-actions/setup-swift@v1
with:
swift-version: "${{steps.get_swift_version.outputs.version}}"
- name: Run integration tests
run: |
python integration-tests/runner.py

43
.github/workflows/swift-qltest.yml vendored Normal file
View File

@@ -0,0 +1,43 @@
name: "Swift: Run QL Tests"
on:
pull_request:
paths:
- "swift/**"
- "misc/bazel/**"
- "*.bazel*"
- .github/workflows/swift-qltest.yml
- .github/actions/fetch-codeql/action.yml
- codeql-workspace.yml
branches:
- main
defaults:
run:
working-directory: swift
jobs:
qlformat:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
- uses: ./.github/actions/fetch-codeql
- name: Check QL formatting
run: find ql "(" -name "*.ql" -or -name "*.qll" ")" -print0 | xargs -0 codeql query format --check-only
qltest:
runs-on: ${{ matrix.os }}
strategy:
fail-fast: false
matrix:
os : [ubuntu-20.04, macos-latest]
steps:
- uses: actions/checkout@v3
- uses: ./.github/actions/fetch-codeql
- uses: bazelbuild/setup-bazelisk@v2
- name: Build Swift extractor
run: |
bazel run //swift:create-extractor-pack
- name: Run QL tests
run: |
codeql test run --threads=0 --ram 5000 --search-path "${{ github.workspace }}/swift/extractor-pack" --check-databases --check-unused-labels --check-repeated-labels --check-redefined-labels --check-use-before-definition ql/test
env:
GITHUB_TOKEN: ${{ github.token }}

View File

@@ -1,103 +0,0 @@
name: "Swift"
on:
pull_request:
paths:
- "swift/**"
- "misc/bazel/**"
- "*.bazel*"
- .github/workflows/swift.yml
- .github/actions/**
- codeql-workspace.yml
- .pre-commit-config.yaml
- "!**/*.md"
- "!**/*.qhelp"
branches:
- main
- rc/*
push:
paths:
- "swift/**"
- "misc/bazel/**"
- "*.bazel*"
- .github/workflows/swift.yml
- .github/actions/**
- codeql-workspace.yml
- "!**/*.md"
- "!**/*.qhelp"
branches:
- main
- rc/*
jobs:
# not using a matrix as you cannot depend on a specific job in a matrix, and we want to start linux checks
# without waiting for the macOS build
build-and-test-macos:
runs-on: macos-12-xl
steps:
- uses: actions/checkout@v3
- uses: ./swift/actions/build-and-test
build-and-test-linux:
runs-on: ubuntu-latest-xl
steps:
- uses: actions/checkout@v3
- uses: ./swift/actions/build-and-test
qltests-linux:
needs: build-and-test-linux
runs-on: ubuntu-latest-xl
steps:
- uses: actions/checkout@v3
- uses: ./swift/actions/run-ql-tests
qltests-macos:
if : ${{ github.event_name == 'pull_request' }}
needs: build-and-test-macos
runs-on: macos-12-xl
steps:
- uses: actions/checkout@v3
- uses: ./swift/actions/run-ql-tests
integration-tests-linux:
needs: build-and-test-linux
runs-on: ubuntu-latest-xl
steps:
- uses: actions/checkout@v3
- uses: ./swift/actions/run-integration-tests
integration-tests-macos:
if : ${{ github.event_name == 'pull_request' }}
needs: build-and-test-macos
runs-on: macos-12-xl
timeout-minutes: 60
steps:
- uses: actions/checkout@v3
- uses: ./swift/actions/run-integration-tests
codegen:
if : ${{ github.event_name == 'pull_request' }}
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
- uses: bazelbuild/setup-bazelisk@v2
- uses: actions/setup-python@v4
with:
python-version-file: 'swift/.python-version'
- uses: pre-commit/action@v3.0.0
name: Check that python code is properly formatted
with:
extra_args: autopep8 --all-files
- uses: ./.github/actions/fetch-codeql
- uses: pre-commit/action@v3.0.0
name: Check that QL generated code was checked in
with:
extra_args: swift-codegen --all-files
- name: Generate C++ files
run: |
bazel run //swift/codegen:codegen -- --generate=trap,cpp --cpp-output=$PWD/generated-cpp-files
- uses: actions/upload-artifact@v3
with:
name: swift-generated-cpp-files
path: generated-cpp-files/**
database-upgrade-scripts:
if : ${{ github.event_name == 'pull_request' }}
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
- uses: ./.github/actions/fetch-codeql
- uses: ./swift/actions/database-upgrade-scripts

2
.gitignore vendored
View File

@@ -27,6 +27,8 @@
# It's useful (though not required) to be able to unpack codeql in the ql checkout itself
/codeql/
csharp/extractor/Semmle.Extraction.CSharp.Driver/Properties/launchSettings.json
# Avoid committing cached package components
.codeql

View File

@@ -19,7 +19,7 @@ repos:
rev: v1.6.0
hooks:
- id: autopep8
files: ^swift/.*\.py
files: ^swift/codegen/.*\.py
- repo: local
hooks:
@@ -31,7 +31,7 @@ repos:
- id: sync-files
name: Fix files required to be identical
files: \.(qll?|qhelp|swift)$|^config/identical-files\.json$
files: \.(qll?|qhelp|swift)$
language: system
entry: python3 config/sync-files.py --latest
pass_filenames: false
@@ -44,7 +44,7 @@ repos:
- id: swift-codegen
name: Run Swift checked in code generation
files: ^swift/(schema.py$|codegen/|.*/generated/|ql/lib/(swift\.dbscheme$|codeql/swift/elements)|ql/\.generated.list)
files: ^swift/(codegen/|.*/generated/|ql/lib/(swift\.dbscheme$|codeql/swift/elements))
language: system
entry: bazel run //swift/codegen -- --quiet
pass_filenames: false

View File

@@ -1,5 +1,3 @@
{
"omnisharp.autoStart": false,
"cmake.sourceDirectory": "${workspaceFolder}/swift",
"cmake.buildDirectory": "${workspaceFolder}/bazel-cmake-build"
"omnisharp.autoStart": false
}

View File

@@ -5,17 +5,24 @@
/javascript/ @github/codeql-javascript
/python/ @github/codeql-python
/ruby/ @github/codeql-ruby
/swift/ @github/codeql-swift
/swift/ @github/codeql-c
/java/kotlin-extractor/ @github/codeql-kotlin
/java/kotlin-explorer/ @github/codeql-kotlin
# ML-powered queries
/javascript/ql/experimental/adaptivethreatmodeling/ @github/codeql-ml-powered-queries-reviewers
# Notify members of codeql-go about PRs to the shared data-flow library files
/java/ql/src/semmle/code/java/dataflow/internal/DataFlowImpl.qll @github/codeql-java @github/codeql-go
/java/ql/src/semmle/code/java/dataflow/internal/DataFlowImpl2.qll @github/codeql-java @github/codeql-go
/java/ql/src/semmle/code/java/dataflow/internal/DataFlowImplCommon.qll @github/codeql-java @github/codeql-go
/java/ql/src/semmle/code/java/dataflow/internal/tainttracking1/TaintTrackingImpl.qll @github/codeql-java @github/codeql-go
/java/ql/src/semmle/code/java/dataflow/internal/tainttracking2/TaintTrackingImpl.qll @github/codeql-java @github/codeql-go
# CodeQL tools and associated docs
/docs/codeql/codeql-cli/ @github/codeql-cli-reviewers
/docs/codeql/codeql-for-visual-studio-code/ @github/codeql-vscode-reviewers
/docs/codeql/ql-language-reference/ @github/codeql-frontend-reviewers
/docs/codeql-cli/ @github/codeql-cli-reviewers
/docs/codeql-for-visual-studio-code/ @github/codeql-vscode-reviewers
/docs/ql-language-reference/ @github/codeql-frontend-reviewers
/docs/query-*-style-guide.md @github/codeql-analysis-reviewers
# QL for QL reviewers
@@ -33,9 +40,8 @@ WORKSPACE.bazel @github/codeql-ci-reviewers
# Workflows
/.github/workflows/ @github/codeql-ci-reviewers
/.github/workflows/atm-* @github/codeql-ml-powered-queries-reviewers
/.github/workflows/go-* @github/codeql-go
/.github/workflows/js-ml-tests.yml @github/codeql-ml-powered-queries-reviewers
/.github/workflows/ql-for-ql-* @github/codeql-ql-for-ql-reviewers
/.github/workflows/ruby-* @github/codeql-ruby
/.github/workflows/swift.yml @github/codeql-swift
/.github/workflows/swift-* @github/codeql-c

View File

@@ -25,7 +25,6 @@ If you have an idea for a query that you would like to share with other CodeQL u
Each language-specific directory contains further subdirectories that group queries based on their `@tags` or purpose.
- Experimental queries and libraries are stored in the `experimental` subdirectory within each language-specific directory in the [CodeQL repository](https://github.com/github/codeql). For example, experimental Java queries and libraries are stored in `java/ql/src/experimental` and any corresponding tests in `java/ql/test/experimental`.
- Experimental queries need to include `experimental` in their `@tags`
- The structure of an `experimental` subdirectory mirrors the structure of its parent directory.
- Select or create an appropriate directory in `experimental` based on the existing directory structure of `experimental` or its parent directory.

View File

@@ -4,7 +4,8 @@ This open source repository contains the standard CodeQL libraries and queries t
## How do I learn CodeQL and run queries?
There is [extensive documentation](https://codeql.github.com/docs/) on getting started with writing CodeQL using the [CodeQL extension for Visual Studio Code](https://codeql.github.com/docs/codeql-for-visual-studio-code/) and the [CodeQL CLI](https://codeql.github.com/docs/codeql-cli/).
There is [extensive documentation](https://codeql.github.com/docs/) on getting started with writing CodeQL.
You can use the [CodeQL for Visual Studio Code](https://codeql.github.com/docs/codeql-for-visual-studio-code/) extension or the [interactive query console](https://lgtm.com/help/lgtm/using-query-console) on LGTM.com (Semmle Legacy product) to try out your queries on any open source project that's currently being analyzed.
## Contributing

View File

@@ -52,7 +52,7 @@
| Unneeded defensive code | More true positive and fewer false positive results | This query now recognizes additional defensive code patterns. |
| Unsafe dynamic method access | Fewer false positive results | This query no longer flags concatenated strings as unsafe method names. |
| Unused parameter | Fewer false positive results | This query no longer flags parameters with leading underscore. |
| Unused variable, import, function or class | Fewer false positive results | This query now flags fewer variables that are implicitly used by JSX elements. It no longer flags variables with a leading underscore and variables in dead code. |
| Unused variable, import, function or class | Fewer false positive results | This query now flags fewer variables that are implictly used by JSX elements. It no longer flags variables with a leading underscore and variables in dead code. |
| Unvalidated dynamic method call | More true positive results | This query now flags concatenated strings as unvalidated method names in more cases. |
| Useless assignment to property. | Fewer false positive results | This query now treats assignments with complex right-hand sides correctly. |
| Useless conditional | Fewer results | Additional defensive coding patterns are now ignored. |

View File

@@ -19,7 +19,7 @@ The following changes in version 1.23 affect C/C++ analysis in all applications.
| Hard-coded Japanese era start date in call (`cpp/japanese-era/constructor-or-method-with-exact-era-date`) | Deprecated | This query has been deprecated. Use the new combined query Hard-coded Japanese era start date (`cpp/japanese-era/exact-era-date`) instead. |
| Hard-coded Japanese era start date in struct (`cpp/japanese-era/struct-with-exact-era-date`) | Deprecated | This query has been deprecated. Use the new combined query Hard-coded Japanese era start date (`cpp/japanese-era/exact-era-date`) instead. |
| Hard-coded Japanese era start date (`cpp/japanese-era/exact-era-date`) | More correct results | This query now checks for the beginning date of the Reiwa era (1st May 2019). |
| Non-constant format string (`cpp/non-constant-format`) | Fewer false positive results | Fixed false positive results triggered by mismatching declarations of a formatting function. |
| Non-constant format string (`cpp/non-constant-format`) | Fewer false positive results | Fixed false positive results triggrered by mismatching declarations of a formatting function. |
| Sign check of bitwise operation (`cpp/bitwise-sign-check`) | Fewer false positive results | Results involving `>=` or `<=` are no longer reported. |
| Too few arguments to formatting function (`cpp/wrong-number-format-arguments`) | Fewer false positive results | Fixed false positive results triggered by mismatching declarations of a formatting function. |
| Too many arguments to formatting function (`cpp/too-many-format-arguments`) | Fewer false positive results | Fixed false positive results triggered by mismatching declarations of a formatting function. |

View File

@@ -91,7 +91,7 @@
## Changes to libraries
* The predicates `RegExpTerm.getSuccessor` and `RegExpTerm.getPredecessor` have been changed to reflect textual, not operational, matching order. This only makes a difference in lookbehind assertions, which are operationally matched backwards. Previously, `getSuccessor` would mimic this, so in an assertion `(?<=ab)` the term `b` would be considered the predecessor, not the successor, of `a`. Textually, however, `a` is still matched before `b`, and this is the order we now follow.
* The predicates `RegExpTerm.getSuccessor` and `RegExpTerm.getPredecessor` have been changed to reflect textual, not operational, matching order. This only makes a difference in lookbehind assertions, which are operationally matched backwards. Previously, `getSuccessor` would mimick this, so in an assertion `(?<=ab)` the term `b` would be considered the predecessor, not the successor, of `a`. Textually, however, `a` is still matched before `b`, and this is the order we now follow.
* An extensible model of the `EventEmitter` pattern has been implemented.
* Taint-tracking configurations now interact differently with the `data` flow label, which may affect queries
that combine taint-tracking and flow labels.

View File

@@ -17,7 +17,6 @@ provide:
# - "javascript/ql/experimental/adaptivethreatmodeling/model/qlpack.yml"
- "javascript/ql/experimental/adaptivethreatmodeling/modelbuilding/qlpack.yml"
- "javascript/ql/experimental/adaptivethreatmodeling/src/qlpack.yml"
- "javascript/ql/experimental/adaptivethreatmodeling/test/qlpack.yml"
- "csharp/ql/campaigns/Solorigate/lib/qlpack.yml"
- "csharp/ql/campaigns/Solorigate/src/qlpack.yml"
- "csharp/ql/campaigns/Solorigate/test/qlpack.yml"
@@ -25,8 +24,7 @@ provide:
- "misc/suite-helpers/qlpack.yml"
- "ruby/extractor-pack/codeql-extractor.yml"
- "swift/extractor-pack/codeql-extractor.yml"
- "swift/integration-tests/qlpack.yml"
- "ql/extractor-pack/codeql-extractor.yml"
- "ql/extractor-pack/codeql-extractor.ym"
versionPolicies:
default:

View File

@@ -1,21 +0,0 @@
const mongoose = require('mongoose');
Logger = require('./logger').Logger;
Note = require('./models/note').Note;
(async () => {
if (process.argv.length != 5) {
Logger.log("Creates a private note. Usage: node add-note.js <token> <title> <body>")
return;
}
// Open the default mongoose connection
await mongoose.connect('mongodb://localhost:27017/notes', { useFindAndModify: false });
const [userToken, title, body] = process.argv.slice(2);
await Note.create({ title, body, userToken });
Logger.log(`Created private note with title ${title} and body ${body} belonging to user with token ${userToken}.`);
await mongoose.connection.close();
})();

View File

@@ -1,68 +0,0 @@
const bodyParser = require('body-parser');
const express = require('express');
const mongoose = require('mongoose');
const notesApi = require('./notes-api');
const usersApi = require('./users-api');
const addSampleData = module.exports.addSampleData = async () => {
const [userA, userB] = await User.create([
{
name: "A",
token: "tokenA"
},
{
name: "B",
token: "tokenB"
}
]);
await Note.create([
{
title: "Public note belonging to A",
body: "This is a public note belonging to A",
isPublic: true,
ownerToken: userA.token
},
{
title: "Public note belonging to B",
body: "This is a public note belonging to B",
isPublic: true,
ownerToken: userB.token
},
{
title: "Private note belonging to A",
body: "This is a private note belonging to A",
ownerToken: userA.token
},
{
title: "Private note belonging to B",
body: "This is a private note belonging to B",
ownerToken: userB.token
}
]);
}
module.exports.startApp = async () => {
// Open the default mongoose connection
await mongoose.connect('mongodb://mongo:27017/notes', { useFindAndModify: false });
// Drop contents of DB
mongoose.connection.dropDatabase();
// Add some sample data
await addSampleData();
const app = express();
app.use(bodyParser.json());
app.use(bodyParser.urlencoded());
app.get('/', async (_req, res) => {
res.send('Hello World');
});
app.use('/api/notes', notesApi.router);
app.use('/api/users', usersApi.router);
app.listen(3000);
Logger.log('Express started on port 3000');
};

View File

@@ -1,7 +0,0 @@
const startApp = require('./app').startApp;
Logger = require('./logger').Logger;
Note = require('./models/note').Note;
User = require('./models/user').User;
startApp();

View File

@@ -1,5 +0,0 @@
module.exports.Logger = class {
log(message, ...objs) {
console.log(message, objs);
}
};

View File

@@ -1,8 +0,0 @@
const mongoose = require('mongoose');
module.exports.Note = mongoose.model('Note', new mongoose.Schema({
title: String,
body: String,
ownerToken: String,
isPublic: Boolean
}));

View File

@@ -1,6 +0,0 @@
const mongoose = require('mongoose');
module.exports.User = mongoose.model('User', new mongoose.Schema({
name: String,
token: String
}));

View File

@@ -1,44 +0,0 @@
const express = require('express')
const router = module.exports.router = express.Router();
function serializeNote(note) {
return {
title: note.title,
body: note.body
};
}
router.post('/find', async (req, res) => {
const notes = await Note.find({
ownerToken: req.body.token
}).exec();
res.json({
notes: notes.map(serializeNote)
});
});
router.get('/findPublic', async (_req, res) => {
const notes = await Note.find({
isPublic: true
}).exec();
res.json({
notes: notes.map(serializeNote)
});
});
router.post('/findVisible', async (req, res) => {
const notes = await Note.find({
$or: [
{
isPublic: true
},
{
ownerToken: req.body.token
}
]
}).exec();
res.json({
notes: notes.map(serializeNote)
});
});

View File

@@ -1,37 +0,0 @@
const mongoose = require('mongoose');
Logger = require('./logger').Logger;
Note = require('./models/note').Note;
User = require('./models/user').User;
(async () => {
if (process.argv.length != 3) {
Logger.log("Outputs all notes visible to a user. Usage: node read-notes.js <token>")
return;
}
// Open the default mongoose connection
await mongoose.connect('mongodb://localhost:27017/notes', { useFindAndModify: false });
const ownerToken = process.argv[2];
const user = await User.findOne({
token: ownerToken
}).exec();
const notes = await Note.find({
$or: [
{ isPublic: true },
{ ownerToken }
]
}).exec();
notes.map(note => {
Logger.log("Title:" + note.title);
Logger.log("By:" + user.name);
Logger.log("Body:" + note.body);
Logger.log();
});
await mongoose.connection.close();
})();

View File

@@ -1,25 +0,0 @@
const express = require('express')
Logger = require('./logger').Logger;
const router = module.exports.router = express.Router();
router.post('/updateName', async (req, res) => {
Logger.log("/updateName called with new name", req.body.name);
await User.findOneAndUpdate({
token: req.body.token
}, {
name: req.body.name
}).exec();
res.json({
name: req.body.name
});
});
router.post('/getName', async (req, res) => {
const user = await User.findOne({
token: req.body.token
}).exec();
res.json({
name: user.name
});
});

View File

@@ -1,5 +1,5 @@
{
"DataFlow Java/C++/C#/Go/Python/Ruby/Swift": [
"DataFlow Java/C++/C#/Python": [
"java/ql/lib/semmle/code/java/dataflow/internal/DataFlowImpl.qll",
"java/ql/lib/semmle/code/java/dataflow/internal/DataFlowImpl2.qll",
"java/ql/lib/semmle/code/java/dataflow/internal/DataFlowImpl3.qll",
@@ -27,32 +27,27 @@
"csharp/ql/lib/semmle/code/csharp/dataflow/internal/DataFlowImpl4.qll",
"csharp/ql/lib/semmle/code/csharp/dataflow/internal/DataFlowImpl5.qll",
"csharp/ql/lib/semmle/code/csharp/dataflow/internal/DataFlowImplForContentDataFlow.qll",
"go/ql/lib/semmle/go/dataflow/internal/DataFlowImpl.qll",
"go/ql/lib/semmle/go/dataflow/internal/DataFlowImpl2.qll",
"go/ql/lib/semmle/go/dataflow/internal/DataFlowImplForStringsNewReplacer.qll",
"python/ql/lib/semmle/python/dataflow/new/internal/DataFlowImpl.qll",
"python/ql/lib/semmle/python/dataflow/new/internal/DataFlowImpl2.qll",
"python/ql/lib/semmle/python/dataflow/new/internal/DataFlowImpl3.qll",
"python/ql/lib/semmle/python/dataflow/new/internal/DataFlowImpl4.qll",
"ruby/ql/lib/codeql/ruby/dataflow/internal/DataFlowImpl.qll",
"ruby/ql/lib/codeql/ruby/dataflow/internal/DataFlowImpl2.qll",
"ruby/ql/lib/codeql/ruby/dataflow/internal/DataFlowImplForRegExp.qll",
"ruby/ql/lib/codeql/ruby/dataflow/internal/DataFlowImplForLibraries.qll",
"ruby/ql/lib/codeql/ruby/dataflow/internal/DataFlowImplForHttpClientLibraries.qll",
"ruby/ql/lib/codeql/ruby/dataflow/internal/DataFlowImplForPathname.qll",
"swift/ql/lib/codeql/swift/dataflow/internal/DataFlowImpl.qll"
],
"DataFlow Java/C++/C#/Go/Python/Ruby/Swift Common": [
"DataFlow Java/C++/C#/Python Common": [
"java/ql/lib/semmle/code/java/dataflow/internal/DataFlowImplCommon.qll",
"cpp/ql/lib/semmle/code/cpp/dataflow/internal/DataFlowImplCommon.qll",
"cpp/ql/lib/semmle/code/cpp/ir/dataflow/internal/DataFlowImplCommon.qll",
"cpp/ql/lib/experimental/semmle/code/cpp/ir/dataflow/internal/DataFlowImplCommon.qll",
"csharp/ql/lib/semmle/code/csharp/dataflow/internal/DataFlowImplCommon.qll",
"go/ql/lib/semmle/go/dataflow/internal/DataFlowImplCommon.qll",
"python/ql/lib/semmle/python/dataflow/new/internal/DataFlowImplCommon.qll",
"ruby/ql/lib/codeql/ruby/dataflow/internal/DataFlowImplCommon.qll",
"swift/ql/lib/codeql/swift/dataflow/internal/DataFlowImplCommon.qll"
],
"TaintTracking::Configuration Java/C++/C#/Go/Python/Ruby/Swift": [
"TaintTracking::Configuration Java/C++/C#/Python": [
"cpp/ql/lib/semmle/code/cpp/dataflow/internal/tainttracking1/TaintTrackingImpl.qll",
"cpp/ql/lib/semmle/code/cpp/dataflow/internal/tainttracking2/TaintTrackingImpl.qll",
"cpp/ql/lib/semmle/code/cpp/ir/dataflow/internal/tainttracking1/TaintTrackingImpl.qll",
@@ -66,8 +61,6 @@
"csharp/ql/lib/semmle/code/csharp/dataflow/internal/tainttracking3/TaintTrackingImpl.qll",
"csharp/ql/lib/semmle/code/csharp/dataflow/internal/tainttracking4/TaintTrackingImpl.qll",
"csharp/ql/lib/semmle/code/csharp/dataflow/internal/tainttracking5/TaintTrackingImpl.qll",
"go/ql/lib/semmle/go/dataflow/internal/tainttracking1/TaintTrackingImpl.qll",
"go/ql/lib/semmle/go/dataflow/internal/tainttracking2/TaintTrackingImpl.qll",
"java/ql/lib/semmle/code/java/dataflow/internal/tainttracking1/TaintTrackingImpl.qll",
"java/ql/lib/semmle/code/java/dataflow/internal/tainttracking2/TaintTrackingImpl.qll",
"java/ql/lib/semmle/code/java/dataflow/internal/tainttracking3/TaintTrackingImpl.qll",
@@ -76,9 +69,10 @@
"python/ql/lib/semmle/python/dataflow/new/internal/tainttracking3/TaintTrackingImpl.qll",
"python/ql/lib/semmle/python/dataflow/new/internal/tainttracking4/TaintTrackingImpl.qll",
"ruby/ql/lib/codeql/ruby/dataflow/internal/tainttracking1/TaintTrackingImpl.qll",
"ruby/ql/lib/codeql/ruby/dataflow/internal/tainttrackingforlibraries/TaintTrackingImpl.qll",
"swift/ql/lib/codeql/swift/dataflow/internal/tainttracking1/TaintTrackingImpl.qll"
],
"DataFlow Java/C++/C#/Python/Ruby/Swift Consistency checks": [
"DataFlow Java/C++/C#/Python Consistency checks": [
"java/ql/lib/semmle/code/java/dataflow/internal/DataFlowImplConsistency.qll",
"cpp/ql/lib/semmle/code/cpp/dataflow/internal/DataFlowImplConsistency.qll",
"cpp/ql/lib/semmle/code/cpp/ir/dataflow/internal/DataFlowImplConsistency.qll",
@@ -88,10 +82,9 @@
"ruby/ql/lib/codeql/ruby/dataflow/internal/DataFlowImplConsistency.qll",
"swift/ql/lib/codeql/swift/dataflow/internal/DataFlowImplConsistency.qll"
],
"DataFlow Java/C#/Go/Ruby/Python/Swift Flow Summaries": [
"DataFlow Java/C#/Ruby/Python/Swift Flow Summaries": [
"java/ql/lib/semmle/code/java/dataflow/internal/FlowSummaryImpl.qll",
"csharp/ql/lib/semmle/code/csharp/dataflow/internal/FlowSummaryImpl.qll",
"go/ql/lib/semmle/go/dataflow/internal/FlowSummaryImpl.qll",
"ruby/ql/lib/codeql/ruby/dataflow/internal/FlowSummaryImpl.qll",
"python/ql/lib/semmle/python/dataflow/new/internal/FlowSummaryImpl.qll",
"swift/ql/lib/codeql/swift/dataflow/internal/FlowSummaryImpl.qll"
@@ -101,8 +94,8 @@
"csharp/ql/lib/semmle/code/csharp/dataflow/internal/rangeanalysis/SsaReadPositionCommon.qll"
],
"Model as Data Generation Java/C# - CaptureModels": [
"java/ql/src/utils/modelgenerator/internal/CaptureModels.qll",
"csharp/ql/src/utils/modelgenerator/internal/CaptureModels.qll"
"java/ql/src/utils/model-generator/internal/CaptureModels.qll",
"csharp/ql/src/utils/model-generator/internal/CaptureModels.qll"
],
"Sign Java/C#": [
"java/ql/lib/semmle/code/java/dataflow/internal/rangeanalysis/Sign.qll",
@@ -403,6 +396,16 @@
"csharp/ql/lib/semmle/code/csharp/dataflow/internal/ControlFlowReachability.qll",
"csharp/ql/lib/semmle/code/csharp/dataflow/internal/rangeanalysis/ControlFlowReachability.qll"
],
"Inline Test Expectations": [
"cpp/ql/test/TestUtilities/InlineExpectationsTest.qll",
"csharp/ql/test/TestUtilities/InlineExpectationsTest.qll",
"java/ql/test/TestUtilities/InlineExpectationsTest.qll",
"python/ql/test/TestUtilities/InlineExpectationsTest.qll",
"ruby/ql/test/TestUtilities/InlineExpectationsTest.qll",
"ql/ql/test/TestUtilities/InlineExpectationsTest.qll",
"go/ql/test/TestUtilities/InlineExpectationsTest.qll",
"swift/ql/test/TestUtilities/InlineExpectationsTest.qll"
],
"C++ ExternalAPIs": [
"cpp/ql/src/Security/CWE/CWE-020/ExternalAPIs.qll",
"cpp/ql/src/Security/CWE/CWE-020/ir/ExternalAPIs.qll"
@@ -461,10 +464,6 @@
"javascript/ql/src/Comments/CommentedOutCodeReferences.inc.qhelp",
"python/ql/src/Lexical/CommentedOutCodeReferences.inc.qhelp"
],
"ThreadResourceAbuse qhelp": [
"java/ql/src/experimental/Security/CWE/CWE-400/LocalThreadResourceAbuse.qhelp",
"java/ql/src/experimental/Security/CWE/CWE-400/ThreadResourceAbuse.qhelp"
],
"IDE Contextual Queries": [
"cpp/ql/lib/IDEContextual.qll",
"csharp/ql/lib/IDEContextual.qll",
@@ -487,6 +486,40 @@
"python/ql/lib/semmle/python/security/internal/SensitiveDataHeuristics.qll",
"ruby/ql/lib/codeql/ruby/security/internal/SensitiveDataHeuristics.qll"
],
"ReDoS Util Python/JS/Ruby/Java": [
"javascript/ql/lib/semmle/javascript/security/regexp/NfaUtils.qll",
"python/ql/lib/semmle/python/security/regexp/NfaUtils.qll",
"ruby/ql/lib/codeql/ruby/security/regexp/NfaUtils.qll",
"java/ql/lib/semmle/code/java/security/regexp/NfaUtils.qll"
],
"ReDoS Exponential Python/JS/Ruby/Java": [
"javascript/ql/lib/semmle/javascript/security/regexp/ExponentialBackTracking.qll",
"python/ql/lib/semmle/python/security/regexp/ExponentialBackTracking.qll",
"ruby/ql/lib/codeql/ruby/security/regexp/ExponentialBackTracking.qll",
"java/ql/lib/semmle/code/java/security/regexp/ExponentialBackTracking.qll"
],
"ReDoS Polynomial Python/JS/Ruby/Java": [
"javascript/ql/lib/semmle/javascript/security/regexp/SuperlinearBackTracking.qll",
"python/ql/lib/semmle/python/security/regexp/SuperlinearBackTracking.qll",
"ruby/ql/lib/codeql/ruby/security/regexp/SuperlinearBackTracking.qll",
"java/ql/lib/semmle/code/java/security/regexp/SuperlinearBackTracking.qll"
],
"RegexpMatching Python/JS/Ruby": [
"javascript/ql/lib/semmle/javascript/security/regexp/RegexpMatching.qll",
"python/ql/lib/semmle/python/security/regexp/RegexpMatching.qll",
"ruby/ql/lib/codeql/ruby/security/regexp/RegexpMatching.qll"
],
"BadTagFilterQuery Python/JS/Ruby": [
"javascript/ql/lib/semmle/javascript/security/BadTagFilterQuery.qll",
"python/ql/lib/semmle/python/security/BadTagFilterQuery.qll",
"ruby/ql/lib/codeql/ruby/security/BadTagFilterQuery.qll"
],
"OverlyLargeRange Python/JS/Ruby/Java": [
"javascript/ql/lib/semmle/javascript/security/OverlyLargeRangeQuery.qll",
"python/ql/lib/semmle/python/security/OverlyLargeRangeQuery.qll",
"ruby/ql/lib/codeql/ruby/security/OverlyLargeRangeQuery.qll",
"java/ql/lib/semmle/code/java/security/OverlyLargeRangeQuery.qll"
],
"CFG": [
"csharp/ql/lib/semmle/code/csharp/controlflow/internal/ControlFlowGraphImplShared.qll",
"ruby/ql/lib/codeql/ruby/controlflow/internal/ControlFlowGraphImplShared.qll",
@@ -496,9 +529,16 @@
"python/ql/lib/semmle/python/dataflow/new/internal/TypeTracker.qll",
"ruby/ql/lib/codeql/ruby/typetracking/TypeTracker.qll"
],
"CodeQL Tutorial": [
"cpp/ql/lib/tutorial.qll",
"csharp/ql/lib/tutorial.qll",
"java/ql/lib/tutorial.qll",
"javascript/ql/lib/tutorial.qll",
"python/ql/lib/tutorial.qll",
"ruby/ql/lib/tutorial.qll"
],
"AccessPathSyntax": [
"csharp/ql/lib/semmle/code/csharp/dataflow/internal/AccessPathSyntax.qll",
"go/ql/lib/semmle/go/dataflow/internal/AccessPathSyntax.qll",
"java/ql/lib/semmle/code/java/dataflow/internal/AccessPathSyntax.qll",
"javascript/ql/lib/semmle/javascript/frameworks/data/internal/AccessPathSyntax.qll",
"ruby/ql/lib/codeql/ruby/dataflow/internal/AccessPathSyntax.qll",
@@ -514,16 +554,16 @@
"ruby/ql/lib/codeql/ruby/internal/ConceptsShared.qll",
"javascript/ql/lib/semmle/javascript/internal/ConceptsShared.qll"
],
"Hostname Regexp queries": [
"javascript/ql/src/Security/CWE-020/HostnameRegexpShared.qll",
"python/ql/src/Security/CWE-020/HostnameRegexpShared.qll",
"ruby/ql/src/queries/security/cwe-020/HostnameRegexpShared.qll"
],
"ApiGraphModels": [
"javascript/ql/lib/semmle/javascript/frameworks/data/internal/ApiGraphModels.qll",
"ruby/ql/lib/codeql/ruby/frameworks/data/internal/ApiGraphModels.qll",
"python/ql/lib/semmle/python/frameworks/data/internal/ApiGraphModels.qll"
],
"ApiGraphModelsExtensions": [
"javascript/ql/lib/semmle/javascript/frameworks/data/internal/ApiGraphModelsExtensions.qll",
"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"
@@ -560,16 +600,8 @@
"swift/ql/test/extractor-tests/patterns/patterns.swift",
"swift/ql/test/library-tests/ast/patterns.swift"
],
"Swift control flow test file": [
"swift/ql/test/library-tests/controlflow/graph/cfg.swift",
"swift/ql/test/library-tests/ast/cfg.swift"
],
"IncompleteMultiCharacterSanitization JS/Ruby": [
"javascript/ql/lib/semmle/javascript/security/IncompleteMultiCharacterSanitizationQuery.qll",
"ruby/ql/lib/codeql/ruby/security/IncompleteMultiCharacterSanitizationQuery.qll"
],
"EncryptionKeySizes Python/Java": [
"python/ql/lib/semmle/python/security/internal/EncryptionKeySizes.qll",
"java/ql/lib/semmle/code/java/security/internal/EncryptionKeySizes.qll"
]
}
}

View File

@@ -257,11 +257,11 @@ namespace Semmle.Autobuild.Cpp.Tests
Actions.GetCurrentDirectory = cwd;
Actions.IsWindows = isWindows;
var options = new CppAutobuildOptions(Actions);
var options = new AutobuildOptions(Actions, Language.Cpp);
return new CppAutobuilder(Actions, options);
}
void TestAutobuilderScript(CppAutobuilder autobuilder, int expectedOutput, int commandsRun)
void TestAutobuilderScript(Autobuilder autobuilder, int expectedOutput, int commandsRun)
{
Assert.Equal(expectedOutput, autobuilder.GetBuildScript().Run(Actions, StartCallback, EndCallback));
@@ -299,7 +299,7 @@ namespace Semmle.Autobuild.Cpp.Tests
{
Actions.RunProcess[@"cmd.exe /C nuget restore C:\Project\test.sln -DisableParallelProcessing"] = 1;
Actions.RunProcess[@"cmd.exe /C C:\Project\.nuget\nuget.exe restore C:\Project\test.sln -DisableParallelProcessing"] = 0;
Actions.RunProcess[@"cmd.exe /C CALL ^""C:\Program Files ^(x86^)\Microsoft Visual Studio 14.0\VC\vcvarsall.bat^"" && set Platform=&& type NUL && msbuild C:\Project\test.sln /t:rebuild /p:Platform=""x86"" /p:Configuration=""Release"""] = 0;
Actions.RunProcess[@"cmd.exe /C CALL ^""C:\Program Files ^(x86^)\Microsoft Visual Studio 14.0\VC\vcvarsall.bat^"" && set Platform=&& type NUL && C:\odasa\tools\odasa index --auto msbuild C:\Project\test.sln /t:rebuild /p:Platform=""x86"" /p:Configuration=""Release"" /p:MvcBuildViews=true"] = 0;
Actions.RunProcessOut[@"C:\Program Files (x86)\Microsoft Visual Studio\Installer\vswhere.exe -prerelease -legacy -property installationPath"] = "";
Actions.RunProcess[@"C:\Program Files (x86)\Microsoft Visual Studio\Installer\vswhere.exe -prerelease -legacy -property installationPath"] = 1;
Actions.RunProcess[@"C:\Program Files (x86)\Microsoft Visual Studio\Installer\vswhere.exe -prerelease -legacy -property installationVersion"] = 0;

View File

@@ -11,12 +11,11 @@
<ItemGroup>
<PackageReference Include="System.IO.FileSystem" Version="4.3.0" />
<PackageReference Include="System.IO.FileSystem.Primitives" Version="4.3.0" />
<PackageReference Include="xunit" Version="2.4.2" />
<PackageReference Include="xunit.runner.visualstudio" Version="2.4.5">
<PackageReference Include="xunit" Version="2.4.1" />
<PackageReference Include="xunit.runner.visualstudio" Version="2.4.1">
<PrivateAssets>all</PrivateAssets>
<IncludeAssets>runtime; build; native; contentfiles; analyzers</IncludeAssets>
</PackageReference>
<PackageReference Include="Microsoft.NET.Test.Sdk" Version="17.4.0" />
</ItemGroup>
<ItemGroup>

View File

@@ -2,26 +2,9 @@
namespace Semmle.Autobuild.Cpp
{
/// <summary>
/// Encapsulates C++ build options.
/// </summary>
public class CppAutobuildOptions : AutobuildOptionsShared
public class CppAutobuilder : Autobuilder
{
public override Language Language => Language.Cpp;
/// <summary>
/// Reads options from environment variables.
/// Throws ArgumentOutOfRangeException for invalid arguments.
/// </summary>
public CppAutobuildOptions(IBuildActions actions) : base(actions)
{
}
}
public class CppAutobuilder : Autobuilder<CppAutobuildOptions>
{
public CppAutobuilder(IBuildActions actions, CppAutobuildOptions options) : base(actions, options) { }
public CppAutobuilder(IBuildActions actions, AutobuildOptions options) : base(actions, options) { }
public override BuildScript GetBuildScript()
{

View File

@@ -11,14 +11,14 @@ namespace Semmle.Autobuild.Cpp
try
{
var actions = SystemBuildActions.Instance;
var options = new CppAutobuildOptions(actions);
var options = new AutobuildOptions(actions, Language.Cpp);
try
{
Console.WriteLine("CodeQL C++ autobuilder");
var builder = new CppAutobuilder(actions, options);
return builder.AttemptBuild();
}
catch (InvalidEnvironmentException ex)
catch(InvalidEnvironmentException ex)
{
Console.WriteLine("The environment is invalid: {0}", ex.Message);
}

View File

@@ -17,7 +17,7 @@
</ItemGroup>
<ItemGroup>
<PackageReference Include="Microsoft.Build" Version="17.3.2" />
<PackageReference Include="Microsoft.Build" Version="16.11.0" />
</ItemGroup>
<ItemGroup>

View File

@@ -13,5 +13,5 @@ predicate isExprWithNewBuiltin(Expr expr) {
from Expr expr, int kind, int kind_new, Location location
where
exprs(expr, kind, location) and
if isExprWithNewBuiltin(expr) then kind_new = 1 else kind_new = kind
if isExprWithNewBuiltin(expr) then kind_new = 0 else kind_new = kind
select expr, kind_new, location

View File

@@ -9,5 +9,5 @@ class Location extends @location_expr {
from Expr expr, int kind, int kind_new, Location location
where
exprs(expr, kind, location) and
if expr instanceof @blockassignexpr then kind_new = 1 else kind_new = kind
if expr instanceof @blockassignexpr then kind_new = 0 else kind_new = kind
select expr, kind_new, location

View File

@@ -1,11 +0,0 @@
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 @float16 or type instanceof @complex_float16
then kind_new = 2
else kind_new = kind
select type, name, kind_new, size, sign, alignment

View File

@@ -1,3 +0,0 @@
description: Introduce (_Complex) _Float16 type
compatibility: backwards
builtintypes.rel: run builtintypes.qlo

View File

@@ -1,2 +0,0 @@
description: Uncomment case splits in dbscheme
compatibility: full

View File

@@ -3,4 +3,4 @@ groups:
- cpp
- examples
dependencies:
codeql/cpp-all: ${workspace}
codeql/cpp-all: "*"

View File

@@ -1,54 +1,3 @@
## 0.5.1
No user-facing changes.
## 0.5.0
### Breaking Changes
The predicates in the `MustFlow::Configuration` class used by the `MustFlow` library (`semmle.code.cpp.ir.dataflow.MustFlow`) have changed to be defined directly in terms of the C++ IR instead of IR dataflow nodes.
### Deprecated APIs
* Deprecated `semmle.code.cpp.ir.dataflow.DefaultTaintTracking`. Use `semmle.code.cpp.ir.dataflow.TaintTracking`.
* Deprecated `semmle.code.cpp.security.TaintTrackingImpl`. Use `semmle.code.cpp.ir.dataflow.TaintTracking`.
* Deprecated `semmle.code.cpp.valuenumbering.GlobalValueNumberingImpl`. Use `semmle.code.cpp.valuenumbering.GlobalValueNumbering`, which exposes the same API.
### Minor Analysis Improvements
* The `ArgvSource` flow source now uses the second parameter of `main` as its source instead of the uses of this parameter.
* The `ArgvSource` flow source has been generalized to handle cases where the argument vector of `main` is not named `argv`.
* The `getaddrinfo` function is now recognized as a flow source.
* The `secure_getenv` and `_wgetenv` functions are now recognized as local flow sources.
* The `scanf` and `fscanf` functions and their variants are now recognized as flow sources.
* Deleted the deprecated `getName` and `getShortName` predicates from the `Folder` class.
## 0.4.6
No user-facing changes.
## 0.4.5
No user-facing changes.
## 0.4.4
No user-facing changes.
## 0.4.3
### Minor Analysis Improvements
* Fixed bugs in the `FormatLiteral` class that were causing `getMaxConvertedLength` and related predicates to return no results when the format literal was `%e`, `%f` or `%g` and an explicit precision was specified.
## 0.4.2
No user-facing changes.
## 0.4.1
No user-facing changes.
## 0.4.0
### Deprecated APIs

View File

@@ -1,3 +0,0 @@
## 0.4.1
No user-facing changes.

View File

@@ -1,3 +0,0 @@
## 0.4.2
No user-facing changes.

View File

@@ -1,5 +0,0 @@
## 0.4.3
### Minor Analysis Improvements
* Fixed bugs in the `FormatLiteral` class that were causing `getMaxConvertedLength` and related predicates to return no results when the format literal was `%e`, `%f` or `%g` and an explicit precision was specified.

View File

@@ -1,3 +0,0 @@
## 0.4.4
No user-facing changes.

View File

@@ -1,3 +0,0 @@
## 0.4.5
No user-facing changes.

View File

@@ -1,3 +0,0 @@
## 0.4.6
No user-facing changes.

View File

@@ -1,20 +0,0 @@
## 0.5.0
### Breaking Changes
The predicates in the `MustFlow::Configuration` class used by the `MustFlow` library (`semmle.code.cpp.ir.dataflow.MustFlow`) have changed to be defined directly in terms of the C++ IR instead of IR dataflow nodes.
### Deprecated APIs
* Deprecated `semmle.code.cpp.ir.dataflow.DefaultTaintTracking`. Use `semmle.code.cpp.ir.dataflow.TaintTracking`.
* Deprecated `semmle.code.cpp.security.TaintTrackingImpl`. Use `semmle.code.cpp.ir.dataflow.TaintTracking`.
* Deprecated `semmle.code.cpp.valuenumbering.GlobalValueNumberingImpl`. Use `semmle.code.cpp.valuenumbering.GlobalValueNumbering`, which exposes the same API.
### Minor Analysis Improvements
* The `ArgvSource` flow source now uses the second parameter of `main` as its source instead of the uses of this parameter.
* The `ArgvSource` flow source has been generalized to handle cases where the argument vector of `main` is not named `argv`.
* The `getaddrinfo` function is now recognized as a flow source.
* The `secure_getenv` and `_wgetenv` functions are now recognized as local flow sources.
* The `scanf` and `fscanf` functions and their variants are now recognized as flow sources.
* Deleted the deprecated `getName` and `getShortName` predicates from the `Folder` class.

View File

@@ -1,3 +0,0 @@
## 0.5.1
No user-facing changes.

View File

@@ -1,2 +1,2 @@
---
lastReleaseVersion: 0.5.1
lastReleaseVersion: 0.4.0

View File

@@ -12,8 +12,8 @@ import IDEContextual
*
* In some cases it is preferable to modify locations (the
* `hasLocationInfo()` predicate) so that they are short, and
* non-overlapping with other locations that might be reported as
* code scanning alerts on GitHub.
* non-overlapping with other locations that might be highlighted in
* the LGTM interface.
*
* We need to give locations that may not be in the database, so
* we use `hasLocationInfo()` rather than `getLocation()`.
@@ -123,13 +123,6 @@ private predicate constructorCallTypeMention(ConstructorCall cc, TypeMention tm)
)
}
/** Holds if `loc` has the container `container` and is on the line starting at `startLine`. */
pragma[nomagic]
private predicate hasContainerAndStartLine(Location loc, Container container, int startLine) {
loc.getStartLine() = startLine and
loc.getContainer() = container
}
/**
* Gets an element, of kind `kind`, that element `e` uses, if any.
* Attention: This predicate yields multiple definitions for a single location.
@@ -166,9 +159,9 @@ Top definitionOf(Top e, string kind) {
// Multiple type mentions can be generated when a typedef is used, and
// in such cases we want to exclude all but the originating typedef.
not exists(Type secondary |
exists(File f, int startline, int startcol |
exists(TypeMention tm, File f, int startline, int startcol |
typeMentionStartLoc(e, result, f, startline, startcol) and
typeMentionStartLoc(_, secondary, f, startline, startcol) and
typeMentionStartLoc(tm, secondary, f, startline, startcol) and
(
result = secondary.(TypedefType).getBaseType() or
result = secondary.(TypedefType).getBaseType().(SpecifiedType).getBaseType()
@@ -191,9 +184,11 @@ Top definitionOf(Top e, string kind) {
kind = "I" and
result = e.(Include).getIncludedFile() and
// exclude `#include` directives containing macros
not exists(MacroInvocation mi, Container container, int startLine |
hasContainerAndStartLine(e.(Include).getLocation(), container, startLine) and
hasContainerAndStartLine(mi.getLocation(), container, startLine)
not exists(MacroInvocation mi, Location l1, Location l2 |
l1 = e.(Include).getLocation() and
l2 = mi.getLocation() and
l1.getContainer() = l2.getContainer() and
l1.getStartLine() = l2.getStartLine()
// (an #include directive must be always on it's own line)
)
) and

View File

@@ -20,8 +20,7 @@ module ProductFlow {
* `source1` and `source2` must belong to the same callable.
*/
predicate isSourcePair(
DataFlow::Node source1, DataFlow::FlowState state1, DataFlow::Node source2,
DataFlow::FlowState state2
DataFlow::Node source1, string state1, DataFlow::Node source2, string state2
) {
state1 = "" and
state2 = "" and
@@ -90,61 +89,6 @@ module ProductFlow {
*/
predicate isBarrierOut2(DataFlow::Node node) { none() }
/*
* Holds if data may flow from `node1` to `node2` in addition to the normal data-flow steps in
* the first projection of the product dataflow graph.
*/
predicate isAdditionalFlowStep1(DataFlow::Node node1, DataFlow::Node node2) { none() }
/**
* Holds if data may flow from `node1` to `node2` in addition to the normal data-flow steps in
* the first projection of the product dataflow graph.
*
* This step is only applicable in `state1` and updates the flow state to `state2`.
*/
predicate isAdditionalFlowStep1(
DataFlow::Node node1, DataFlow::FlowState state1, DataFlow::Node node2,
DataFlow::FlowState state2
) {
state1 instanceof DataFlow::FlowStateEmpty and
state2 instanceof DataFlow::FlowStateEmpty and
this.isAdditionalFlowStep1(node1, node2)
}
/**
* Holds if data may flow from `node1` to `node2` in addition to the normal data-flow steps in
* the second projection of the product dataflow graph.
*/
predicate isAdditionalFlowStep2(DataFlow::Node node1, DataFlow::Node node2) { none() }
/**
* Holds if data may flow from `node1` to `node2` in addition to the normal data-flow steps in
* the second projection of the product dataflow graph.
*
* This step is only applicable in `state1` and updates the flow state to `state2`.
*/
predicate isAdditionalFlowStep2(
DataFlow::Node node1, DataFlow::FlowState state1, DataFlow::Node node2,
DataFlow::FlowState state2
) {
state1 instanceof DataFlow::FlowStateEmpty and
state2 instanceof DataFlow::FlowStateEmpty and
this.isAdditionalFlowStep2(node1, node2)
}
/**
* Holds if data flow into `node` is prohibited in the first projection of the product
* dataflow graph.
*/
predicate isBarrierIn1(DataFlow::Node node) { none() }
/**
* Holds if data flow into `node` is prohibited in the second projection of the product
* dataflow graph.
*/
predicate isBarrierIn2(DataFlow::Node node) { none() }
predicate hasFlowPath(
DataFlow::PathNode source1, DataFlow2::PathNode source2, DataFlow::PathNode sink1,
DataFlow2::PathNode sink2
@@ -159,78 +103,54 @@ module ProductFlow {
class Conf1 extends DataFlow::Configuration {
Conf1() { this = "Conf1" }
override predicate isSource(DataFlow::Node source, DataFlow::FlowState state) {
override predicate isSource(DataFlow::Node source, string state) {
exists(Configuration conf | conf.isSourcePair(source, state, _, _))
}
override predicate isSink(DataFlow::Node sink, DataFlow::FlowState state) {
override predicate isSink(DataFlow::Node sink, string state) {
exists(Configuration conf | conf.isSinkPair(sink, state, _, _))
}
override predicate isBarrier(DataFlow::Node node, DataFlow::FlowState state) {
override predicate isBarrier(DataFlow::Node node, string state) {
exists(Configuration conf | conf.isBarrier1(node, state))
}
override predicate isBarrierOut(DataFlow::Node node) {
exists(Configuration conf | conf.isBarrierOut1(node))
}
override predicate isAdditionalFlowStep(
DataFlow::Node node1, DataFlow::FlowState state1, DataFlow::Node node2,
DataFlow::FlowState state2
) {
exists(Configuration conf | conf.isAdditionalFlowStep1(node1, state1, node2, state2))
}
override predicate isBarrierIn(DataFlow::Node node) {
exists(Configuration conf | conf.isBarrierIn1(node))
}
}
class Conf2 extends DataFlow2::Configuration {
Conf2() { this = "Conf2" }
override predicate isSource(DataFlow::Node source, DataFlow::FlowState state) {
exists(Configuration conf, DataFlow::PathNode source1 |
conf.isSourcePair(source1.getNode(), source1.getState(), source, state) and
any(Conf1 c).hasFlowPath(source1, _)
override predicate isSource(DataFlow::Node source, string state) {
exists(Configuration conf, DataFlow::Node source1 |
conf.isSourcePair(source1, _, source, state) and
any(Conf1 c).hasFlow(source1, _)
)
}
override predicate isSink(DataFlow::Node sink, DataFlow::FlowState state) {
exists(Configuration conf, DataFlow::PathNode sink1 |
conf.isSinkPair(sink1.getNode(), sink1.getState(), sink, state) and
any(Conf1 c).hasFlowPath(_, sink1)
override predicate isSink(DataFlow::Node sink, string state) {
exists(Configuration conf, DataFlow::Node sink1 |
conf.isSinkPair(sink1, _, sink, state) and any(Conf1 c).hasFlow(_, sink1)
)
}
override predicate isBarrier(DataFlow::Node node, DataFlow::FlowState state) {
override predicate isBarrier(DataFlow::Node node, string state) {
exists(Configuration conf | conf.isBarrier2(node, state))
}
override predicate isBarrierOut(DataFlow::Node node) {
exists(Configuration conf | conf.isBarrierOut2(node))
}
override predicate isAdditionalFlowStep(
DataFlow::Node node1, DataFlow::FlowState state1, DataFlow::Node node2,
DataFlow::FlowState state2
) {
exists(Configuration conf | conf.isAdditionalFlowStep2(node1, state1, node2, state2))
}
override predicate isBarrierIn(DataFlow::Node node) {
exists(Configuration conf | conf.isBarrierIn2(node))
}
}
}
pragma[nomagic]
private predicate reachableInterprocEntry(
Configuration conf, DataFlow::PathNode source1, DataFlow2::PathNode source2,
DataFlow::PathNode node1, DataFlow2::PathNode node2
) {
conf.isSourcePair(node1.getNode(), node1.getState(), node2.getNode(), node2.getState()) and
conf.isSourcePair(node1.getNode(), _, node2.getNode(), _) and
node1 = source1 and
node2 = source2
or
@@ -293,7 +213,7 @@ module ProductFlow {
) {
exists(DataFlow::PathNode mid1, DataFlow2::PathNode mid2 |
reachableInterprocEntry(conf, source1, source2, mid1, mid2) and
conf.isSinkPair(sink1.getNode(), sink1.getState(), sink2.getNode(), sink2.getState()) and
conf.isSinkPair(sink1.getNode(), _, sink2.getNode(), _) and
localPathStep1*(mid1, sink1) and
localPathStep2*(mid2, sink2)
)

View File

@@ -915,57 +915,18 @@ private module Cached {
TDataFlowCallNone() or
TDataFlowCallSome(DataFlowCall call)
cached
newtype TParamNodeOption =
TParamNodeNone() or
TParamNodeSome(ParamNode p)
cached
newtype TReturnCtx =
TReturnCtxNone() or
TReturnCtxNoFlowThrough() or
TReturnCtxMaybeFlowThrough(ReturnPosition pos)
cached
newtype TTypedContentApprox =
MkTypedContentApprox(ContentApprox c, DataFlowType t) {
exists(Content cont |
c = getContentApprox(cont) and
store(_, cont, _, _, t)
)
}
cached
newtype TTypedContent = MkTypedContent(Content c, DataFlowType t) { store(_, c, _, _, t) }
cached
TypedContent getATypedContent(TypedContentApprox c) {
exists(ContentApprox cls, DataFlowType t, Content cont |
c = MkTypedContentApprox(cls, pragma[only_bind_into](t)) and
result = MkTypedContent(cont, pragma[only_bind_into](t)) and
cls = getContentApprox(cont)
)
}
cached
newtype TAccessPathFront =
TFrontNil(DataFlowType t) or
TFrontHead(TypedContent tc)
cached
newtype TApproxAccessPathFront =
TApproxFrontNil(DataFlowType t) or
TApproxFrontHead(TypedContentApprox tc)
cached
newtype TAccessPathFrontOption =
TAccessPathFrontNone() or
TAccessPathFrontSome(AccessPathFront apf)
cached
newtype TApproxAccessPathFrontOption =
TApproxAccessPathFrontNone() or
TApproxAccessPathFrontSome(ApproxAccessPathFront apf)
}
/**
@@ -1343,113 +1304,6 @@ class DataFlowCallOption extends TDataFlowCallOption {
}
}
/** An optional `ParamNode`. */
class ParamNodeOption extends TParamNodeOption {
string toString() {
this = TParamNodeNone() and
result = "(none)"
or
exists(ParamNode p |
this = TParamNodeSome(p) and
result = p.toString()
)
}
}
/**
* A return context used to calculate flow summaries in reverse flow.
*
* The possible values are:
*
* - `TReturnCtxNone()`: no return flow.
* - `TReturnCtxNoFlowThrough()`: return flow, but flow through is not possible.
* - `TReturnCtxMaybeFlowThrough(ReturnPosition pos)`: return flow, of kind `pos`, and
* flow through may be possible.
*/
class ReturnCtx extends TReturnCtx {
string toString() {
this = TReturnCtxNone() and
result = "(none)"
or
this = TReturnCtxNoFlowThrough() and
result = "(no flow through)"
or
exists(ReturnPosition pos |
this = TReturnCtxMaybeFlowThrough(pos) and
result = pos.toString()
)
}
}
/** An approximated `Content` tagged with the type of a containing object. */
class TypedContentApprox extends MkTypedContentApprox {
private ContentApprox c;
private DataFlowType t;
TypedContentApprox() { this = MkTypedContentApprox(c, t) }
/** Gets a typed content approximated by this value. */
TypedContent getATypedContent() { result = getATypedContent(this) }
/** Gets the container type. */
DataFlowType getContainerType() { result = t }
/** Gets a textual representation of this approximated content. */
string toString() { result = c.toString() }
}
/**
* The front of an approximated access path. This is either a head or a nil.
*/
abstract class ApproxAccessPathFront extends TApproxAccessPathFront {
abstract string toString();
abstract DataFlowType getType();
abstract boolean toBoolNonEmpty();
pragma[nomagic]
TypedContent getAHead() {
exists(TypedContentApprox cont |
this = TApproxFrontHead(cont) and
result = cont.getATypedContent()
)
}
}
class ApproxAccessPathFrontNil extends ApproxAccessPathFront, TApproxFrontNil {
private DataFlowType t;
ApproxAccessPathFrontNil() { this = TApproxFrontNil(t) }
override string toString() { result = ppReprType(t) }
override DataFlowType getType() { result = t }
override boolean toBoolNonEmpty() { result = false }
}
class ApproxAccessPathFrontHead extends ApproxAccessPathFront, TApproxFrontHead {
private TypedContentApprox tc;
ApproxAccessPathFrontHead() { this = TApproxFrontHead(tc) }
override string toString() { result = tc.toString() }
override DataFlowType getType() { result = tc.getContainerType() }
override boolean toBoolNonEmpty() { result = true }
}
/** An optional approximated access path front. */
class ApproxAccessPathFrontOption extends TApproxAccessPathFrontOption {
string toString() {
this = TApproxAccessPathFrontNone() and result = "<none>"
or
this = TApproxAccessPathFrontSome(any(ApproxAccessPathFront apf | result = apf.toString()))
}
}
/** A `Content` tagged with the type of a containing object. */
class TypedContent extends MkTypedContent {
private Content c;
@@ -1482,7 +1336,7 @@ abstract class AccessPathFront extends TAccessPathFront {
abstract DataFlowType getType();
abstract ApproxAccessPathFront toApprox();
abstract boolean toBoolNonEmpty();
TypedContent getHead() { this = TFrontHead(result) }
}
@@ -1496,7 +1350,7 @@ class AccessPathFrontNil extends AccessPathFront, TFrontNil {
override DataFlowType getType() { result = t }
override ApproxAccessPathFront toApprox() { result = TApproxFrontNil(t) }
override boolean toBoolNonEmpty() { result = false }
}
class AccessPathFrontHead extends AccessPathFront, TFrontHead {
@@ -1508,7 +1362,7 @@ class AccessPathFrontHead extends AccessPathFront, TFrontHead {
override DataFlowType getType() { result = tc.getContainerType() }
override ApproxAccessPathFront toApprox() { result.getAHead() = tc }
override boolean toBoolNonEmpty() { result = true }
}
/** An optional access path front. */

View File

@@ -45,16 +45,6 @@ module Consistency {
) {
none()
}
/** Holds if `(c, pos, p)` should be excluded from the consistency test `uniqueParameterNodeAtPosition`. */
predicate uniqueParameterNodeAtPositionExclude(DataFlowCallable c, ParameterPosition pos, Node p) {
none()
}
/** Holds if `(c, pos, p)` should be excluded from the consistency test `uniqueParameterNodePosition`. */
predicate uniqueParameterNodePositionExclude(DataFlowCallable c, ParameterPosition pos, Node p) {
none()
}
}
private class RelevantNode extends Node {
@@ -111,7 +101,9 @@ module Consistency {
exists(int c |
c =
strictcount(Node n |
not n.hasLocationInfo(_, _, _, _, _) and
not exists(string filepath, int startline, int startcolumn, int endline, int endcolumn |
n.hasLocationInfo(filepath, startline, startcolumn, endline, endcolumn)
) and
not any(ConsistencyConfiguration conf).missingLocationExclude(n)
) and
msg = "Nodes without location: " + c
@@ -144,18 +136,6 @@ module Consistency {
msg = "Local flow step does not preserve enclosing callable."
}
query predicate readStepIsLocal(Node n1, Node n2, string msg) {
readStep(n1, _, n2) and
nodeGetEnclosingCallable(n1) != nodeGetEnclosingCallable(n2) and
msg = "Read step does not preserve enclosing callable."
}
query predicate storeStepIsLocal(Node n1, Node n2, string msg) {
storeStep(n1, _, n2) and
nodeGetEnclosingCallable(n1) != nodeGetEnclosingCallable(n2) and
msg = "Store step does not preserve enclosing callable."
}
private DataFlowType typeRepr() { result = getNodeType(_) }
query predicate compatibleTypesReflexive(DataFlowType t, string msg) {
@@ -252,27 +232,4 @@ module Consistency {
not callable = viableCallable(call) and
not any(ConsistencyConfiguration c).viableImplInCallContextTooLargeExclude(call, ctx, callable)
}
query predicate uniqueParameterNodeAtPosition(
DataFlowCallable c, ParameterPosition pos, Node p, string msg
) {
not any(ConsistencyConfiguration conf).uniqueParameterNodeAtPositionExclude(c, pos, p) and
isParameterNode(p, c, pos) and
not exists(unique(Node p0 | isParameterNode(p0, c, pos))) and
msg = "Parameters with overlapping positions."
}
query predicate uniqueParameterNodePosition(
DataFlowCallable c, ParameterPosition pos, Node p, string msg
) {
not any(ConsistencyConfiguration conf).uniqueParameterNodePositionExclude(c, pos, p) and
isParameterNode(p, c, pos) and
not exists(unique(ParameterPosition pos0 | isParameterNode(p, c, pos0))) and
msg = "Parameter node with multiple positions."
}
query predicate uniqueContentApprox(Content c, string msg) {
not exists(unique(ContentApprox approx | approx = getContentApprox(c))) and
msg = "Non-unique content approximation."
}
}

View File

@@ -137,7 +137,7 @@ private newtype TReturnKind =
exists(IndirectReturnNode return, ReturnIndirectionInstruction returnInd |
returnInd.hasIndex(argumentIndex) and
return.getAddressOperand() = returnInd.getSourceAddressOperand() and
indirectionIndex = return.getIndirectionIndex()
indirectionIndex = return.getIndirectionIndex() - 1 // We subtract one because the return loads the value.
)
}
@@ -197,7 +197,7 @@ class ReturnIndirectionNode extends IndirectReturnNode, ReturnNode {
exists(int argumentIndex, ReturnIndirectionInstruction returnInd |
returnInd.hasIndex(argumentIndex) and
this.getAddressOperand() = returnInd.getSourceAddressOperand() and
result = TIndirectReturnKind(argumentIndex, this.getIndirectionIndex()) and
result = TIndirectReturnKind(argumentIndex, this.getIndirectionIndex() - 1) and
hasNonInitializeParameterDef(returnInd.getIRVariable())
)
or
@@ -241,7 +241,7 @@ private Instruction getANonConversionUse(Operand operand) {
/**
* Gets the operand that represents the first use of the value of `call` following
* a sequence of conversion-like instructions.
* a sequnce of conversion-like instructions.
*/
predicate operandForfullyConvertedCall(Operand operand, CallInstruction call) {
exists(getANonConversionUse(operand)) and
@@ -254,7 +254,7 @@ predicate operandForfullyConvertedCall(Operand operand, CallInstruction call) {
/**
* Gets the instruction that represents the first use of the value of `call` following
* a sequence of conversion-like instructions.
* a sequnce of conversion-like instructions.
*
* This predicate only holds if there is no suitable operand (i.e., no operand of a non-
* conversion instruction) to use to represent the value of `call` after conversions.
@@ -365,7 +365,7 @@ predicate jumpStep(Node n1, Node n2) {
predicate storeStep(Node node1, Content c, PostFieldUpdateNode node2) {
exists(int indirectionIndex1, int numberOfLoads, StoreInstruction store |
nodeHasInstruction(node1, store, pragma[only_bind_into](indirectionIndex1)) and
node2.getIndirectionIndex() = 1 and
node2.getIndirectionIndex() = 0 and
numberOfLoadsFromOperand(node2.getFieldAddress(), store.getDestinationAddressOperand(),
numberOfLoads)
|
@@ -465,20 +465,20 @@ predicate clearsContent(Node n, Content c) {
predicate expectsContent(Node n, ContentSet c) { none() }
/** Gets the type of `n` used for type pruning. */
DataFlowType getNodeType(Node n) {
IRType getNodeType(Node n) {
suppressUnusedNode(n) and
result instanceof VoidType // stub implementation
result instanceof IRVoidType // stub implementation
}
/** Gets a string representation of a type returned by `getNodeType`. */
string ppReprType(DataFlowType t) { none() } // stub implementation
string ppReprType(IRType t) { none() } // stub implementation
/**
* Holds if `t1` and `t2` are compatible, that is, whether data can flow from
* a node of type `t1` to a node of type `t2`.
*/
pragma[inline]
predicate compatibleTypes(DataFlowType t1, DataFlowType t2) {
predicate compatibleTypes(IRType t1, IRType t2) {
any() // stub implementation
}
@@ -502,7 +502,7 @@ class DataFlowCallable = Cpp::Declaration;
class DataFlowExpr = Expr;
class DataFlowType = Type;
class DataFlowType = IRType;
/** A function call relevant for data flow. */
class DataFlowCall extends CallInstruction {
@@ -551,13 +551,6 @@ predicate additionalLambdaFlowStep(Node nodeFrom, Node nodeTo, boolean preserves
*/
predicate allowParameterReturnInSelf(ParameterNode p) { none() }
/** An approximated `Content`. */
class ContentApprox = Unit;
/** Gets an approximated value for content `c`. */
pragma[inline]
ContentApprox getContentApprox(Content c) { any() }
private class MyConsistencyConfiguration extends Consistency::ConsistencyConfiguration {
override predicate argHasPostUpdateExclude(ArgumentNode n) {
// The rules for whether an IR argument gets a post-update node are too

View File

@@ -38,12 +38,13 @@ private module Cached {
TVariableNode(Variable var) or
TPostFieldUpdateNode(FieldAddress operand, int indirectionIndex) {
indirectionIndex =
[1 .. Ssa::countIndirectionsForCppType(operand.getObjectAddress().getResultLanguageType())]
[0 .. Ssa::countIndirectionsForCppType(operand.getObjectAddress().getResultLanguageType()) -
1]
} or
TSsaPhiNode(Ssa::PhiNode phi) or
TIndirectArgumentOutNode(ArgumentOperand operand, int indirectionIndex) {
Ssa::isModifiableByCall(operand) and
indirectionIndex = [1 .. Ssa::countIndirectionsForCppType(operand.getLanguageType())]
indirectionIndex = [0 .. Ssa::countIndirectionsForCppType(operand.getLanguageType()) - 1]
} or
TIndirectOperand(Operand op, int indirectionIndex) {
Ssa::hasIndirectOperand(op, indirectionIndex)
@@ -112,7 +113,7 @@ class Node extends TIRDataFlowNode {
Declaration getFunction() { none() } // overridden in subclasses
/** Gets the type of this node. */
DataFlowType getType() { none() } // overridden in subclasses
IRType getType() { none() } // overridden in subclasses
/** Gets the instruction corresponding to this node, if any. */
Instruction asInstruction() { result = this.(InstructionNode).getInstruction() }
@@ -229,13 +230,7 @@ class Node extends TIRDataFlowNode {
Expr asIndirectArgument() { result = this.asIndirectArgument(_) }
/** Gets the positional parameter corresponding to this node, if any. */
Parameter asParameter() { result = this.asParameter(0) }
/**
* Gets the uninitialized local variable corresponding to this node, if
* any.
*/
LocalVariable asUninitialized() { result = this.(UninitializedNode).getLocalVariable() }
Parameter asParameter() { result = asParameter(0) }
/**
* Gets the positional parameter corresponding to the node that represents
@@ -278,7 +273,7 @@ class Node extends TIRDataFlowNode {
/**
* Gets an upper bound on the type of this node.
*/
DataFlowType getTypeBound() { result = this.getType() }
IRType getTypeBound() { result = this.getType() }
/** Gets the location of this element. */
cached
@@ -327,7 +322,7 @@ class InstructionNode extends Node, TInstructionNode {
override Declaration getFunction() { result = instr.getEnclosingFunction() }
override DataFlowType getType() { result = instr.getResultType() }
override IRType getType() { result = instr.getResultIRType() }
final override Location getLocationImpl() { result = instr.getLocation() }
@@ -353,32 +348,13 @@ class OperandNode extends Node, TOperandNode {
override Declaration getFunction() { result = op.getUse().getEnclosingFunction() }
override DataFlowType getType() { result = op.getType() }
override IRType getType() { result = op.getIRType() }
final override Location getLocationImpl() { result = op.getLocation() }
override string toStringImpl() { result = this.getOperand().toString() }
}
/**
* Returns `t`, but stripped of the `n` outermost pointers, references, etc.
*
* For example, `stripPointers(int*&, 2)` is `int` and `stripPointers(int*, 0)` is `int*`.
*/
private Type stripPointers(Type t, int n) {
result = t and n = 0
or
result = stripPointers(t.(PointerType).getBaseType(), n - 1)
or
result = stripPointers(t.(ArrayType).getBaseType(), n - 1)
or
result = stripPointers(t.(ReferenceType).getBaseType(), n - 1)
or
result = stripPointers(t.(PointerToMemberType).getBaseType(), n - 1)
or
result = stripPointers(t.(FunctionPointerIshType).getBaseType(), n - 1)
}
/**
* INTERNAL: do not use.
*
@@ -394,6 +370,8 @@ class PostFieldUpdateNode extends TPostFieldUpdateNode, PartialDefinitionNode {
override Declaration getEnclosingCallable() { result = this.getFunction() }
override IRType getType() { result = fieldAddress.getIRType() }
FieldAddress getFieldAddress() { result = fieldAddress }
Field getUpdatedField() { result = fieldAddress.getField() }
@@ -401,8 +379,10 @@ class PostFieldUpdateNode extends TPostFieldUpdateNode, PartialDefinitionNode {
int getIndirectionIndex() { result = indirectionIndex }
override Node getPreUpdateNode() {
// + 1 because we're storing into an lvalue, and the original node should be the rvalue of
// the same address.
hasOperandAndIndex(result, pragma[only_bind_into](fieldAddress).getObjectAddressOperand(),
indirectionIndex)
indirectionIndex + 1)
}
override Expr getDefinedExpr() {
@@ -431,7 +411,7 @@ class SsaPhiNode extends Node, TSsaPhiNode {
override Declaration getFunction() { result = phi.getBasicBlock().getEnclosingFunction() }
override DataFlowType getType() { result = this.getAnInput().getType() }
override IRType getType() { result instanceof IRVoidType }
final override Location getLocationImpl() { result = phi.getBasicBlock().getLocation() }
@@ -474,6 +454,8 @@ class SideEffectOperandNode extends Node, IndirectOperand {
override Function getFunction() { result = call.getEnclosingFunction() }
override IRType getType() { result instanceof IRVoidType }
Expr getArgument() { result = call.getArgument(argumentIndex).getUnconvertedResultExpression() }
}
@@ -496,6 +478,8 @@ class IndirectParameterNode extends Node, IndirectInstruction {
override Function getFunction() { result = this.getInstruction().getEnclosingFunction() }
override IRType getType() { result instanceof IRVoidType }
override string toStringImpl() {
result = this.getParameter().toString() + " indirection"
or
@@ -520,6 +504,8 @@ class IndirectReturnNode extends IndirectOperand {
Operand getAddressOperand() { result = operand }
override Declaration getEnclosingCallable() { result = this.getFunction() }
override IRType getType() { result instanceof IRVoidType }
}
/**
@@ -550,7 +536,9 @@ class IndirectArgumentOutNode extends Node, TIndirectArgumentOutNode, PostUpdate
override Function getFunction() { result = this.getCallInstruction().getEnclosingFunction() }
override Node getPreUpdateNode() { hasOperandAndIndex(result, operand, indirectionIndex) }
override IRType getType() { result instanceof IRVoidType }
override Node getPreUpdateNode() { hasOperandAndIndex(result, operand, indirectionIndex + 1) }
override string toStringImpl() {
// This string should be unique enough to be helpful but common enough to
@@ -606,38 +594,6 @@ class IndirectReturnOutNode extends Node {
int getIndirectionIndex() { result = indirectionIndex }
}
private PointerType getGLValueType(Type t, int indirectionIndex) {
result.getBaseType() = stripPointers(t, indirectionIndex - 1)
}
bindingset[isGLValue]
private DataFlowType getTypeImpl(Type t, int indirectionIndex, boolean isGLValue) {
if isGLValue = true
then
result = getGLValueType(t, indirectionIndex)
or
// Ideally, the above case would cover all glvalue cases. However, consider the case where
// the database consists only of:
// ```
// void test() {
// int* x;
// x = nullptr;
// }
// ```
// and we want to compute the type of `*x` in the assignment `x = nullptr`. Here, `x` is an lvalue
// of type int* (which morally is an int**). So when we call `getTypeImpl` it will be with the
// parameters:
// - t = int*
// - indirectionIndex = 1 (when we want to model the dataflow node corresponding to *x)
// - isGLValue = true
// In this case, `getTypeImpl(t, indirectionIndex, isGLValue)` should give back `int**`. In this
// case, however, `int**` does not exist in the database. So instead we return int* (which is
// wrong, but at least we have a type).
not exists(getGLValueType(t, indirectionIndex)) and
result = stripPointers(t, indirectionIndex - 1)
else result = stripPointers(t, indirectionIndex)
}
/**
* INTERNAL: Do not use.
*
@@ -659,11 +615,7 @@ class IndirectOperand extends Node, TIndirectOperand {
override Declaration getEnclosingCallable() { result = this.getFunction() }
override DataFlowType getType() {
exists(boolean isGLValue | if operand.isGLValue() then isGLValue = true else isGLValue = false |
result = getTypeImpl(operand.getType().getUnspecifiedType(), indirectionIndex, isGLValue)
)
}
override IRType getType() { result = this.getOperand().getIRType() }
final override Location getLocationImpl() { result = this.getOperand().getLocation() }
@@ -672,25 +624,6 @@ class IndirectOperand extends Node, TIndirectOperand {
}
}
/**
* The value of an uninitialized local variable, viewed as a node in a data
* flow graph.
*/
class UninitializedNode extends Node {
LocalVariable v;
UninitializedNode() {
exists(Ssa::Def def |
def.getDefiningInstruction() instanceof UninitializedInstruction and
Ssa::nodeToDefOrUse(this, def) and
v = def.getSourceVariable().getBaseVariable().(Ssa::BaseIRVariable).getIRVariable().getAst()
)
}
/** Gets the uninitialized local variable corresponding to this node. */
LocalVariable getLocalVariable() { result = v }
}
/**
* INTERNAL: Do not use.
*
@@ -712,11 +645,7 @@ class IndirectInstruction extends Node, TIndirectInstruction {
override Declaration getEnclosingCallable() { result = this.getFunction() }
override DataFlowType getType() {
exists(boolean isGLValue | if instr.isGLValue() then isGLValue = true else isGLValue = false |
result = getTypeImpl(instr.getResultType().getUnspecifiedType(), indirectionIndex, isGLValue)
)
}
override IRType getType() { result = this.getInstruction().getResultIRType() }
final override Location getLocationImpl() { result = this.getInstruction().getLocation() }
@@ -746,7 +675,7 @@ predicate exprNodeShouldBeOperand(Node node, Expr e) {
/**
* Holds if `load` is a `LoadInstruction` that is the result of evaluating `e`
* and `node` is an `IndirectOperandNode` that should map `node.asExpr()` to `e`.
* and `node` is an `IndirctOperandNode` that should map `node.asExpr()` to `e`.
*
* We map `e` to `node.asExpr()` when `node` semantically represents the
* same value as `load`. A subsequent flow step will flow `node` to
@@ -930,8 +859,6 @@ abstract class PostUpdateNode extends Node {
* Gets the node before the state update.
*/
abstract Node getPreUpdateNode();
final override DataFlowType getType() { result = this.getPreUpdateNode().getType() }
}
/**
@@ -995,7 +922,7 @@ class VariableNode extends Node, TVariableNode {
result = v
}
override DataFlowType getType() { result = v.getType() }
override IRType getType() { result.getCanonicalLanguageType().hasUnspecifiedType(v.getType(), _) }
final override Location getLocationImpl() { result = v.getLocation() }
@@ -1148,7 +1075,7 @@ predicate simpleLocalFlowStep(Node nodeFrom, Node nodeTo) {
store.getDestinationAddressOperand() = address
)
or
Ssa::outNodeHasAddressAndIndex(nodeFrom, address, indirectionIndex)
Ssa::outNodeHasAddressAndIndex(nodeFrom, address, indirectionIndex - 1)
)
}

View File

@@ -41,7 +41,7 @@ Node callOutput(CallInstruction call, FunctionOutput output) {
// The side effect of a call on the value pointed to by an argument or qualifier
exists(int index, int indirectionIndex |
result.(IndirectArgumentOutNode).getArgumentIndex() = index and
result.(IndirectArgumentOutNode).getIndirectionIndex() = indirectionIndex and
result.(IndirectArgumentOutNode).getIndirectionIndex() + 1 = indirectionIndex and
result.(IndirectArgumentOutNode).getCallInstruction() = call and
output.isParameterDerefOrQualifierObject(index, indirectionIndex)
)

View File

@@ -100,7 +100,7 @@ private string getNodeProperty(DataFlow::Node node, string key) {
or
// Is there partial flow from a source to this node?
// This property will only be emitted if partial flow is enabled by overriding
// `DataFlow::Configuration::explorationLimit()`.
// `DataFlow::Configration::explorationLimit()`.
key = "pflow" and
result =
strictconcat(DataFlow::PartialPathNode sourceNode, DataFlow::PartialPathNode destNode, int dist,

View File

@@ -11,9 +11,7 @@ private import DataFlowUtil
* corresponding `(Indirect)OperandNode`.
*/
predicate ignoreOperand(Operand operand) {
operand = any(Instruction instr | ignoreInstruction(instr)).getAnOperand() or
operand = any(Instruction instr | ignoreInstruction(instr)).getAUse() or
operand instanceof MemoryOperand
operand = any(Instruction instr | ignoreInstruction(instr)).getAnOperand()
}
/**

View File

@@ -36,7 +36,7 @@ private module SourceVariables {
override string toString() { result = var.toString() }
override DataFlowType getType() { result = var.getType() }
override DataFlowType getType() { result = var.getIRType() }
}
class BaseCallVariable extends BaseSourceVariable, TBaseCallVariable {
@@ -48,7 +48,7 @@ private module SourceVariables {
override string toString() { result = call.toString() }
override DataFlowType getType() { result = call.getResultType() }
override DataFlowType getType() { result = call.getResultIRType() }
}
private newtype TSourceVariable =

View File

@@ -218,7 +218,7 @@ private predicate allocation(Instruction array, Length length, int delta) {
length.(VNLength).getInstruction().getConvertedResultExpression() = lengthExpr
)
or
not deconstructMallocSizeExpr(alloc.getSizeExpr(), _, _) and
not exists(int d | deconstructMallocSizeExpr(alloc.getSizeExpr(), _, d)) and
length.(VNLength).getInstruction().getConvertedResultExpression() = alloc.getSizeExpr() and
delta = 0
)

View File

@@ -543,7 +543,9 @@ private predicate boundedPhiCand(
PhiInstruction phi, boolean upper, Bound b, int delta, boolean fromBackEdge, int origdelta,
Reason reason
) {
boundedPhiInp(phi, _, b, delta, upper, fromBackEdge, origdelta, reason)
exists(PhiInputOperand op |
boundedPhiInp(phi, op, b, delta, upper, fromBackEdge, origdelta, reason)
)
}
/**

View File

@@ -292,8 +292,12 @@ module SemanticExprConfig {
final Location getLocation() { result = super.getLocation() }
}
private class ValueNumberBound extends Bound instanceof IRBound::ValueNumberBound {
override string toString() { result = IRBound::ValueNumberBound.super.toString() }
private class ValueNumberBound extends Bound {
IRBound::ValueNumberBound bound;
ValueNumberBound() { bound = this }
override string toString() { result = bound.toString() }
}
predicate zeroBound(Bound bound) { bound instanceof IRBound::ZeroBound }

View File

@@ -1,29 +0,0 @@
private import RangeAnalysisStage
module FloatDelta implements DeltaSig {
class Delta = float;
bindingset[d]
bindingset[result]
float toFloat(Delta d) { result = d }
bindingset[d]
bindingset[result]
int toInt(Delta d) { result = d }
bindingset[n]
bindingset[result]
Delta fromInt(int n) { result = n }
bindingset[f]
Delta fromFloat(float f) {
result =
min(float diff, float res |
diff = (res - f) and res = f.ceil()
or
diff = (f - res) and res = f.floor()
|
res order by diff
)
}
}

View File

@@ -4,338 +4,308 @@
* variable), and `v` is an integer in the range `[0 .. m-1]`.
*/
/*
* The main recursion has base cases in both `ssaModulus` (for guarded reads) and `semExprModulus`
* (for constant values). The most interesting recursive case is `phiModulusRankStep`, which
* handles phi inputs.
*/
private import ModulusAnalysisSpecific::Private
private import experimental.semmle.code.cpp.semantic.Semantic
private import ConstantAnalysis
private import RangeUtils
private import RangeAnalysisStage
module ModulusAnalysis<DeltaSig D, BoundSig<D> Bounds, UtilSig<D> U> {
/**
* Holds if `e + delta` equals `v` at `pos`.
*/
private predicate valueFlowStepSsa(SemSsaVariable v, SemSsaReadPosition pos, SemExpr e, int delta) {
U::semSsaUpdateStep(v, e, D::fromInt(delta)) and pos.hasReadOfVar(v)
or
exists(SemGuard guard, boolean testIsTrue |
pos.hasReadOfVar(v) and
guard = U::semEqFlowCond(v, e, D::fromInt(delta), true, testIsTrue) and
semGuardDirectlyControlsSsaRead(guard, pos, testIsTrue)
)
}
/**
* Holds if `e + delta` equals `v` at `pos`.
*/
private predicate valueFlowStepSsa(SemSsaVariable v, SemSsaReadPosition pos, SemExpr e, int delta) {
semSsaUpdateStep(v, e, delta) and pos.hasReadOfVar(v)
or
exists(SemGuard guard, boolean testIsTrue |
pos.hasReadOfVar(v) and
guard = semEqFlowCond(v, e, delta, true, testIsTrue) and
semGuardDirectlyControlsSsaRead(guard, pos, testIsTrue)
)
}
/**
* Holds if `add` is the addition of `larg` and `rarg`, neither of which are
* `ConstantIntegerExpr`s.
*/
private predicate nonConstAddition(SemExpr add, SemExpr larg, SemExpr rarg) {
exists(SemAddExpr a | a = add |
larg = a.getLeftOperand() and
rarg = a.getRightOperand()
) and
not larg instanceof SemConstantIntegerExpr and
not rarg instanceof SemConstantIntegerExpr
}
/**
* Holds if `add` is the addition of `larg` and `rarg`, neither of which are
* `ConstantIntegerExpr`s.
*/
private predicate nonConstAddition(SemExpr add, SemExpr larg, SemExpr rarg) {
exists(SemAddExpr a | a = add |
larg = a.getLeftOperand() and
rarg = a.getRightOperand()
) and
not larg instanceof SemConstantIntegerExpr and
not rarg instanceof SemConstantIntegerExpr
}
/**
* Holds if `sub` is the subtraction of `larg` and `rarg`, where `rarg` is not
* a `ConstantIntegerExpr`.
*/
private predicate nonConstSubtraction(SemExpr sub, SemExpr larg, SemExpr rarg) {
exists(SemSubExpr s | s = sub |
larg = s.getLeftOperand() and
rarg = s.getRightOperand()
) and
not rarg instanceof SemConstantIntegerExpr
}
/**
* Holds if `sub` is the subtraction of `larg` and `rarg`, where `rarg` is not
* a `ConstantIntegerExpr`.
*/
private predicate nonConstSubtraction(SemExpr sub, SemExpr larg, SemExpr rarg) {
exists(SemSubExpr s | s = sub |
larg = s.getLeftOperand() and
rarg = s.getRightOperand()
) and
not rarg instanceof SemConstantIntegerExpr
}
/** Gets an expression that is the remainder modulo `mod` of `arg`. */
private SemExpr modExpr(SemExpr arg, int mod) {
exists(SemRemExpr rem |
result = rem and
arg = rem.getLeftOperand() and
rem.getRightOperand().(SemConstantIntegerExpr).getIntValue() = mod and
mod >= 2
)
or
exists(SemConstantIntegerExpr c |
mod = 2.pow([1 .. 30]) and
c.getIntValue() = mod - 1 and
result.(SemBitAndExpr).hasOperands(arg, c)
)
}
/** Gets an expression that is the remainder modulo `mod` of `arg`. */
private SemExpr modExpr(SemExpr arg, int mod) {
exists(SemRemExpr rem |
result = rem and
arg = rem.getLeftOperand() and
rem.getRightOperand().(SemConstantIntegerExpr).getIntValue() = mod and
mod >= 2
)
or
exists(SemConstantIntegerExpr c |
mod = 2.pow([1 .. 30]) and
c.getIntValue() = mod - 1 and
result.(SemBitAndExpr).hasOperands(arg, c)
)
}
/**
* Gets a guard that tests whether `v` is congruent with `val` modulo `mod` on
* its `testIsTrue` branch.
*/
private SemGuard moduloCheck(SemSsaVariable v, int val, int mod, boolean testIsTrue) {
exists(SemExpr rem, SemConstantIntegerExpr c, int r, boolean polarity |
result.isEquality(rem, c, polarity) and
c.getIntValue() = r and
rem = modExpr(v.getAUse(), mod) and
(
testIsTrue = polarity and val = r
or
testIsTrue = polarity.booleanNot() and
mod = 2 and
val = 1 - r and
(r = 0 or r = 1)
)
)
}
/**
* Holds if a guard ensures that `v` at `pos` is congruent with `val` modulo `mod`.
*/
private predicate moduloGuardedRead(SemSsaVariable v, SemSsaReadPosition pos, int val, int mod) {
exists(SemGuard guard, boolean testIsTrue |
pos.hasReadOfVar(v) and
guard = moduloCheck(v, val, mod, testIsTrue) and
semGuardControlsSsaRead(guard, pos, testIsTrue)
)
}
/** Holds if `factor` is a power of 2 that divides `mask`. */
bindingset[mask]
private predicate andmaskFactor(int mask, int factor) {
mask % factor = 0 and
factor = 2.pow([1 .. 30])
}
/** Holds if `e` is evenly divisible by `factor`. */
private predicate evenlyDivisibleExpr(SemExpr e, int factor) {
exists(SemConstantIntegerExpr c, int k | k = c.getIntValue() |
e.(SemMulExpr).getAnOperand() = c and factor = k.abs() and factor >= 2
/**
* Gets a guard that tests whether `v` is congruent with `val` modulo `mod` on
* its `testIsTrue` branch.
*/
private SemGuard moduloCheck(SemSsaVariable v, int val, int mod, boolean testIsTrue) {
exists(SemExpr rem, SemConstantIntegerExpr c, int r, boolean polarity |
result.isEquality(rem, c, polarity) and
c.getIntValue() = r and
rem = modExpr(v.getAUse(), mod) and
(
testIsTrue = polarity and val = r
or
e.(SemShiftLeftExpr).getRightOperand() = c and factor = 2.pow(k) and k > 0
or
e.(SemBitAndExpr).getAnOperand() = c and factor = max(int f | andmaskFactor(k, f))
testIsTrue = polarity.booleanNot() and
mod = 2 and
val = 1 - r and
(r = 0 or r = 1)
)
}
)
}
/**
* Holds if `rix` is the number of input edges to `phi`.
*/
private predicate maxPhiInputRank(SemSsaPhiNode phi, int rix) {
rix = max(int r | rankedPhiInput(phi, _, _, r))
}
/**
* Holds if a guard ensures that `v` at `pos` is congruent with `val` modulo `mod`.
*/
private predicate moduloGuardedRead(SemSsaVariable v, SemSsaReadPosition pos, int val, int mod) {
exists(SemGuard guard, boolean testIsTrue |
pos.hasReadOfVar(v) and
guard = moduloCheck(v, val, mod, testIsTrue) and
semGuardControlsSsaRead(guard, pos, testIsTrue)
)
}
/**
* Gets the remainder of `val` modulo `mod`.
*
* For `mod = 0` the result equals `val` and for `mod > 1` the result is within
* the range `[0 .. mod-1]`.
*/
bindingset[val, mod]
private int remainder(int val, int mod) {
mod = 0 and result = val
/** Holds if `factor` is a power of 2 that divides `mask`. */
bindingset[mask]
private predicate andmaskFactor(int mask, int factor) {
mask % factor = 0 and
factor = 2.pow([1 .. 30])
}
/** Holds if `e` is evenly divisible by `factor`. */
private predicate evenlyDivisibleExpr(SemExpr e, int factor) {
exists(SemConstantIntegerExpr c, int k | k = c.getIntValue() |
e.(SemMulExpr).getAnOperand() = c and factor = k.abs() and factor >= 2
or
mod > 1 and result = ((val % mod) + mod) % mod
}
/**
* Holds if `inp` is an input to `phi` and equals `phi` modulo `mod` along `edge`.
*/
private predicate phiSelfModulus(
SemSsaPhiNode phi, SemSsaVariable inp, SemSsaReadPositionPhiInputEdge edge, int mod
) {
exists(Bounds::SemSsaBound phibound, int v, int m |
edge.phiInput(phi, inp) and
phibound.getAVariable() = phi and
ssaModulus(inp, edge, phibound, v, m) and
mod = m.gcd(v) and
mod != 1
)
}
/**
* Holds if `b + val` modulo `mod` is a candidate congruence class for `phi`.
*/
private predicate phiModulusInit(SemSsaPhiNode phi, Bounds::SemBound b, int val, int mod) {
exists(SemSsaVariable inp, SemSsaReadPositionPhiInputEdge edge |
edge.phiInput(phi, inp) and
ssaModulus(inp, edge, b, val, mod)
)
}
/**
* Holds if all inputs to `phi` numbered `1` to `rix` are equal to `b + val` modulo `mod`.
*/
pragma[nomagic]
private predicate phiModulusRankStep(
SemSsaPhiNode phi, Bounds::SemBound b, int val, int mod, int rix
) {
/*
* base case. If any phi input is equal to `b + val` modulo `mod`, that's a potential congruence
* class for the phi node.
*/
rix = 0 and
phiModulusInit(phi, b, val, mod)
e.(SemShiftLeftExpr).getRightOperand() = c and factor = 2.pow(k) and k > 0
or
exists(SemSsaVariable inp, SemSsaReadPositionPhiInputEdge edge, int v1, int m1 |
mod != 1 and
val = remainder(v1, mod)
|
/*
* Recursive case. If `inp` = `b + v2` mod `m2`, we combine that with the preceding potential
* congruence class `b + v1` mod `m1`. The result will be the congruence class of `v1` modulo
* the greatest common denominator of `m1`, `m2`, and `v1 - v2`.
*/
e.(SemBitAndExpr).getAnOperand() = c and factor = max(int f | andmaskFactor(k, f))
)
}
exists(int v2, int m2 |
rankedPhiInput(pragma[only_bind_out](phi), inp, edge, rix) and
phiModulusRankStep(phi, b, v1, m1, rix - 1) and
ssaModulus(inp, edge, b, v2, m2) and
mod = m1.gcd(m2).gcd(v1 - v2)
)
or
/*
* Recursive case. If `inp` = `phi` mod `m2`, we combine that with the preceding potential
* congruence class `b + v1` mod `m1`. The result will be a congruence class modulo the greatest
* common denominator of `m1` and `m2`.
*/
/**
* Holds if `rix` is the number of input edges to `phi`.
*/
private predicate maxPhiInputRank(SemSsaPhiNode phi, int rix) {
rix = max(int r | rankedPhiInput(phi, _, _, r))
}
exists(int m2 |
rankedPhiInput(phi, inp, edge, rix) and
phiModulusRankStep(phi, b, v1, m1, rix - 1) and
phiSelfModulus(phi, inp, edge, m2) and
mod = m1.gcd(m2)
)
/**
* Gets the remainder of `val` modulo `mod`.
*
* For `mod = 0` the result equals `val` and for `mod > 1` the result is within
* the range `[0 .. mod-1]`.
*/
bindingset[val, mod]
private int remainder(int val, int mod) {
mod = 0 and result = val
or
mod > 1 and result = ((val % mod) + mod) % mod
}
/**
* Holds if `inp` is an input to `phi` and equals `phi` modulo `mod` along `edge`.
*/
private predicate phiSelfModulus(
SemSsaPhiNode phi, SemSsaVariable inp, SemSsaReadPositionPhiInputEdge edge, int mod
) {
exists(SemSsaBound phibound, int v, int m |
edge.phiInput(phi, inp) and
phibound.getAVariable() = phi and
ssaModulus(inp, edge, phibound, v, m) and
mod = m.gcd(v) and
mod != 1
)
}
/**
* Holds if `b + val` modulo `mod` is a candidate congruence class for `phi`.
*/
private predicate phiModulusInit(SemSsaPhiNode phi, SemBound b, int val, int mod) {
exists(SemSsaVariable inp, SemSsaReadPositionPhiInputEdge edge |
edge.phiInput(phi, inp) and
ssaModulus(inp, edge, b, val, mod)
)
}
/**
* Holds if all inputs to `phi` numbered `1` to `rix` are equal to `b + val` modulo `mod`.
*/
pragma[nomagic]
private predicate phiModulusRankStep(SemSsaPhiNode phi, SemBound b, int val, int mod, int rix) {
rix = 0 and
phiModulusInit(phi, b, val, mod)
or
exists(SemSsaVariable inp, SemSsaReadPositionPhiInputEdge edge, int v1, int m1 |
mod != 1 and
val = remainder(v1, mod)
|
exists(int v2, int m2 |
rankedPhiInput(pragma[only_bind_out](phi), inp, edge, rix) and
phiModulusRankStep(phi, b, v1, m1, rix - 1) and
ssaModulus(inp, edge, b, v2, m2) and
mod = m1.gcd(m2).gcd(v1 - v2)
)
}
/**
* Holds if `phi` is equal to `b + val` modulo `mod`.
*/
private predicate phiModulus(SemSsaPhiNode phi, Bounds::SemBound b, int val, int mod) {
exists(int r |
maxPhiInputRank(phi, r) and
phiModulusRankStep(phi, b, val, mod, r)
or
exists(int m2 |
rankedPhiInput(phi, inp, edge, rix) and
phiModulusRankStep(phi, b, v1, m1, rix - 1) and
phiSelfModulus(phi, inp, edge, m2) and
mod = m1.gcd(m2)
)
}
)
}
/**
* Holds if `v` at `pos` is equal to `b + val` modulo `mod`.
*/
private predicate ssaModulus(
SemSsaVariable v, SemSsaReadPosition pos, Bounds::SemBound b, int val, int mod
) {
phiModulus(v, b, val, mod) and pos.hasReadOfVar(v)
/**
* Holds if `phi` is equal to `b + val` modulo `mod`.
*/
private predicate phiModulus(SemSsaPhiNode phi, SemBound b, int val, int mod) {
exists(int r |
maxPhiInputRank(phi, r) and
phiModulusRankStep(phi, b, val, mod, r)
)
}
/**
* Holds if `v` at `pos` is equal to `b + val` modulo `mod`.
*/
private predicate ssaModulus(SemSsaVariable v, SemSsaReadPosition pos, SemBound b, int val, int mod) {
phiModulus(v, b, val, mod) and pos.hasReadOfVar(v)
or
b.(SemSsaBound).getAVariable() = v and pos.hasReadOfVar(v) and val = 0 and mod = 0
or
exists(SemExpr e, int val0, int delta |
semExprModulus(e, b, val0, mod) and
valueFlowStepSsa(v, pos, e, delta) and
val = remainder(val0 + delta, mod)
)
or
moduloGuardedRead(v, pos, val, mod) and b instanceof SemZeroBound
}
/**
* Holds if `e` is equal to `b + val` modulo `mod`.
*
* There are two cases for the modulus:
* - `mod = 0`: The equality `e = b + val` is an ordinary equality.
* - `mod > 1`: `val` lies within the range `[0 .. mod-1]`.
*/
cached
predicate semExprModulus(SemExpr e, SemBound b, int val, int mod) {
not ignoreExprModulus(e) and
(
e = b.getExpr(val) and mod = 0
or
b.(Bounds::SemSsaBound).getAVariable() = v and pos.hasReadOfVar(v) and val = 0 and mod = 0
evenlyDivisibleExpr(e, mod) and
val = 0 and
b instanceof SemZeroBound
or
exists(SemExpr e, int val0, int delta |
semExprModulus(e, b, val0, mod) and
valueFlowStepSsa(v, pos, e, delta) and
exists(SemSsaVariable v, SemSsaReadPositionBlock bb |
ssaModulus(v, bb, b, val, mod) and
e = v.getAUse() and
bb.getAnExpr() = e
)
or
exists(SemExpr mid, int val0, int delta |
semExprModulus(mid, b, val0, mod) and
semValueFlowStep(e, mid, delta) and
val = remainder(val0 + delta, mod)
)
or
moduloGuardedRead(v, pos, val, mod) and b instanceof Bounds::SemZeroBound
}
/**
* Holds if `e` is equal to `b + val` modulo `mod`.
*
* There are two cases for the modulus:
* - `mod = 0`: The equality `e = b + val` is an ordinary equality.
* - `mod > 1`: `val` lies within the range `[0 .. mod-1]`.
*/
cached
predicate semExprModulus(SemExpr e, Bounds::SemBound b, int val, int mod) {
not ignoreExprModulus(e) and
(
e = b.getExpr(D::fromInt(val)) and mod = 0
or
evenlyDivisibleExpr(e, mod) and
val = 0 and
b instanceof Bounds::SemZeroBound
or
exists(SemSsaVariable v, SemSsaReadPositionBlock bb |
ssaModulus(v, bb, b, val, mod) and
e = v.getAUse() and
bb.getAnExpr() = e
)
or
exists(SemExpr mid, int val0, int delta |
semExprModulus(mid, b, val0, mod) and
U::semValueFlowStep(e, mid, D::fromInt(delta)) and
val = remainder(val0 + delta, mod)
)
or
exists(SemConditionalExpr cond, int v1, int v2, int m1, int m2 |
cond = e and
condExprBranchModulus(cond, true, b, v1, m1) and
condExprBranchModulus(cond, false, b, v2, m2) and
mod = m1.gcd(m2).gcd(v1 - v2) and
mod != 1 and
val = remainder(v1, mod)
)
or
exists(Bounds::SemBound b1, Bounds::SemBound b2, int v1, int v2, int m1, int m2 |
addModulus(e, true, b1, v1, m1) and
addModulus(e, false, b2, v2, m2) and
mod = m1.gcd(m2) and
mod != 1 and
val = remainder(v1 + v2, mod)
|
b = b1 and b2 instanceof Bounds::SemZeroBound
or
b = b2 and b1 instanceof Bounds::SemZeroBound
)
or
exists(int v1, int v2, int m1, int m2 |
subModulus(e, true, b, v1, m1) and
subModulus(e, false, any(Bounds::SemZeroBound zb), v2, m2) and
mod = m1.gcd(m2) and
mod != 1 and
val = remainder(v1 - v2, mod)
)
exists(SemConditionalExpr cond, int v1, int v2, int m1, int m2 |
cond = e and
condExprBranchModulus(cond, true, b, v1, m1) and
condExprBranchModulus(cond, false, b, v2, m2) and
mod = m1.gcd(m2).gcd(v1 - v2) and
mod != 1 and
val = remainder(v1, mod)
)
}
private predicate condExprBranchModulus(
SemConditionalExpr cond, boolean branch, Bounds::SemBound b, int val, int mod
) {
semExprModulus(cond.getBranchExpr(branch), b, val, mod)
}
private predicate addModulus(SemExpr add, boolean isLeft, Bounds::SemBound b, int val, int mod) {
exists(SemExpr larg, SemExpr rarg | nonConstAddition(add, larg, rarg) |
semExprModulus(larg, b, val, mod) and isLeft = true
or
exists(SemBound b1, SemBound b2, int v1, int v2, int m1, int m2 |
addModulus(e, true, b1, v1, m1) and
addModulus(e, false, b2, v2, m2) and
mod = m1.gcd(m2) and
mod != 1 and
val = remainder(v1 + v2, mod)
|
b = b1 and b2 instanceof SemZeroBound
or
semExprModulus(rarg, b, val, mod) and isLeft = false
b = b2 and b1 instanceof SemZeroBound
)
}
private predicate subModulus(SemExpr sub, boolean isLeft, Bounds::SemBound b, int val, int mod) {
exists(SemExpr larg, SemExpr rarg | nonConstSubtraction(sub, larg, rarg) |
semExprModulus(larg, b, val, mod) and isLeft = true
or
semExprModulus(rarg, b, val, mod) and isLeft = false
or
exists(int v1, int v2, int m1, int m2 |
subModulus(e, true, b, v1, m1) and
subModulus(e, false, any(SemZeroBound zb), v2, m2) and
mod = m1.gcd(m2) and
mod != 1 and
val = remainder(v1 - v2, mod)
)
)
}
private predicate condExprBranchModulus(
SemConditionalExpr cond, boolean branch, SemBound b, int val, int mod
) {
semExprModulus(cond.getBranchExpr(branch), b, val, mod)
}
private predicate addModulus(SemExpr add, boolean isLeft, SemBound b, int val, int mod) {
exists(SemExpr larg, SemExpr rarg | nonConstAddition(add, larg, rarg) |
semExprModulus(larg, b, val, mod) and isLeft = true
or
semExprModulus(rarg, b, val, mod) and isLeft = false
)
}
private predicate subModulus(SemExpr sub, boolean isLeft, SemBound b, int val, int mod) {
exists(SemExpr larg, SemExpr rarg | nonConstSubtraction(sub, larg, rarg) |
semExprModulus(larg, b, val, mod) and isLeft = true
or
semExprModulus(rarg, b, val, mod) and isLeft = false
)
}
/**
* Holds if `inp` is an input to `phi` along `edge` and this input has index `r`
* in an arbitrary 1-based numbering of the input edges to `phi`.
*/
private predicate rankedPhiInput(
SemSsaPhiNode phi, SemSsaVariable inp, SemSsaReadPositionPhiInputEdge edge, int r
) {
edge.phiInput(phi, inp) and
edge =
rank[r](SemSsaReadPositionPhiInputEdge e |
e.phiInput(phi, _)
|
e order by e.getOrigBlock().getUniqueId()
)
}
/**
* Holds if `inp` is an input to `phi` along `edge` and this input has index `r`
* in an arbitrary 1-based numbering of the input edges to `phi`.
*/
private predicate rankedPhiInput(
SemSsaPhiNode phi, SemSsaVariable inp, SemSsaReadPositionPhiInputEdge edge, int r
) {
edge.phiInput(phi, inp) and
edge =
rank[r](SemSsaReadPositionPhiInputEdge e |
e.phiInput(phi, _)
|
e order by e.getOrigBlock().getUniqueId()
)
}
}

View File

@@ -1,24 +1,832 @@
private import RangeAnalysisStage
private import RangeAnalysisSpecific
private import experimental.semmle.code.cpp.semantic.analysis.FloatDelta
/**
* Provides classes and predicates for range analysis.
*
* An inferred bound can either be a specific integer, the abstract value of an
* SSA variable, or the abstract value of an interesting expression. The latter
* category includes array lengths that are not SSA variables.
*
* If an inferred bound relies directly on a condition, then this condition is
* reported as the reason for the bound.
*/
/*
* This library tackles range analysis as a flow problem. Consider e.g.:
* ```
* len = arr.length;
* if (x < len) { ... y = x-1; ... y ... }
* ```
* In this case we would like to infer `y <= arr.length - 2`, and this is
* accomplished by tracking the bound through a sequence of steps:
* ```
* arr.length --> len = .. --> x < len --> x-1 --> y = .. --> y
* ```
*
* In its simplest form the step relation `E1 --> E2` relates two expressions
* such that `E1 <= B` implies `E2 <= B` for any `B` (with a second separate
* step relation handling lower bounds). Examples of such steps include
* assignments `E2 = E1` and conditions `x <= E1` where `E2` is a use of `x`
* guarded by the condition.
*
* In order to handle subtractions and additions with constants, and strict
* comparisons, the step relation is augmented with an integer delta. With this
* generalization `E1 --(delta)--> E2` relates two expressions and an integer
* such that `E1 <= B` implies `E2 <= B + delta` for any `B`. This corresponds
* to the predicate `boundFlowStep`.
*
* The complete range analysis is then implemented as the transitive closure of
* the step relation summing the deltas along the way. If `E1` transitively
* steps to `E2`, `delta` is the sum of deltas along the path, and `B` is an
* interesting bound equal to the value of `E1` then `E2 <= B + delta`. This
* corresponds to the predicate `bounded`.
*
* Phi nodes need a little bit of extra handling. Consider `x0 = phi(x1, x2)`.
* There are essentially two cases:
* - If `x1 <= B + d1` and `x2 <= B + d2` then `x0 <= B + max(d1,d2)`.
* - If `x1 <= B + d1` and `x2 <= x0 + d2` with `d2 <= 0` then `x0 <= B + d1`.
* The first case is for whenever a bound can be proven without taking looping
* into account. The second case is relevant when `x2` comes from a back-edge
* where we can prove that the variable has been non-increasing through the
* loop-iteration as this means that any upper bound that holds prior to the
* loop also holds for the variable during the loop.
* This generalizes to a phi node with `n` inputs, so if
* `x0 = phi(x1, ..., xn)` and `xi <= B + delta` for one of the inputs, then we
* also have `x0 <= B + delta` if we can prove either:
* - `xj <= B + d` with `d <= delta` or
* - `xj <= x0 + d` with `d <= 0`
* for each input `xj`.
*
* As all inferred bounds can be related directly to a path in the source code
* the only source of non-termination is if successive redundant (and thereby
* increasingly worse) bounds are calculated along a loop in the source code.
* We prevent this by weakening the bound to a small finite set of bounds when
* a path follows a second back-edge (we postpone weakening till the second
* back-edge as a precise bound might require traversing a loop once).
*/
private import RangeAnalysisSpecific as Specific
private import RangeUtils
private import experimental.semmle.code.cpp.semantic.SemanticBound as SemanticBound
private import SignAnalysisCommon
private import ModulusAnalysis
private import experimental.semmle.code.cpp.semantic.Semantic
private import ConstantAnalysis
module Bounds implements BoundSig<FloatDelta> {
class SemBound instanceof SemanticBound::SemBound {
string toString() { result = super.toString() }
SemExpr getExpr(float delta) { result = super.getExpr(delta) }
cached
private module RangeAnalysisCache {
cached
module RangeAnalysisPublic {
/**
* Holds if `b + delta` is a valid bound for `e`.
* - `upper = true` : `e <= b + delta`
* - `upper = false` : `e >= b + delta`
*
* The reason for the bound is given by `reason` and may be either a condition
* or `NoReason` if the bound was proven directly without the use of a bounding
* condition.
*/
cached
predicate semBounded(SemExpr e, SemBound b, int delta, boolean upper, SemReason reason) {
bounded(e, b, delta, upper, _, _, reason) and
bestBound(e, b, delta, upper)
}
}
class SemZeroBound extends SemBound instanceof SemanticBound::SemZeroBound { }
class SemSsaBound extends SemBound instanceof SemanticBound::SemSsaBound {
SemSsaVariable getAVariable() { result = this.(SemanticBound::SemSsaBound).getAVariable() }
/**
* Holds if `guard = boundFlowCond(_, _, _, _, _) or guard = eqFlowCond(_, _, _, _, _)`.
*/
cached
predicate possibleReason(SemGuard guard) {
guard = boundFlowCond(_, _, _, _, _) or guard = semEqFlowCond(_, _, _, _, _)
}
}
private module CppRangeAnalysis =
RangeStage<FloatDelta, Bounds, CppLangImpl, RangeUtil<FloatDelta, CppLangImpl>>;
private import RangeAnalysisCache
import RangeAnalysisPublic
import CppRangeAnalysis
/**
* Holds if `b + delta` is a valid bound for `e` and this is the best such delta.
* - `upper = true` : `e <= b + delta`
* - `upper = false` : `e >= b + delta`
*/
private predicate bestBound(SemExpr e, SemBound b, int delta, boolean upper) {
delta = min(int d | bounded(e, b, d, upper, _, _, _)) and upper = true
or
delta = max(int d | bounded(e, b, d, upper, _, _, _)) and upper = false
}
/**
* Holds if `comp` corresponds to:
* - `upper = true` : `v <= e + delta` or `v < e + delta`
* - `upper = false` : `v >= e + delta` or `v > e + delta`
*/
private predicate boundCondition(
SemRelationalExpr comp, SemSsaVariable v, SemExpr e, int delta, boolean upper
) {
comp.getLesserOperand() = semSsaRead(v, delta) and e = comp.getGreaterOperand() and upper = true
or
comp.getGreaterOperand() = semSsaRead(v, delta) and e = comp.getLesserOperand() and upper = false
or
exists(SemSubExpr sub, SemConstantIntegerExpr c, int d |
// (v - d) - e < c
comp.getLesserOperand() = sub and
comp.getGreaterOperand() = c and
sub.getLeftOperand() = semSsaRead(v, d) and
sub.getRightOperand() = e and
upper = true and
delta = d + c.getIntValue()
or
// (v - d) - e > c
comp.getGreaterOperand() = sub and
comp.getLesserOperand() = c and
sub.getLeftOperand() = semSsaRead(v, d) and
sub.getRightOperand() = e and
upper = false and
delta = d + c.getIntValue()
or
// e - (v - d) < c
comp.getLesserOperand() = sub and
comp.getGreaterOperand() = c and
sub.getLeftOperand() = e and
sub.getRightOperand() = semSsaRead(v, d) and
upper = false and
delta = d - c.getIntValue()
or
// e - (v - d) > c
comp.getGreaterOperand() = sub and
comp.getLesserOperand() = c and
sub.getLeftOperand() = e and
sub.getRightOperand() = semSsaRead(v, d) and
upper = true and
delta = d - c.getIntValue()
)
}
/**
* Holds if `comp` is a comparison between `x` and `y` for which `y - x` has a
* fixed value modulo some `mod > 1`, such that the comparison can be
* strengthened by `strengthen` when evaluating to `testIsTrue`.
*/
private predicate modulusComparison(SemRelationalExpr comp, boolean testIsTrue, int strengthen) {
exists(
SemBound b, int v1, int v2, int mod1, int mod2, int mod, boolean resultIsStrict, int d, int k
|
// If `x <= y` and `x =(mod) b + v1` and `y =(mod) b + v2` then
// `0 <= y - x =(mod) v2 - v1`. By choosing `k =(mod) v2 - v1` with
// `0 <= k < mod` we get `k <= y - x`. If the resulting comparison is
// strict then the strengthening amount is instead `k - 1` modulo `mod`:
// `x < y` means `0 <= y - x - 1 =(mod) k - 1` so `k - 1 <= y - x - 1` and
// thus `k - 1 < y - x` with `0 <= k - 1 < mod`.
semExprModulus(comp.getLesserOperand(), b, v1, mod1) and
semExprModulus(comp.getGreaterOperand(), b, v2, mod2) and
mod = mod1.gcd(mod2) and
mod != 1 and
(testIsTrue = true or testIsTrue = false) and
(
if comp.isStrict()
then resultIsStrict = testIsTrue
else resultIsStrict = testIsTrue.booleanNot()
) and
(
resultIsStrict = true and d = 1
or
resultIsStrict = false and d = 0
) and
(
testIsTrue = true and k = v2 - v1
or
testIsTrue = false and k = v1 - v2
) and
strengthen = (((k - d) % mod) + mod) % mod
)
}
/**
* Gets a condition that tests whether `v` is bounded by `e + delta`.
*
* If the condition evaluates to `testIsTrue`:
* - `upper = true` : `v <= e + delta`
* - `upper = false` : `v >= e + delta`
*/
private SemGuard boundFlowCond(
SemSsaVariable v, SemExpr e, int delta, boolean upper, boolean testIsTrue
) {
exists(
SemRelationalExpr comp, int d1, int d2, int d3, int strengthen, boolean compIsUpper,
boolean resultIsStrict
|
comp = result.asExpr() and
boundCondition(comp, v, e, d1, compIsUpper) and
(testIsTrue = true or testIsTrue = false) and
upper = compIsUpper.booleanXor(testIsTrue.booleanNot()) and
(
if comp.isStrict()
then resultIsStrict = testIsTrue
else resultIsStrict = testIsTrue.booleanNot()
) and
(
if
getTrackedTypeForSsaVariable(v) instanceof SemIntegerType or
getTrackedTypeForSsaVariable(v) instanceof SemAddressType
then
upper = true and strengthen = -1
or
upper = false and strengthen = 1
else strengthen = 0
) and
(
exists(int k | modulusComparison(comp, testIsTrue, k) and d2 = strengthen * k)
or
not modulusComparison(comp, testIsTrue, _) and d2 = 0
) and
// A strict inequality `x < y` can be strengthened to `x <= y - 1`.
(
resultIsStrict = true and d3 = strengthen
or
resultIsStrict = false and d3 = 0
) and
delta = d1 + d2 + d3
)
or
exists(boolean testIsTrue0 |
semImplies_v2(result, testIsTrue, boundFlowCond(v, e, delta, upper, testIsTrue0), testIsTrue0)
)
or
result = semEqFlowCond(v, e, delta, true, testIsTrue) and
(upper = true or upper = false)
or
// guard that tests whether `v2` is bounded by `e + delta + d1 - d2` and
// exists a guard `guardEq` such that `v = v2 - d1 + d2`.
exists(SemSsaVariable v2, SemGuard guardEq, boolean eqIsTrue, int d1, int d2 |
guardEq = semEqFlowCond(v, semSsaRead(v2, d1), d2, true, eqIsTrue) and
result = boundFlowCond(v2, e, delta + d1 - d2, upper, testIsTrue) and
// guardEq needs to control guard
guardEq.directlyControls(result.getBasicBlock(), eqIsTrue)
)
}
private newtype TSemReason =
TSemNoReason() or
TSemCondReason(SemGuard guard) { possibleReason(guard) }
/**
* A reason for an inferred bound. This can either be `CondReason` if the bound
* is due to a specific condition, or `NoReason` if the bound is inferred
* without going through a bounding condition.
*/
abstract class SemReason extends TSemReason {
/** Gets a textual representation of this reason. */
abstract string toString();
}
/**
* A reason for an inferred bound that indicates that the bound is inferred
* without going through a bounding condition.
*/
class SemNoReason extends SemReason, TSemNoReason {
override string toString() { result = "NoReason" }
}
/** A reason for an inferred bound pointing to a condition. */
class SemCondReason extends SemReason, TSemCondReason {
/** Gets the condition that is the reason for the bound. */
SemGuard getCond() { this = TSemCondReason(result) }
override string toString() { result = getCond().toString() }
}
/**
* Holds if `e + delta` is a valid bound for `v` at `pos`.
* - `upper = true` : `v <= e + delta`
* - `upper = false` : `v >= e + delta`
*/
private predicate boundFlowStepSsa(
SemSsaVariable v, SemSsaReadPosition pos, SemExpr e, int delta, boolean upper, SemReason reason
) {
semSsaUpdateStep(v, e, delta) and
pos.hasReadOfVar(v) and
(upper = true or upper = false) and
reason = TSemNoReason()
or
exists(SemGuard guard, boolean testIsTrue |
pos.hasReadOfVar(v) and
guard = boundFlowCond(v, e, delta, upper, testIsTrue) and
semGuardDirectlyControlsSsaRead(guard, pos, testIsTrue) and
reason = TSemCondReason(guard)
)
}
/** Holds if `v != e + delta` at `pos` and `v` is of integral type. */
private predicate unequalFlowStepIntegralSsa(
SemSsaVariable v, SemSsaReadPosition pos, SemExpr e, int delta, SemReason reason
) {
getTrackedTypeForSsaVariable(v) instanceof SemIntegerType and
exists(SemGuard guard, boolean testIsTrue |
pos.hasReadOfVar(v) and
guard = semEqFlowCond(v, e, delta, false, testIsTrue) and
semGuardDirectlyControlsSsaRead(guard, pos, testIsTrue) and
reason = TSemCondReason(guard)
)
}
/**
* An expression that does conversion, boxing, or unboxing
*/
private class ConvertOrBoxExpr extends SemUnaryExpr {
ConvertOrBoxExpr() {
this instanceof SemConvertExpr
or
this instanceof SemBoxExpr
or
this instanceof SemUnboxExpr
}
}
/**
* A cast that can be ignored for the purpose of range analysis.
*/
private class SafeCastExpr extends ConvertOrBoxExpr {
SafeCastExpr() {
conversionCannotOverflow(getTrackedType(pragma[only_bind_into](getOperand())),
getTrackedType(this))
}
}
/**
* Holds if `typ` is a small integral type with the given lower and upper bounds.
*/
private predicate typeBound(SemIntegerType typ, int lowerbound, int upperbound) {
exists(int bitSize | bitSize = typ.getByteSize() * 8 |
bitSize < 32 and
(
if typ.isSigned()
then (
upperbound = 1.bitShiftLeft(bitSize - 1) - 1 and
lowerbound = -upperbound - 1
) else (
lowerbound = 0 and
upperbound = 1.bitShiftLeft(bitSize) - 1
)
)
)
}
/**
* A cast to a small integral type that may overflow or underflow.
*/
private class NarrowingCastExpr extends ConvertOrBoxExpr {
NarrowingCastExpr() {
not this instanceof SafeCastExpr and
typeBound(getTrackedType(this), _, _)
}
/** Gets the lower bound of the resulting type. */
int getLowerBound() { typeBound(getTrackedType(this), result, _) }
/** Gets the upper bound of the resulting type. */
int getUpperBound() { typeBound(getTrackedType(this), _, result) }
}
/** Holds if `e >= 1` as determined by sign analysis. */
private predicate strictlyPositiveIntegralExpr(SemExpr e) {
semStrictlyPositive(e) and getTrackedType(e) instanceof SemIntegerType
}
/** Holds if `e <= -1` as determined by sign analysis. */
private predicate strictlyNegativeIntegralExpr(SemExpr e) {
semStrictlyNegative(e) and getTrackedType(e) instanceof SemIntegerType
}
/**
* Holds if `e1 + delta` is a valid bound for `e2`.
* - `upper = true` : `e2 <= e1 + delta`
* - `upper = false` : `e2 >= e1 + delta`
*/
private predicate boundFlowStep(SemExpr e2, SemExpr e1, int delta, boolean upper) {
semValueFlowStep(e2, e1, delta) and
(upper = true or upper = false)
or
e2.(SafeCastExpr).getOperand() = e1 and
delta = 0 and
(upper = true or upper = false)
or
exists(SemExpr x | e2.(SemAddExpr).hasOperands(e1, x) |
// `x instanceof ConstantIntegerExpr` is covered by valueFlowStep
not x instanceof SemConstantIntegerExpr and
not e1 instanceof SemConstantIntegerExpr and
if strictlyPositiveIntegralExpr(x)
then upper = false and delta = 1
else
if semPositive(x)
then upper = false and delta = 0
else
if strictlyNegativeIntegralExpr(x)
then upper = true and delta = -1
else
if semNegative(x)
then upper = true and delta = 0
else none()
)
or
exists(SemExpr x, SemSubExpr sub |
e2 = sub and
sub.getLeftOperand() = e1 and
sub.getRightOperand() = x
|
// `x instanceof ConstantIntegerExpr` is covered by valueFlowStep
not x instanceof SemConstantIntegerExpr and
if strictlyPositiveIntegralExpr(x)
then upper = true and delta = -1
else
if semPositive(x)
then upper = true and delta = 0
else
if strictlyNegativeIntegralExpr(x)
then upper = false and delta = 1
else
if semNegative(x)
then upper = false and delta = 0
else none()
)
or
e2.(SemRemExpr).getRightOperand() = e1 and
semPositive(e1) and
delta = -1 and
upper = true
or
e2.(SemRemExpr).getLeftOperand() = e1 and semPositive(e1) and delta = 0 and upper = true
or
e2.(SemBitAndExpr).getAnOperand() = e1 and
semPositive(e1) and
delta = 0 and
upper = true
or
e2.(SemBitOrExpr).getAnOperand() = e1 and
semPositive(e2) and
delta = 0 and
upper = false
or
Specific::hasBound(e2, e1, delta, upper)
}
/** Holds if `e2 = e1 * factor` and `factor > 0`. */
private predicate boundFlowStepMul(SemExpr e2, SemExpr e1, int factor) {
exists(SemConstantIntegerExpr c, int k | k = c.getIntValue() and k > 0 |
e2.(SemMulExpr).hasOperands(e1, c) and factor = k
or
exists(SemShiftLeftExpr e |
e = e2 and e.getLeftOperand() = e1 and e.getRightOperand() = c and factor = 2.pow(k)
)
)
}
/**
* Holds if `e2 = e1 / factor` and `factor > 0`.
*
* This conflates division, right shift, and unsigned right shift and is
* therefore only valid for non-negative numbers.
*/
private predicate boundFlowStepDiv(SemExpr e2, SemExpr e1, int factor) {
exists(SemConstantIntegerExpr c, int k | k = c.getIntValue() and k > 0 |
exists(SemDivExpr e |
e = e2 and e.getLeftOperand() = e1 and e.getRightOperand() = c and factor = k
)
or
exists(SemShiftRightExpr e |
e = e2 and e.getLeftOperand() = e1 and e.getRightOperand() = c and factor = 2.pow(k)
)
or
exists(SemShiftRightUnsignedExpr e |
e = e2 and e.getLeftOperand() = e1 and e.getRightOperand() = c and factor = 2.pow(k)
)
)
}
/**
* Holds if `b + delta` is a valid bound for `v` at `pos`.
* - `upper = true` : `v <= b + delta`
* - `upper = false` : `v >= b + delta`
*/
private predicate boundedSsa(
SemSsaVariable v, SemSsaReadPosition pos, SemBound b, int delta, boolean upper,
boolean fromBackEdge, int origdelta, SemReason reason
) {
exists(SemExpr mid, int d1, int d2, SemReason r1, SemReason r2 |
boundFlowStepSsa(v, pos, mid, d1, upper, r1) and
bounded(mid, b, d2, upper, fromBackEdge, origdelta, r2) and
// upper = true: v <= mid + d1 <= b + d1 + d2 = b + delta
// upper = false: v >= mid + d1 >= b + d1 + d2 = b + delta
delta = d1 + d2 and
(if r1 instanceof SemNoReason then reason = r2 else reason = r1)
)
or
exists(int d, SemReason r1, SemReason r2 |
boundedSsa(v, pos, b, d, upper, fromBackEdge, origdelta, r2) or
boundedPhi(v, b, d, upper, fromBackEdge, origdelta, r2)
|
unequalIntegralSsa(v, pos, b, d, r1) and
(
upper = true and delta = d - 1
or
upper = false and delta = d + 1
) and
(
reason = r1
or
reason = r2 and not r2 instanceof SemNoReason
)
)
}
/**
* Holds if `v != b + delta` at `pos` and `v` is of integral type.
*/
private predicate unequalIntegralSsa(
SemSsaVariable v, SemSsaReadPosition pos, SemBound b, int delta, SemReason reason
) {
exists(SemExpr e, int d1, int d2 |
unequalFlowStepIntegralSsa(v, pos, e, d1, reason) and
boundedUpper(e, b, d1) and
boundedLower(e, b, d2) and
delta = d2 + d1
)
}
/**
* Holds if `b + delta` is an upper bound for `e`.
*
* This predicate only exists to prevent a bad standard order in `unequalIntegralSsa`.
*/
pragma[nomagic]
private predicate boundedUpper(SemExpr e, SemBound b, int delta) {
bounded(e, b, delta, true, _, _, _)
}
/**
* Holds if `b + delta` is a lower bound for `e`.
*
* This predicate only exists to prevent a bad standard order in `unequalIntegralSsa`.
*/
pragma[nomagic]
private predicate boundedLower(SemExpr e, SemBound b, int delta) {
bounded(e, b, delta, false, _, _, _)
}
/** Weakens a delta to lie in the range `[-1..1]`. */
bindingset[delta, upper]
private int weakenDelta(boolean upper, int delta) {
delta in [-1 .. 1] and result = delta
or
upper = true and result = -1 and delta < -1
or
upper = false and result = 1 and delta > 1
}
/**
* Holds if `b + delta` is a valid bound for `inp` when used as an input to
* `phi` along `edge`.
* - `upper = true` : `inp <= b + delta`
* - `upper = false` : `inp >= b + delta`
*/
private predicate boundedPhiInp(
SemSsaPhiNode phi, SemSsaVariable inp, SemSsaReadPositionPhiInputEdge edge, SemBound b, int delta,
boolean upper, boolean fromBackEdge, int origdelta, SemReason reason
) {
edge.phiInput(phi, inp) and
exists(int d, boolean fromBackEdge0 |
boundedSsa(inp, edge, b, d, upper, fromBackEdge0, origdelta, reason)
or
boundedPhi(inp, b, d, upper, fromBackEdge0, origdelta, reason)
or
b.(SemSsaBound).getAVariable() = inp and
d = 0 and
(upper = true or upper = false) and
fromBackEdge0 = false and
origdelta = 0 and
reason = TSemNoReason()
|
if semBackEdge(phi, inp, edge)
then
fromBackEdge = true and
(
fromBackEdge0 = true and delta = weakenDelta(upper, d - origdelta) + origdelta
or
fromBackEdge0 = false and delta = d
)
else (
delta = d and fromBackEdge = fromBackEdge0
)
)
}
/**
* Holds if `b + delta` is a valid bound for `inp` when used as an input to
* `phi` along `edge`.
* - `upper = true` : `inp <= b + delta`
* - `upper = false` : `inp >= b + delta`
*
* Equivalent to `boundedPhiInp(phi, inp, edge, b, delta, upper, _, _, _)`.
*/
pragma[noinline]
private predicate boundedPhiInp1(
SemSsaPhiNode phi, SemBound b, boolean upper, SemSsaVariable inp,
SemSsaReadPositionPhiInputEdge edge, int delta
) {
boundedPhiInp(phi, inp, edge, b, delta, upper, _, _, _)
}
/**
* Holds if `phi` is a valid bound for `inp` when used as an input to `phi`
* along `edge`.
* - `upper = true` : `inp <= phi`
* - `upper = false` : `inp >= phi`
*/
private predicate selfBoundedPhiInp(
SemSsaPhiNode phi, SemSsaVariable inp, SemSsaReadPositionPhiInputEdge edge, boolean upper
) {
exists(int d, SemSsaBound phibound |
phibound.getAVariable() = phi and
boundedPhiInp(phi, inp, edge, phibound, d, upper, _, _, _) and
(
upper = true and d <= 0
or
upper = false and d >= 0
)
)
}
/**
* Holds if `b + delta` is a valid bound for some input, `inp`, to `phi`, and
* thus a candidate bound for `phi`.
* - `upper = true` : `inp <= b + delta`
* - `upper = false` : `inp >= b + delta`
*/
pragma[noinline]
private predicate boundedPhiCand(
SemSsaPhiNode phi, boolean upper, SemBound b, int delta, boolean fromBackEdge, int origdelta,
SemReason reason
) {
exists(SemSsaVariable inp, SemSsaReadPositionPhiInputEdge edge |
boundedPhiInp(phi, inp, edge, b, delta, upper, fromBackEdge, origdelta, reason)
)
}
/**
* Holds if the candidate bound `b + delta` for `phi` is valid for the phi input
* `inp` along `edge`.
*/
private predicate boundedPhiCandValidForEdge(
SemSsaPhiNode phi, SemBound b, int delta, boolean upper, boolean fromBackEdge, int origdelta,
SemReason reason, SemSsaVariable inp, SemSsaReadPositionPhiInputEdge edge
) {
boundedPhiCand(phi, upper, b, delta, fromBackEdge, origdelta, reason) and
(
exists(int d | boundedPhiInp1(phi, b, upper, inp, edge, d) | upper = true and d <= delta)
or
exists(int d | boundedPhiInp1(phi, b, upper, inp, edge, d) | upper = false and d >= delta)
or
selfBoundedPhiInp(phi, inp, edge, upper)
)
}
/**
* Holds if `b + delta` is a valid bound for `phi`.
* - `upper = true` : `phi <= b + delta`
* - `upper = false` : `phi >= b + delta`
*/
private predicate boundedPhi(
SemSsaPhiNode phi, SemBound b, int delta, boolean upper, boolean fromBackEdge, int origdelta,
SemReason reason
) {
forex(SemSsaVariable inp, SemSsaReadPositionPhiInputEdge edge | edge.phiInput(phi, inp) |
boundedPhiCandValidForEdge(phi, b, delta, upper, fromBackEdge, origdelta, reason, inp, edge)
)
}
/**
* Holds if `e` has an upper (for `upper = true`) or lower
* (for `upper = false`) bound of `b`.
*/
private predicate baseBound(SemExpr e, int b, boolean upper) {
Specific::hasConstantBound(e, b, upper)
or
upper = false and
b = 0 and
semPositive(e.(SemBitAndExpr).getAnOperand()) and
// REVIEW: We let the language opt out here to preserve original results.
not Specific::ignoreZeroLowerBound(e)
}
/**
* Holds if the value being cast has an upper (for `upper = true`) or lower
* (for `upper = false`) bound within the bounds of the resulting type.
* For `upper = true` this means that the cast will not overflow and for
* `upper = false` this means that the cast will not underflow.
*/
private predicate safeNarrowingCast(NarrowingCastExpr cast, boolean upper) {
exists(int bound | bounded(cast.getOperand(), any(SemZeroBound zb), bound, upper, _, _, _) |
upper = true and bound <= cast.getUpperBound()
or
upper = false and bound >= cast.getLowerBound()
)
}
pragma[noinline]
private predicate boundedCastExpr(
NarrowingCastExpr cast, SemBound b, int delta, boolean upper, boolean fromBackEdge, int origdelta,
SemReason reason
) {
bounded(cast.getOperand(), b, delta, upper, fromBackEdge, origdelta, reason)
}
/**
* Holds if `b + delta` is a valid bound for `e`.
* - `upper = true` : `e <= b + delta`
* - `upper = false` : `e >= b + delta`
*/
private predicate bounded(
SemExpr e, SemBound b, int delta, boolean upper, boolean fromBackEdge, int origdelta,
SemReason reason
) {
not Specific::ignoreExprBound(e) and
(
e = b.getExpr(delta) and
(upper = true or upper = false) and
fromBackEdge = false and
origdelta = delta and
reason = TSemNoReason()
or
baseBound(e, delta, upper) and
b instanceof SemZeroBound and
fromBackEdge = false and
origdelta = delta and
reason = TSemNoReason()
or
exists(SemSsaVariable v, SemSsaReadPositionBlock bb |
boundedSsa(v, bb, b, delta, upper, fromBackEdge, origdelta, reason) and
e = v.getAUse() and
bb.getBlock() = e.getBasicBlock()
)
or
exists(SemExpr mid, int d1, int d2 |
boundFlowStep(e, mid, d1, upper) and
// Constants have easy, base-case bounds, so let's not infer any recursive bounds.
not e instanceof SemConstantIntegerExpr and
bounded(mid, b, d2, upper, fromBackEdge, origdelta, reason) and
// upper = true: e <= mid + d1 <= b + d1 + d2 = b + delta
// upper = false: e >= mid + d1 >= b + d1 + d2 = b + delta
delta = d1 + d2
)
or
exists(SemSsaPhiNode phi |
boundedPhi(phi, b, delta, upper, fromBackEdge, origdelta, reason) and
e = phi.getAUse()
)
or
exists(SemExpr mid, int factor, int d |
boundFlowStepMul(e, mid, factor) and
not e instanceof SemConstantIntegerExpr and
bounded(mid, b, d, upper, fromBackEdge, origdelta, reason) and
b instanceof SemZeroBound and
delta = d * factor
)
or
exists(SemExpr mid, int factor, int d |
boundFlowStepDiv(e, mid, factor) and
not e instanceof SemConstantIntegerExpr and
bounded(mid, b, d, upper, fromBackEdge, origdelta, reason) and
b instanceof SemZeroBound and
d >= 0 and
delta = d / factor
)
or
exists(NarrowingCastExpr cast |
cast = e and
safeNarrowingCast(cast, upper.booleanNot()) and
boundedCastExpr(cast, b, delta, upper, fromBackEdge, origdelta, reason)
)
or
exists(
SemConditionalExpr cond, int d1, int d2, boolean fbe1, boolean fbe2, int od1, int od2,
SemReason r1, SemReason r2
|
cond = e and
boundedConditionalExpr(cond, b, upper, true, d1, fbe1, od1, r1) and
boundedConditionalExpr(cond, b, upper, false, d2, fbe2, od2, r2) and
(
delta = d1 and fromBackEdge = fbe1 and origdelta = od1 and reason = r1
or
delta = d2 and fromBackEdge = fbe2 and origdelta = od2 and reason = r2
)
|
upper = true and delta = d1.maximum(d2)
or
upper = false and delta = d1.minimum(d2)
)
)
}
private predicate boundedConditionalExpr(
SemConditionalExpr cond, SemBound b, boolean upper, boolean branch, int delta,
boolean fromBackEdge, int origdelta, SemReason reason
) {
bounded(cond.getBranchExpr(branch), b, delta, upper, fromBackEdge, origdelta, reason)
}

View File

@@ -3,90 +3,86 @@
*/
private import experimental.semmle.code.cpp.semantic.Semantic
private import RangeAnalysisStage
private import experimental.semmle.code.cpp.semantic.analysis.FloatDelta
module CppLangImpl implements LangSig<FloatDelta> {
/**
* Holds if the specified expression should be excluded from the result of `ssaRead()`.
*
* This predicate is to keep the results identical to the original Java implementation. It should be
* removed once we have the new implementation matching the old results exactly.
*/
predicate ignoreSsaReadCopy(SemExpr e) { none() }
/**
* Holds if the specified expression should be excluded from the result of `ssaRead()`.
*
* This predicate is to keep the results identical to the original Java implementation. It should be
* removed once we have the new implementation matching the old results exactly.
*/
predicate ignoreSsaReadCopy(SemExpr e) { none() }
/**
* Ignore the bound on this expression.
*
* This predicate is to keep the results identical to the original Java implementation. It should be
* removed once we have the new implementation matching the old results exactly.
*/
predicate ignoreExprBound(SemExpr e) { none() }
/**
* Ignore the bound on this expression.
*
* This predicate is to keep the results identical to the original Java implementation. It should be
* removed once we have the new implementation matching the old results exactly.
*/
predicate ignoreExprBound(SemExpr e) { none() }
/**
* Ignore any inferred zero lower bound on this expression.
*
* This predicate is to keep the results identical to the original Java implementation. It should be
* removed once we have the new implementation matching the old results exactly.
*/
predicate ignoreZeroLowerBound(SemExpr e) { none() }
/**
* Ignore any inferred zero lower bound on this expression.
*
* This predicate is to keep the results identical to the original Java implementation. It should be
* removed once we have the new implementation matching the old results exactly.
*/
predicate ignoreZeroLowerBound(SemExpr e) { none() }
/**
* Holds if the specified expression should be excluded from the result of `ssaRead()`.
*
* This predicate is to keep the results identical to the original Java implementation. It should be
* removed once we have the new implementation matching the old results exactly.
*/
predicate ignoreSsaReadArithmeticExpr(SemExpr e) { none() }
/**
* Holds if the specified expression should be excluded from the result of `ssaRead()`.
*
* This predicate is to keep the results identical to the original Java implementation. It should be
* removed once we have the new implementation matching the old results exactly.
*/
predicate ignoreSsaReadArithmeticExpr(SemExpr e) { none() }
/**
* Holds if the specified variable should be excluded from the result of `ssaRead()`.
*
* This predicate is to keep the results identical to the original Java implementation. It should be
* removed once we have the new implementation matching the old results exactly.
*/
predicate ignoreSsaReadAssignment(SemSsaVariable v) { none() }
/**
* Holds if the specified variable should be excluded from the result of `ssaRead()`.
*
* This predicate is to keep the results identical to the original Java implementation. It should be
* removed once we have the new implementation matching the old results exactly.
*/
predicate ignoreSsaReadAssignment(SemSsaVariable v) { none() }
/**
* Adds additional results to `ssaRead()` that are specific to Java.
*
* This predicate handles propagation of offsets for post-increment and post-decrement expressions
* in exactly the same way as the old Java implementation. Once the new implementation matches the
* old one, we should remove this predicate and propagate deltas for all similar patterns, whether
* or not they come from a post-increment/decrement expression.
*/
SemExpr specificSsaRead(SemSsaVariable v, float delta) { none() }
/**
* Adds additional results to `ssaRead()` that are specific to Java.
*
* This predicate handles propagation of offsets for post-increment and post-decrement expressions
* in exactly the same way as the old Java implementation. Once the new implementation matches the
* old one, we should remove this predicate and propagate deltas for all similar patterns, whether
* or not they come from a post-increment/decrement expression.
*/
SemExpr specificSsaRead(SemSsaVariable v, int delta) { none() }
/**
* Holds if `e >= bound` (if `upper = false`) or `e <= bound` (if `upper = true`).
*/
predicate hasConstantBound(SemExpr e, float bound, boolean upper) { none() }
/**
* Holds if `e >= bound` (if `upper = false`) or `e <= bound` (if `upper = true`).
*/
predicate hasConstantBound(SemExpr e, int bound, boolean upper) { none() }
/**
* Holds if `e >= bound + delta` (if `upper = false`) or `e <= bound + delta` (if `upper = true`).
*/
predicate hasBound(SemExpr e, SemExpr bound, float delta, boolean upper) { none() }
/**
* Holds if `e >= bound + delta` (if `upper = false`) or `e <= bound + delta` (if `upper = true`).
*/
predicate hasBound(SemExpr e, SemExpr bound, int delta, boolean upper) { none() }
/**
* Holds if the value of `dest` is known to be `src + delta`.
*/
predicate additionalValueFlowStep(SemExpr dest, SemExpr src, float delta) { none() }
/**
* Holds if the value of `dest` is known to be `src + delta`.
*/
predicate additionalValueFlowStep(SemExpr dest, SemExpr src, int delta) { none() }
/**
* Gets the type that range analysis should use to track the result of the specified expression,
* if a type other than the original type of the expression is to be used.
*
* This predicate is commonly used in languages that support immutable "boxed" types that are
* actually references but whose values can be tracked as the type contained in the box.
*/
SemType getAlternateType(SemExpr e) { none() }
/**
* Gets the type that range analysis should use to track the result of the specified expression,
* if a type other than the original type of the expression is to be used.
*
* This predicate is commonly used in languages that support immutable "boxed" types that are
* actually references but whose values can be tracked as the type contained in the box.
*/
SemType getAlternateType(SemExpr e) { none() }
/**
* Gets the type that range analysis should use to track the result of the specified source
* variable, if a type other than the original type of the expression is to be used.
*
* This predicate is commonly used in languages that support immutable "boxed" types that are
* actually references but whose values can be tracked as the type contained in the box.
*/
SemType getAlternateTypeForSsaVariable(SemSsaVariable var) { none() }
}
/**
* Gets the type that range analysis should use to track the result of the specified source
* variable, if a type other than the original type of the expression is to be used.
*
* This predicate is commonly used in languages that support immutable "boxed" types that are
* actually references but whose values can be tracked as the type contained in the box.
*/
SemType getAlternateTypeForSsaVariable(SemSsaVariable var) { none() }

View File

@@ -3,138 +3,133 @@
*/
private import experimental.semmle.code.cpp.semantic.Semantic
private import RangeAnalysisSpecific
private import RangeAnalysisStage as Range
private import RangeAnalysisSpecific as Specific
private import ConstantAnalysis
module RangeUtil<Range::DeltaSig D, Range::LangSig<D> Lang> implements Range::UtilSig<D> {
/**
* Gets an expression that equals `v - d`.
*/
SemExpr semSsaRead(SemSsaVariable v, D::Delta delta) {
// There are various language-specific extension points that can be removed once we no longer
// expect to match the original Java implementation's results exactly.
result = v.getAUse() and delta = D::fromInt(0)
or
exists(D::Delta d1, SemConstantIntegerExpr c |
result.(SemAddExpr).hasOperands(semSsaRead(v, d1), c) and
delta = D::fromFloat(D::toFloat(d1) - c.getIntValue()) and
not Lang::ignoreSsaReadArithmeticExpr(result)
)
or
exists(SemSubExpr sub, D::Delta d1, SemConstantIntegerExpr c |
result = sub and
sub.getLeftOperand() = semSsaRead(v, d1) and
sub.getRightOperand() = c and
delta = D::fromFloat(D::toFloat(d1) + c.getIntValue()) and
not Lang::ignoreSsaReadArithmeticExpr(result)
)
or
result = v.(SemSsaExplicitUpdate).getSourceExpr() and
delta = D::fromFloat(0) and
not Lang::ignoreSsaReadAssignment(v)
or
result = Lang::specificSsaRead(v, delta)
or
result.(SemCopyValueExpr).getOperand() = semSsaRead(v, delta) and
not Lang::ignoreSsaReadCopy(result)
or
result.(SemStoreExpr).getOperand() = semSsaRead(v, delta)
}
/**
* Gets a condition that tests whether `v` equals `e + delta`.
*
* If the condition evaluates to `testIsTrue`:
* - `isEq = true` : `v == e + delta`
* - `isEq = false` : `v != e + delta`
*/
SemGuard semEqFlowCond(
SemSsaVariable v, SemExpr e, D::Delta delta, boolean isEq, boolean testIsTrue
) {
exists(boolean eqpolarity |
result.isEquality(semSsaRead(v, delta), e, eqpolarity) and
(testIsTrue = true or testIsTrue = false) and
eqpolarity.booleanXor(testIsTrue).booleanNot() = isEq
)
or
exists(boolean testIsTrue0 |
semImplies_v2(result, testIsTrue, semEqFlowCond(v, e, delta, isEq, testIsTrue0), testIsTrue0)
)
}
/**
* Holds if `v` is an `SsaExplicitUpdate` that equals `e + delta`.
*/
predicate semSsaUpdateStep(SemSsaExplicitUpdate v, SemExpr e, D::Delta delta) {
exists(SemExpr defExpr | defExpr = v.getSourceExpr() |
defExpr.(SemCopyValueExpr).getOperand() = e and delta = D::fromFloat(0)
or
defExpr.(SemStoreExpr).getOperand() = e and delta = D::fromFloat(0)
or
defExpr.(SemAddOneExpr).getOperand() = e and delta = D::fromFloat(1)
or
defExpr.(SemSubOneExpr).getOperand() = e and delta = D::fromFloat(-1)
or
e = defExpr and
not (
defExpr instanceof SemCopyValueExpr or
defExpr instanceof SemStoreExpr or
defExpr instanceof SemAddOneExpr or
defExpr instanceof SemSubOneExpr
) and
delta = D::fromFloat(0)
)
}
/**
* Holds if `e1 + delta` equals `e2`.
*/
predicate semValueFlowStep(SemExpr e2, SemExpr e1, D::Delta delta) {
e2.(SemCopyValueExpr).getOperand() = e1 and delta = D::fromFloat(0)
or
e2.(SemStoreExpr).getOperand() = e1 and delta = D::fromFloat(0)
or
e2.(SemAddOneExpr).getOperand() = e1 and delta = D::fromFloat(1)
or
e2.(SemSubOneExpr).getOperand() = e1 and delta = D::fromFloat(-1)
or
Lang::additionalValueFlowStep(e2, e1, delta)
or
exists(SemExpr x | e2.(SemAddExpr).hasOperands(e1, x) |
D::fromInt(x.(SemConstantIntegerExpr).getIntValue()) = delta
)
or
exists(SemExpr x, SemSubExpr sub |
e2 = sub and
sub.getLeftOperand() = e1 and
sub.getRightOperand() = x
|
D::fromInt(-x.(SemConstantIntegerExpr).getIntValue()) = delta
)
}
/**
* Gets the type used to track the specified expression's range information.
*
* Usually, this just `e.getSemType()`, but the language can override this to track immutable boxed
* primitive types as the underlying primitive type.
*/
SemType getTrackedType(SemExpr e) {
result = Lang::getAlternateType(e)
or
not exists(Lang::getAlternateType(e)) and result = e.getSemType()
}
/**
* Gets the type used to track the specified source variable's range information.
*
* Usually, this just `e.getType()`, but the language can override this to track immutable boxed
* primitive types as the underlying primitive type.
*/
SemType getTrackedTypeForSsaVariable(SemSsaVariable var) {
result = Lang::getAlternateTypeForSsaVariable(var)
or
not exists(Lang::getAlternateTypeForSsaVariable(var)) and result = var.getType()
}
/**
* Gets an expression that equals `v - d`.
*/
SemExpr semSsaRead(SemSsaVariable v, int delta) {
// There are various language-specific extension points that can be removed once we no longer
// expect to match the original Java implementation's results exactly.
result = v.getAUse() and delta = 0
or
exists(int d1, SemConstantIntegerExpr c |
result.(SemAddExpr).hasOperands(semSsaRead(v, d1), c) and
delta = d1 - c.getIntValue() and
not Specific::ignoreSsaReadArithmeticExpr(result)
)
or
exists(SemSubExpr sub, int d1, SemConstantIntegerExpr c |
result = sub and
sub.getLeftOperand() = semSsaRead(v, d1) and
sub.getRightOperand() = c and
delta = d1 + c.getIntValue() and
not Specific::ignoreSsaReadArithmeticExpr(result)
)
or
result = v.(SemSsaExplicitUpdate).getSourceExpr() and
delta = 0 and
not Specific::ignoreSsaReadAssignment(v)
or
result = Specific::specificSsaRead(v, delta)
or
result.(SemCopyValueExpr).getOperand() = semSsaRead(v, delta) and
not Specific::ignoreSsaReadCopy(result)
or
result.(SemStoreExpr).getOperand() = semSsaRead(v, delta)
}
/**
* Gets a condition that tests whether `v` equals `e + delta`.
*
* If the condition evaluates to `testIsTrue`:
* - `isEq = true` : `v == e + delta`
* - `isEq = false` : `v != e + delta`
*/
SemGuard semEqFlowCond(SemSsaVariable v, SemExpr e, int delta, boolean isEq, boolean testIsTrue) {
exists(boolean eqpolarity |
result.isEquality(semSsaRead(v, delta), e, eqpolarity) and
(testIsTrue = true or testIsTrue = false) and
eqpolarity.booleanXor(testIsTrue).booleanNot() = isEq
)
or
exists(boolean testIsTrue0 |
semImplies_v2(result, testIsTrue, semEqFlowCond(v, e, delta, isEq, testIsTrue0), testIsTrue0)
)
}
/**
* Holds if `v` is an `SsaExplicitUpdate` that equals `e + delta`.
*/
predicate semSsaUpdateStep(SemSsaExplicitUpdate v, SemExpr e, int delta) {
exists(SemExpr defExpr | defExpr = v.getSourceExpr() |
defExpr.(SemCopyValueExpr).getOperand() = e and delta = 0
or
defExpr.(SemStoreExpr).getOperand() = e and delta = 0
or
defExpr.(SemAddOneExpr).getOperand() = e and delta = 1
or
defExpr.(SemSubOneExpr).getOperand() = e and delta = -1
or
e = defExpr and
not (
defExpr instanceof SemCopyValueExpr or
defExpr instanceof SemStoreExpr or
defExpr instanceof SemAddOneExpr or
defExpr instanceof SemSubOneExpr
) and
delta = 0
)
}
/**
* Holds if `e1 + delta` equals `e2`.
*/
predicate semValueFlowStep(SemExpr e2, SemExpr e1, int delta) {
e2.(SemCopyValueExpr).getOperand() = e1 and delta = 0
or
e2.(SemStoreExpr).getOperand() = e1 and delta = 0
or
e2.(SemAddOneExpr).getOperand() = e1 and delta = 1
or
e2.(SemSubOneExpr).getOperand() = e1 and delta = -1
or
Specific::additionalValueFlowStep(e2, e1, delta)
or
exists(SemExpr x | e2.(SemAddExpr).hasOperands(e1, x) |
x.(SemConstantIntegerExpr).getIntValue() = delta
)
or
exists(SemExpr x, SemSubExpr sub |
e2 = sub and
sub.getLeftOperand() = e1 and
sub.getRightOperand() = x
|
x.(SemConstantIntegerExpr).getIntValue() = -delta
)
}
/**
* Gets the type used to track the specified expression's range information.
*
* Usually, this just `e.getSemType()`, but the language can override this to track immutable boxed
* primitive types as the underlying primitive type.
*/
SemType getTrackedType(SemExpr e) {
result = Specific::getAlternateType(e)
or
not exists(Specific::getAlternateType(e)) and result = e.getSemType()
}
/**
* Gets the type used to track the specified source variable's range information.
*
* Usually, this just `e.getType()`, but the language can override this to track immutable boxed
* primitive types as the underlying primitive type.
*/
SemType getTrackedTypeForSsaVariable(SemSsaVariable var) {
result = Specific::getAlternateTypeForSsaVariable(var)
or
not exists(Specific::getAlternateTypeForSsaVariable(var)) and result = var.getType()
}

View File

@@ -6,494 +6,496 @@
* three-valued domain `{negative, zero, positive}`.
*/
private import RangeAnalysisStage
private import SignAnalysisSpecific as Specific
private import experimental.semmle.code.cpp.semantic.Semantic
private import ConstantAnalysis
private import RangeUtils
private import Sign
module SignAnalysis<DeltaSig D, UtilSig<D> Utils> {
/**
* An SSA definition for which the analysis can compute the sign.
*
* The actual computation of the sign is done in an override of the `getSign()` predicate. The
* charpred of any subclass must _not_ invoke `getSign()`, directly or indirectly. This ensures
* that the charpred does not introduce negative recursion. The `getSign()` predicate may be
* recursive.
*/
abstract private class SignDef instanceof SemSsaVariable {
final string toString() { result = super.toString() }
/**
* An SSA definition for which the analysis can compute the sign.
*
* The actual computation of the sign is done in an override of the `getSign()` predicate. The
* charpred of any subclass must _not_ invoke `getSign()`, directly or indirectly. This ensures
* that the charpred does not introduce negative recursion. The `getSign()` predicate may be
* recursive.
*/
abstract private class SignDef instanceof SemSsaVariable {
final string toString() { result = super.toString() }
/** Gets the possible signs of this SSA definition. */
abstract Sign getSign();
}
/** Gets the possible signs of this SSA definition. */
abstract Sign getSign();
}
/** An SSA definition whose sign is computed based on standard flow. */
abstract private class FlowSignDef extends SignDef {
abstract override Sign getSign();
}
/** An SSA definition whose sign is computed based on standard flow. */
abstract private class FlowSignDef extends SignDef {
abstract override Sign getSign();
}
/** An SSA definition whose sign is determined by the sign of that definitions source expression. */
private class ExplicitSignDef extends FlowSignDef instanceof SemSsaExplicitUpdate {
final override Sign getSign() { result = semExprSign(super.getSourceExpr()) }
}
/** An SSA definition whose sign is determined by the sign of that definitions source expression. */
private class ExplicitSignDef extends FlowSignDef {
SemSsaExplicitUpdate update;
/** An SSA Phi definition, whose sign is the union of the signs of its inputs. */
private class PhiSignDef extends FlowSignDef instanceof SemSsaPhiNode {
final override Sign getSign() {
exists(SemSsaVariable inp, SemSsaReadPositionPhiInputEdge edge |
edge.phiInput(this, inp) and
result = semSsaSign(inp, edge)
)
}
}
ExplicitSignDef() { update = this }
/** An SSA definition whose sign is computed by a language-specific implementation. */
abstract class CustomSignDef extends SignDef {
abstract override Sign getSign();
}
final override Sign getSign() { result = semExprSign(update.getSourceExpr()) }
}
/**
* An expression for which the analysis can compute the sign.
*
* The actual computation of the sign is done in an override of the `getSign()` predicate. The
* charpred of any subclass must _not_ invoke `getSign()`, directly or indirectly. This ensures
* that the charpred does not introduce negative recursion. The `getSign()` predicate may be
* recursive.
*
* Concrete implementations extend one of the following subclasses:
* - `ConstantSignExpr`, for expressions with a compile-time constant value.
* - `FlowSignExpr`, for expressions whose sign can be computed from the signs of their operands.
* - `CustomsignExpr`, for expressions whose sign can be computed by a language-specific
* implementation.
*
* If the same expression matches more than one of the above subclasses, the sign is computed as
* follows:
* - The sign of a `ConstantSignExpr` is computed solely from `ConstantSignExpr.getSign()`,
* regardless of any other subclasses.
* - If a non-`ConstantSignExpr` expression matches exactly one of `FlowSignExpr` or
* `CustomSignExpr`, the sign is computed by that class' `getSign()` predicate.
* - If a non-`ConstantSignExpr` expression matches both `FlowSignExpr` and `CustomSignExpr`, the
* sign is the _intersection_ of the signs of those two classes' `getSign()` predicates. Thus,
* both classes have the opportunity to _restrict_ the set of possible signs, not to generate new
* possible signs.
* - If an expression does not match any of the three subclasses, then it can have any sign.
*
* Note that the `getSign()` predicate is introduced only in subclasses of `SignExpr`.
*/
abstract class SignExpr instanceof SemExpr {
SignExpr() { not Specific::ignoreExprSign(this) }
/** An SSA Phi definition, whose sign is the union of the signs of its inputs. */
private class PhiSignDef extends FlowSignDef {
SemSsaPhiNode phi;
final string toString() { result = super.toString() }
PhiSignDef() { phi = this }
abstract Sign getSign();
}
/** An expression whose sign is determined by its constant numeric value. */
private class ConstantSignExpr extends SignExpr {
ConstantSignExpr() {
this instanceof SemConstantIntegerExpr or
exists(this.(SemNumericLiteralExpr).getApproximateFloatValue())
}
final override Sign getSign() {
exists(int i | this.(SemConstantIntegerExpr).getIntValue() = i |
i < 0 and result = TNeg()
or
i = 0 and result = TZero()
or
i > 0 and result = TPos()
)
or
not exists(this.(SemConstantIntegerExpr).getIntValue()) and
exists(float f | f = this.(SemNumericLiteralExpr).getApproximateFloatValue() |
f < 0 and result = TNeg()
or
f = 0 and result = TZero()
or
f > 0 and result = TPos()
)
}
}
abstract private class NonConstantSignExpr extends SignExpr {
NonConstantSignExpr() { not this instanceof ConstantSignExpr }
final override Sign getSign() {
// The result is the _intersection_ of the signs computed from flow and by the language.
(result = this.(FlowSignExpr).getSignRestriction() or not this instanceof FlowSignExpr) and
(result = this.(CustomSignExpr).getSignRestriction() or not this instanceof CustomSignExpr)
}
}
/** An expression whose sign is computed from the signs of its operands. */
abstract private class FlowSignExpr extends NonConstantSignExpr {
abstract Sign getSignRestriction();
}
/** An expression whose sign is computed by a language-specific implementation. */
abstract class CustomSignExpr extends NonConstantSignExpr {
abstract Sign getSignRestriction();
}
/** An expression whose sign is unknown. */
private class UnknownSignExpr extends SignExpr {
UnknownSignExpr() {
not this instanceof FlowSignExpr and
not this instanceof CustomSignExpr and
not this instanceof ConstantSignExpr and
(
// Only track numeric types.
Utils::getTrackedType(this) instanceof SemNumericType
or
// Unless the language says to track this expression anyway.
Specific::trackUnknownNonNumericExpr(this)
)
}
final override Sign getSign() { semAnySign(result) }
}
/**
* A `Load` expression whose sign is computed from the sign of its SSA definition, restricted by
* inference from any intervening guards.
*/
class UseSignExpr extends FlowSignExpr {
SemSsaVariable v;
UseSignExpr() { v.getAUse() = this }
override Sign getSignRestriction() {
// Propagate via SSA
// Propagate the sign from the def of `v`, incorporating any inference from guards.
result = semSsaSign(v, any(SemSsaReadPositionBlock bb | bb.getAnExpr() = this))
or
// No block for this read. Just use the sign of the def.
// REVIEW: How can this happen?
not exists(SemSsaReadPositionBlock bb | bb.getAnExpr() = this) and
result = semSsaDefSign(v)
}
}
/** A binary expression whose sign is computed from the signs of its operands. */
private class BinarySignExpr extends FlowSignExpr {
SemBinaryExpr binary;
BinarySignExpr() { binary = this }
override Sign getSignRestriction() {
exists(SemExpr left, SemExpr right |
binaryExprOperands(binary, left, right) and
result =
semExprSign(pragma[only_bind_out](left))
.applyBinaryOp(semExprSign(pragma[only_bind_out](right)), binary.getOpcode())
)
or
exists(SemDivExpr div | div = binary |
result = semExprSign(div.getLeftOperand()) and
result != TZero() and
div.getRightOperand().(SemFloatingPointLiteralExpr).getFloatValue() = 0
)
}
}
pragma[nomagic]
private predicate binaryExprOperands(SemBinaryExpr binary, SemExpr left, SemExpr right) {
binary.getLeftOperand() = left and binary.getRightOperand() = right
}
/**
* A `Convert`, `Box`, or `Unbox` expression.
*/
private class SemCastExpr instanceof SemUnaryExpr {
string toString() { result = super.toString() }
SemCastExpr() {
this instanceof SemConvertExpr
or
this instanceof SemBoxExpr
or
this instanceof SemUnboxExpr
}
}
/** A unary expression whose sign is computed from the sign of its operand. */
private class UnarySignExpr extends FlowSignExpr {
SemUnaryExpr unary;
UnarySignExpr() { unary = this and not this instanceof SemCastExpr }
override Sign getSignRestriction() {
result =
semExprSign(pragma[only_bind_out](unary.getOperand())).applyUnaryOp(unary.getOpcode())
}
}
/**
* A `Convert`, `Box`, or `Unbox` expression, whose sign is computed based on
* the sign of its operand and the source and destination types.
*/
abstract private class CastSignExpr extends FlowSignExpr {
SemUnaryExpr cast;
CastSignExpr() { cast = this and cast instanceof SemCastExpr }
override Sign getSignRestriction() { result = semExprSign(cast.getOperand()) }
}
/**
* A `Convert` expression.
*/
private class ConvertSignExpr extends CastSignExpr {
override SemConvertExpr cast;
}
/**
* A `Box` expression.
*/
private class BoxSignExpr extends CastSignExpr {
override SemBoxExpr cast;
}
/**
* An `Unbox` expression.
*/
private class UnboxSignExpr extends CastSignExpr {
override SemUnboxExpr cast;
UnboxSignExpr() {
exists(SemType fromType | fromType = Utils::getTrackedType(cast.getOperand()) |
// Only numeric source types are handled here.
fromType instanceof SemNumericType
)
}
}
private predicate unknownSign(SemExpr e) { e instanceof UnknownSignExpr }
/**
* Holds if `lowerbound` is a lower bound for `v` at `pos`. This is restricted
* to only include bounds for which we might determine a sign.
*/
private predicate lowerBound(
SemExpr lowerbound, SemSsaVariable v, SemSsaReadPosition pos, boolean isStrict
) {
exists(boolean testIsTrue, SemRelationalExpr comp |
pos.hasReadOfVar(v) and
semGuardControlsSsaRead(semGetComparisonGuard(comp), pos, testIsTrue) and
not unknownSign(lowerbound)
|
testIsTrue = true and
comp.getLesserOperand() = lowerbound and
comp.getGreaterOperand() = Utils::semSsaRead(v, D::fromInt(0)) and
(if comp.isStrict() then isStrict = true else isStrict = false)
or
testIsTrue = false and
comp.getGreaterOperand() = lowerbound and
comp.getLesserOperand() = Utils::semSsaRead(v, D::fromInt(0)) and
(if comp.isStrict() then isStrict = false else isStrict = true)
final override Sign getSign() {
exists(SemSsaVariable inp, SemSsaReadPositionPhiInputEdge edge |
edge.phiInput(phi, inp) and
result = semSsaSign(inp, edge)
)
}
/**
* Holds if `upperbound` is an upper bound for `v` at `pos`. This is restricted
* to only include bounds for which we might determine a sign.
*/
private predicate upperBound(
SemExpr upperbound, SemSsaVariable v, SemSsaReadPosition pos, boolean isStrict
) {
exists(boolean testIsTrue, SemRelationalExpr comp |
pos.hasReadOfVar(v) and
semGuardControlsSsaRead(semGetComparisonGuard(comp), pos, testIsTrue) and
not unknownSign(upperbound)
|
testIsTrue = true and
comp.getGreaterOperand() = upperbound and
comp.getLesserOperand() = Utils::semSsaRead(v, D::fromInt(0)) and
(if comp.isStrict() then isStrict = true else isStrict = false)
or
testIsTrue = false and
comp.getLesserOperand() = upperbound and
comp.getGreaterOperand() = Utils::semSsaRead(v, D::fromInt(0)) and
(if comp.isStrict() then isStrict = false else isStrict = true)
)
}
/**
* Holds if `eqbound` is an equality/inequality for `v` at `pos`. This is
* restricted to only include bounds for which we might determine a sign. The
* boolean `isEq` gives the polarity:
* - `isEq = true` : `v = eqbound`
* - `isEq = false` : `v != eqbound`
*/
private predicate eqBound(SemExpr eqbound, SemSsaVariable v, SemSsaReadPosition pos, boolean isEq) {
exists(SemGuard guard, boolean testIsTrue, boolean polarity |
pos.hasReadOfVar(v) and
semGuardControlsSsaRead(guard, pos, testIsTrue) and
guard.isEquality(eqbound, Utils::semSsaRead(v, D::fromInt(0)), polarity) and
isEq = polarity.booleanXor(testIsTrue).booleanNot() and
not unknownSign(eqbound)
)
}
/**
* Holds if `bound` is a bound for `v` at `pos` that needs to be positive in
* order for `v` to be positive.
*/
private predicate posBound(SemExpr bound, SemSsaVariable v, SemSsaReadPosition pos) {
upperBound(bound, v, pos, _) or
eqBound(bound, v, pos, true)
}
/**
* Holds if `bound` is a bound for `v` at `pos` that needs to be negative in
* order for `v` to be negative.
*/
private predicate negBound(SemExpr bound, SemSsaVariable v, SemSsaReadPosition pos) {
lowerBound(bound, v, pos, _) or
eqBound(bound, v, pos, true)
}
/**
* Holds if `bound` is a bound for `v` at `pos` that can restrict whether `v`
* can be zero.
*/
private predicate zeroBound(SemExpr bound, SemSsaVariable v, SemSsaReadPosition pos) {
lowerBound(bound, v, pos, _) or
upperBound(bound, v, pos, _) or
eqBound(bound, v, pos, _)
}
/** Holds if `bound` allows `v` to be positive at `pos`. */
private predicate posBoundOk(SemExpr bound, SemSsaVariable v, SemSsaReadPosition pos) {
posBound(bound, v, pos) and TPos() = semExprSign(bound)
}
/** Holds if `bound` allows `v` to be negative at `pos`. */
private predicate negBoundOk(SemExpr bound, SemSsaVariable v, SemSsaReadPosition pos) {
negBound(bound, v, pos) and TNeg() = semExprSign(bound)
}
/** Holds if `bound` allows `v` to be zero at `pos`. */
private predicate zeroBoundOk(SemExpr bound, SemSsaVariable v, SemSsaReadPosition pos) {
lowerBound(bound, v, pos, _) and TNeg() = semExprSign(bound)
or
lowerBound(bound, v, pos, false) and TZero() = semExprSign(bound)
or
upperBound(bound, v, pos, _) and TPos() = semExprSign(bound)
or
upperBound(bound, v, pos, false) and TZero() = semExprSign(bound)
or
eqBound(bound, v, pos, true) and TZero() = semExprSign(bound)
or
eqBound(bound, v, pos, false) and TZero() != semExprSign(bound)
}
/**
* Holds if there is a bound that might restrict whether `v` has the sign `s`
* at `pos`.
*/
private predicate hasGuard(SemSsaVariable v, SemSsaReadPosition pos, Sign s) {
s = TPos() and posBound(_, v, pos)
or
s = TNeg() and negBound(_, v, pos)
or
s = TZero() and zeroBound(_, v, pos)
}
/**
* Gets a possible sign of `v` at `pos` based on its definition, where the sign
* might be ruled out by a guard.
*/
pragma[noinline]
private Sign guardedSsaSign(SemSsaVariable v, SemSsaReadPosition pos) {
result = semSsaDefSign(v) and
pos.hasReadOfVar(v) and
hasGuard(v, pos, result)
}
/**
* Gets a possible sign of `v` at `pos` based on its definition, where no guard
* can rule it out.
*/
pragma[noinline]
private Sign unguardedSsaSign(SemSsaVariable v, SemSsaReadPosition pos) {
result = semSsaDefSign(v) and
pos.hasReadOfVar(v) and
not hasGuard(v, pos, result)
}
/**
* Gets a possible sign of `v` at read position `pos`, where a guard could have
* ruled out the sign but does not.
* This does not check that the definition of `v` also allows the sign.
*/
private Sign guardedSsaSignOk(SemSsaVariable v, SemSsaReadPosition pos) {
result = TPos() and
forex(SemExpr bound | posBound(bound, v, pos) | posBoundOk(bound, v, pos))
or
result = TNeg() and
forex(SemExpr bound | negBound(bound, v, pos) | negBoundOk(bound, v, pos))
or
result = TZero() and
forex(SemExpr bound | zeroBound(bound, v, pos) | zeroBoundOk(bound, v, pos))
}
/** Gets a possible sign for `v` at `pos`. */
private Sign semSsaSign(SemSsaVariable v, SemSsaReadPosition pos) {
result = unguardedSsaSign(v, pos)
or
result = guardedSsaSign(v, pos) and
result = guardedSsaSignOk(v, pos)
}
/** Gets a possible sign for `v`. */
pragma[nomagic]
Sign semSsaDefSign(SemSsaVariable v) { result = v.(SignDef).getSign() }
/** Gets a possible sign for `e`. */
cached
Sign semExprSign(SemExpr e) {
exists(Sign s | s = e.(SignExpr).getSign() |
if
Utils::getTrackedType(e) instanceof SemUnsignedIntegerType and
s = TNeg() and
not Specific::ignoreTypeRestrictions(e)
then result = TPos()
else result = s
)
}
/**
* Dummy predicate that holds for any sign. This is added to improve readability
* of cases where the sign is unrestricted.
*/
predicate semAnySign(Sign s) { any() }
/** Holds if `e` can be positive and cannot be negative. */
predicate semPositive(SemExpr e) {
semExprSign(e) = TPos() and
not semExprSign(e) = TNeg()
}
/** Holds if `e` can be negative and cannot be positive. */
predicate semNegative(SemExpr e) {
semExprSign(e) = TNeg() and
not semExprSign(e) = TPos()
}
/** Holds if `e` is strictly positive. */
predicate semStrictlyPositive(SemExpr e) {
semExprSign(e) = TPos() and
not semExprSign(e) = TNeg() and
not semExprSign(e) = TZero()
}
/** Holds if `e` is strictly negative. */
predicate semStrictlyNegative(SemExpr e) {
semExprSign(e) = TNeg() and
not semExprSign(e) = TPos() and
not semExprSign(e) = TZero()
}
}
/** An SSA definition whose sign is computed by a language-specific implementation. */
abstract class CustomSignDef extends SignDef {
abstract override Sign getSign();
}
/**
* An expression for which the analysis can compute the sign.
*
* The actual computation of the sign is done in an override of the `getSign()` predicate. The
* charpred of any subclass must _not_ invoke `getSign()`, directly or indirectly. This ensures
* that the charpred does not introduce negative recursion. The `getSign()` predicate may be
* recursive.
*
* Concrete implementations extend one of the following subclasses:
* - `ConstantSignExpr`, for expressions with a compile-time constant value.
* - `FlowSignExpr`, for expressions whose sign can be computed from the signs of their operands.
* - `CustomsignExpr`, for expressions shose sign can be computed by a language-specific
* implementation.
*
* If the same expression matches more than one of the above subclasses, the sign is computed as
* follows:
* - The sign of a `ConstantSignExpr` is computed solely from `ConstantSignExpr.getSign()`,
* regardless of any other subclasses.
* - If a non-`ConstantSignExpr` expression matches exactly one of `FlowSignExpr` or
* `CustomSignExpr`, the sign is computed by that class' `getSign()` predicate.
* - If a non-`ConstantSignExpr` expression matches both `FlowSignExpr` and `CustomSignExpr`, the
* sign is the _intersection_ of the signs of those two classes' `getSign()` predicates. Thus,
* both classes have the opportunity to _restrict_ the set of possible signs, not to generate new
* possible signs.
* - If an expression does not match any of the three subclasses, then it can have any sign.
*
* Note that the `getSign()` predicate is introduced only in subclasses of `SignExpr`.
*/
abstract class SignExpr instanceof SemExpr {
SignExpr() { not Specific::ignoreExprSign(this) }
final string toString() { result = super.toString() }
abstract Sign getSign();
}
/** An expression whose sign is determined by its constant numeric value. */
private class ConstantSignExpr extends SignExpr {
ConstantSignExpr() {
this instanceof SemConstantIntegerExpr or
exists(this.(SemNumericLiteralExpr).getApproximateFloatValue())
}
final override Sign getSign() {
exists(int i | this.(SemConstantIntegerExpr).getIntValue() = i |
i < 0 and result = TNeg()
or
i = 0 and result = TZero()
or
i > 0 and result = TPos()
)
or
not exists(this.(SemConstantIntegerExpr).getIntValue()) and
exists(float f | f = this.(SemNumericLiteralExpr).getApproximateFloatValue() |
f < 0 and result = TNeg()
or
f = 0 and result = TZero()
or
f > 0 and result = TPos()
)
}
}
abstract private class NonConstantSignExpr extends SignExpr {
NonConstantSignExpr() { not this instanceof ConstantSignExpr }
final override Sign getSign() {
// The result is the _intersection_ of the signs computed from flow and by the language.
(result = this.(FlowSignExpr).getSignRestriction() or not this instanceof FlowSignExpr) and
(result = this.(CustomSignExpr).getSignRestriction() or not this instanceof CustomSignExpr)
}
}
/** An expression whose sign is computed from the signs of its operands. */
abstract private class FlowSignExpr extends NonConstantSignExpr {
abstract Sign getSignRestriction();
}
/** An expression whose sign is computed by a language-specific implementation. */
abstract class CustomSignExpr extends NonConstantSignExpr {
abstract Sign getSignRestriction();
}
/** An expression whose sign is unknown. */
private class UnknownSignExpr extends SignExpr {
UnknownSignExpr() {
not this instanceof FlowSignExpr and
not this instanceof CustomSignExpr and
not this instanceof ConstantSignExpr and
(
// Only track numeric types.
getTrackedType(this) instanceof SemNumericType
or
// Unless the language says to track this expression anyway.
Specific::trackUnknownNonNumericExpr(this)
)
}
final override Sign getSign() { semAnySign(result) }
}
/**
* A `Load` expression whose sign is computed from the sign of its SSA definition, restricted by
* inference from any intervening guards.
*/
class UseSignExpr extends FlowSignExpr {
SemSsaVariable v;
UseSignExpr() { v.getAUse() = this }
override Sign getSignRestriction() {
// Propagate via SSA
// Propagate the sign from the def of `v`, incorporating any inference from guards.
result = semSsaSign(v, any(SemSsaReadPositionBlock bb | bb.getAnExpr() = this))
or
// No block for this read. Just use the sign of the def.
// REVIEW: How can this happen?
not exists(SemSsaReadPositionBlock bb | bb.getAnExpr() = this) and
result = semSsaDefSign(v)
}
}
/** A binary expression whose sign is computed from the signs of its operands. */
private class BinarySignExpr extends FlowSignExpr {
SemBinaryExpr binary;
BinarySignExpr() { binary = this }
override Sign getSignRestriction() {
exists(SemExpr left, SemExpr right |
binaryExprOperands(binary, left, right) and
result =
semExprSign(pragma[only_bind_out](left))
.applyBinaryOp(semExprSign(pragma[only_bind_out](right)), binary.getOpcode())
)
or
exists(SemDivExpr div | div = binary |
result = semExprSign(div.getLeftOperand()) and
result != TZero() and
div.getRightOperand().(SemFloatingPointLiteralExpr).getFloatValue() = 0
)
}
}
pragma[nomagic]
private predicate binaryExprOperands(SemBinaryExpr binary, SemExpr left, SemExpr right) {
binary.getLeftOperand() = left and binary.getRightOperand() = right
}
/**
* A `Convert`, `Box`, or `Unbox` expression.
*/
private class SemCastExpr extends SemUnaryExpr {
SemCastExpr() {
this instanceof SemConvertExpr
or
this instanceof SemBoxExpr
or
this instanceof SemUnboxExpr
}
}
/** A unary expression whose sign is computed from the sign of its operand. */
private class UnarySignExpr extends FlowSignExpr {
SemUnaryExpr unary;
UnarySignExpr() { unary = this and not this instanceof SemCastExpr }
override Sign getSignRestriction() {
result = semExprSign(pragma[only_bind_out](unary.getOperand())).applyUnaryOp(unary.getOpcode())
}
}
/**
* A `Convert`, `Box`, or `Unbox` expression, whose sign is computed based on
* the sign of its operand and the source and destination types.
*/
abstract private class CastSignExpr extends FlowSignExpr {
SemUnaryExpr cast;
CastSignExpr() { cast = this and cast instanceof SemCastExpr }
override Sign getSignRestriction() { result = semExprSign(cast.getOperand()) }
}
/**
* A `Convert` expression.
*/
private class ConvertSignExpr extends CastSignExpr {
override SemConvertExpr cast;
}
/**
* A `Box` expression.
*/
private class BoxSignExpr extends CastSignExpr {
override SemBoxExpr cast;
}
/**
* An `Unbox` expression.
*/
private class UnboxSignExpr extends CastSignExpr {
override SemUnboxExpr cast;
UnboxSignExpr() {
exists(SemType fromType | fromType = getTrackedType(cast.getOperand()) |
// Only numeric source types are handled here.
fromType instanceof SemNumericType
)
}
}
private predicate unknownSign(SemExpr e) { e instanceof UnknownSignExpr }
/**
* Holds if `lowerbound` is a lower bound for `v` at `pos`. This is restricted
* to only include bounds for which we might determine a sign.
*/
private predicate lowerBound(
SemExpr lowerbound, SemSsaVariable v, SemSsaReadPosition pos, boolean isStrict
) {
exists(boolean testIsTrue, SemRelationalExpr comp |
pos.hasReadOfVar(v) and
semGuardControlsSsaRead(semGetComparisonGuard(comp), pos, testIsTrue) and
not unknownSign(lowerbound)
|
testIsTrue = true and
comp.getLesserOperand() = lowerbound and
comp.getGreaterOperand() = semSsaRead(v, 0) and
(if comp.isStrict() then isStrict = true else isStrict = false)
or
testIsTrue = false and
comp.getGreaterOperand() = lowerbound and
comp.getLesserOperand() = semSsaRead(v, 0) and
(if comp.isStrict() then isStrict = false else isStrict = true)
)
}
/**
* Holds if `upperbound` is an upper bound for `v` at `pos`. This is restricted
* to only include bounds for which we might determine a sign.
*/
private predicate upperBound(
SemExpr upperbound, SemSsaVariable v, SemSsaReadPosition pos, boolean isStrict
) {
exists(boolean testIsTrue, SemRelationalExpr comp |
pos.hasReadOfVar(v) and
semGuardControlsSsaRead(semGetComparisonGuard(comp), pos, testIsTrue) and
not unknownSign(upperbound)
|
testIsTrue = true and
comp.getGreaterOperand() = upperbound and
comp.getLesserOperand() = semSsaRead(v, 0) and
(if comp.isStrict() then isStrict = true else isStrict = false)
or
testIsTrue = false and
comp.getLesserOperand() = upperbound and
comp.getGreaterOperand() = semSsaRead(v, 0) and
(if comp.isStrict() then isStrict = false else isStrict = true)
)
}
/**
* Holds if `eqbound` is an equality/inequality for `v` at `pos`. This is
* restricted to only include bounds for which we might determine a sign. The
* boolean `isEq` gives the polarity:
* - `isEq = true` : `v = eqbound`
* - `isEq = false` : `v != eqbound`
*/
private predicate eqBound(SemExpr eqbound, SemSsaVariable v, SemSsaReadPosition pos, boolean isEq) {
exists(SemGuard guard, boolean testIsTrue, boolean polarity |
pos.hasReadOfVar(v) and
semGuardControlsSsaRead(guard, pos, testIsTrue) and
guard.isEquality(eqbound, semSsaRead(v, 0), polarity) and
isEq = polarity.booleanXor(testIsTrue).booleanNot() and
not unknownSign(eqbound)
)
}
/**
* Holds if `bound` is a bound for `v` at `pos` that needs to be positive in
* order for `v` to be positive.
*/
private predicate posBound(SemExpr bound, SemSsaVariable v, SemSsaReadPosition pos) {
upperBound(bound, v, pos, _) or
eqBound(bound, v, pos, true)
}
/**
* Holds if `bound` is a bound for `v` at `pos` that needs to be negative in
* order for `v` to be negative.
*/
private predicate negBound(SemExpr bound, SemSsaVariable v, SemSsaReadPosition pos) {
lowerBound(bound, v, pos, _) or
eqBound(bound, v, pos, true)
}
/**
* Holds if `bound` is a bound for `v` at `pos` that can restrict whether `v`
* can be zero.
*/
private predicate zeroBound(SemExpr bound, SemSsaVariable v, SemSsaReadPosition pos) {
lowerBound(bound, v, pos, _) or
upperBound(bound, v, pos, _) or
eqBound(bound, v, pos, _)
}
/** Holds if `bound` allows `v` to be positive at `pos`. */
private predicate posBoundOk(SemExpr bound, SemSsaVariable v, SemSsaReadPosition pos) {
posBound(bound, v, pos) and TPos() = semExprSign(bound)
}
/** Holds if `bound` allows `v` to be negative at `pos`. */
private predicate negBoundOk(SemExpr bound, SemSsaVariable v, SemSsaReadPosition pos) {
negBound(bound, v, pos) and TNeg() = semExprSign(bound)
}
/** Holds if `bound` allows `v` to be zero at `pos`. */
private predicate zeroBoundOk(SemExpr bound, SemSsaVariable v, SemSsaReadPosition pos) {
lowerBound(bound, v, pos, _) and TNeg() = semExprSign(bound)
or
lowerBound(bound, v, pos, false) and TZero() = semExprSign(bound)
or
upperBound(bound, v, pos, _) and TPos() = semExprSign(bound)
or
upperBound(bound, v, pos, false) and TZero() = semExprSign(bound)
or
eqBound(bound, v, pos, true) and TZero() = semExprSign(bound)
or
eqBound(bound, v, pos, false) and TZero() != semExprSign(bound)
}
/**
* Holds if there is a bound that might restrict whether `v` has the sign `s`
* at `pos`.
*/
private predicate hasGuard(SemSsaVariable v, SemSsaReadPosition pos, Sign s) {
s = TPos() and posBound(_, v, pos)
or
s = TNeg() and negBound(_, v, pos)
or
s = TZero() and zeroBound(_, v, pos)
}
/**
* Gets a possible sign of `v` at `pos` based on its definition, where the sign
* might be ruled out by a guard.
*/
pragma[noinline]
private Sign guardedSsaSign(SemSsaVariable v, SemSsaReadPosition pos) {
result = semSsaDefSign(v) and
pos.hasReadOfVar(v) and
hasGuard(v, pos, result)
}
/**
* Gets a possible sign of `v` at `pos` based on its definition, where no guard
* can rule it out.
*/
pragma[noinline]
private Sign unguardedSsaSign(SemSsaVariable v, SemSsaReadPosition pos) {
result = semSsaDefSign(v) and
pos.hasReadOfVar(v) and
not hasGuard(v, pos, result)
}
/**
* Gets a possible sign of `v` at read position `pos`, where a guard could have
* ruled out the sign but does not.
* This does not check that the definition of `v` also allows the sign.
*/
private Sign guardedSsaSignOk(SemSsaVariable v, SemSsaReadPosition pos) {
result = TPos() and
forex(SemExpr bound | posBound(bound, v, pos) | posBoundOk(bound, v, pos))
or
result = TNeg() and
forex(SemExpr bound | negBound(bound, v, pos) | negBoundOk(bound, v, pos))
or
result = TZero() and
forex(SemExpr bound | zeroBound(bound, v, pos) | zeroBoundOk(bound, v, pos))
}
/** Gets a possible sign for `v` at `pos`. */
private Sign semSsaSign(SemSsaVariable v, SemSsaReadPosition pos) {
result = unguardedSsaSign(v, pos)
or
result = guardedSsaSign(v, pos) and
result = guardedSsaSignOk(v, pos)
}
/** Gets a possible sign for `v`. */
pragma[nomagic]
Sign semSsaDefSign(SemSsaVariable v) { result = v.(SignDef).getSign() }
/** Gets a possible sign for `e`. */
cached
Sign semExprSign(SemExpr e) {
exists(Sign s | s = e.(SignExpr).getSign() |
if
getTrackedType(e) instanceof SemUnsignedIntegerType and
s = TNeg() and
not Specific::ignoreTypeRestrictions(e)
then result = TPos()
else result = s
)
}
/**
* Dummy predicate that holds for any sign. This is added to improve readability
* of cases where the sign is unrestricted.
*/
predicate semAnySign(Sign s) { any() }
/** Holds if `e` can be positive and cannot be negative. */
predicate semPositive(SemExpr e) {
semExprSign(e) = TPos() and
not semExprSign(e) = TNeg()
}
/** Holds if `e` can be negative and cannot be positive. */
predicate semNegative(SemExpr e) {
semExprSign(e) = TNeg() and
not semExprSign(e) = TPos()
}
/** Holds if `e` is strictly positive. */
predicate semStrictlyPositive(SemExpr e) {
semExprSign(e) = TPos() and
not semExprSign(e) = TNeg() and
not semExprSign(e) = TZero()
}
/** Holds if `e` is strictly negative. */
predicate semStrictlyNegative(SemExpr e) {
semExprSign(e) = TNeg() and
not semExprSign(e) = TPos() and
not semExprSign(e) = TZero()
}

Some files were not shown because too many files have changed in this diff Show More