Build binaries compatible with cf-protection

This works around a weird error I was getting when trying to link a
translation unit that included Internal.h with libconflict-set-static.a
with clang + gnu objcopy

.gitea/workflows/ci.yml

@@ -1,282 +0,0 @@
-name: CI
-
-on: [push, pull_request]
-
-jobs:
-  build-image:
-    strategy:
-      fail-fast: false
-      matrix:
-        include:
-          - runner: ubuntu-latest-amd64
-            arch: amd64
-          - runner: ubuntu-latest-arm64
-            arch: arm64
-    runs-on: ${{ matrix.runner }}
-    steps:
-      - uses: actions/checkout@v4
-
-      - name: Log in to registry
-        env:
-          REGISTRY_USER: ${{ secrets.REGISTRY_USER }}
-          REGISTRY_TOKEN: ${{ secrets.REGISTRY_TOKEN }}
-        run: |
-          echo "$REGISTRY_TOKEN" \
-            | docker login -u "$REGISTRY_USER" --password-stdin git.weaselab.dev
-
-      - name: Build and push image if changed
-        run: |
-          image=git.weaselab.dev/weaselab/conflict-set-ci
-          hash="$(sha256sum Dockerfile .pre-commit-config.yaml | sha256sum | cut -c 1-16)"
-          latest="$image:latest-${{ matrix.arch }}"
-          current="$(docker buildx imagetools inspect "$latest" \
-            --format '{{index .Image.Config.Labels "dev.weaselab.ci-hash"}}' 2> /dev/null || true)"
-          if [ "$current" = "$hash" ]; then
-            echo "$latest is up to date"
-          else
-            docker build --push --label "dev.weaselab.ci-hash=$hash" -t "$latest" .
-          fi
-
-  pre-commit:
-    needs: build-image
-    runs-on: ubuntu-latest-amd64
-    container:
-      image: git.weaselab.dev/weaselab/conflict-set-ci:latest-amd64
-      credentials:
-        username: ${{ secrets.REGISTRY_USER }}
-        password: ${{ secrets.REGISTRY_TOKEN }}
-    steps:
-      - uses: actions/checkout@v4
-
-      - name: Run pre-commit
-        env:
-          # use the hooks pre-installed in the image
-          HOME: /tmp
-        run: |
-          git config --global --add safe.directory "$PWD"
-          pre-commit run --all-files --show-diff-on-failure
-
-  test:
-    needs: build-image
-    strategy:
-      fail-fast: false
-      matrix:
-        include:
-          - name: 64-bit-versions
-            cmake_args: -DCMAKE_CXX_FLAGS=-DUSE_64_BIT=1
-          - name: debug
-            cmake_args: -DCMAKE_BUILD_TYPE=Debug
-          - name: simd-fallback
-            cmake_args: -DUSE_SIMD_FALLBACK=ON
-          - name: gcc
-            cmake_args: -DCMAKE_C_COMPILER=gcc -DCMAKE_CXX_COMPILER=g++
-    runs-on: ubuntu-latest-amd64
-    container:
-      image: git.weaselab.dev/weaselab/conflict-set-ci:latest-amd64
-      credentials:
-        username: ${{ secrets.REGISTRY_USER }}
-        password: ${{ secrets.REGISTRY_TOKEN }}
-    steps:
-      - uses: actions/checkout@v4
-
-      - uses: actions/cache@v4
-        with:
-          path: .ccache
-          key: ccache-${{ matrix.name }}-${{ gitea.sha }}
-          restore-keys: |
-            ccache-${{ matrix.name }}-
-
-      - name: Build
-        run: |
-          export CCACHE_DIR="$GITHUB_WORKSPACE/.ccache"
-          rm -rf build
-          cmake -S . -B build -G Ninja -DCMAKE_CXX_COMPILER_LAUNCHER=ccache ${{ matrix.cmake_args }}
-          ninja -C build
-          ccache -s
-
-      - name: Test
-        run: |
-          cd build
-          ctest --no-compress-output --test-output-size-passed 100000 --test-output-size-failed 100000 -T Test -j "$(nproc)" --timeout 90 > /dev/null
-
-      - name: Upload test results to MinIO
-        if: always()
-        env:
-          MINIO_ACCESS_KEY: ${{ secrets.MINIO_ACCESS_KEY }}
-          MC_HOST_minio: https://${{ secrets.MINIO_ACCESS_KEY }}:${{ secrets.MINIO_SECRET_KEY }}@minio.weaselab.dev
-        run: |
-          if [ -z "$MINIO_ACCESS_KEY" ]; then
-            echo "MinIO credentials not configured; skipping upload"
-            exit 0
-          fi
-          zstd build/Testing/*/Test.xml
-          mc cp build/Testing/*/Test.xml.zst "minio/jenkins/conflict-set/${{ gitea.run_number }}/${{ matrix.name }}/"
-
-      - name: Test summary
-        if: always()
-        run: |
-          python3 ctest_summary.py build/Testing/*/Test.xml \
-            --link "https://minio.weaselab.dev/jenkins/conflict-set/${{ gitea.run_number }}/${{ matrix.name }}/Test.xml.zst" \
-            | tee -a "$GITHUB_STEP_SUMMARY"
-
-  release:
-    needs: build-image
-    strategy:
-      fail-fast: false
-      matrix:
-        include:
-          - runner: ubuntu-latest-amd64
-            arch: amd64
-          - runner: ubuntu-latest-arm64
-            arch: arm64
-    runs-on: ${{ matrix.runner }}
-    container:
-      image: git.weaselab.dev/weaselab/conflict-set-ci:latest-${{ matrix.arch }}
-      credentials:
-        username: ${{ secrets.REGISTRY_USER }}
-        password: ${{ secrets.REGISTRY_TOKEN }}
-    steps:
-      - uses: actions/checkout@v4
-
-      - uses: actions/cache@v4
-        with:
-          path: .ccache
-          key: ccache-release-${{ matrix.arch }}-${{ gitea.sha }}
-          restore-keys: |
-            ccache-release-${{ matrix.arch }}-
-
-      - name: Build
-        run: |
-          export CCACHE_DIR="$GITHUB_WORKSPACE/.ccache"
-          rm -rf build
-          cmake -S . -B build -G Ninja -DCMAKE_CXX_COMPILER_LAUNCHER=ccache -DCMAKE_CXX_FLAGS=-DNVALGRIND
-          ninja -C build
-          ccache -s
-
-      - name: Test
-        run: |
-          cd build
-          # On arm64, valgrind needs the MAKE_MEM_DEFINED client requests for
-          # https://git.weaselab.dev/weaselab/conflict-set/issues/39, but this
-          # build has -DNVALGRIND, which compiles them out. Skip valgrind
-          # tests here; they run annotated in the test job.
-          ctest --no-compress-output --test-output-size-passed 100000 --test-output-size-failed 100000 ${{ matrix.arch == 'arm64' && '-E valgrind' || '' }} -T Test -j "$(nproc)" --timeout 90 > /dev/null
-
-      - name: Package
-        run: |
-          cd build
-          cpack -G DEB
-          cpack -G RPM
-
-      - name: Build paper
-        if: matrix.arch == 'amd64'
-        run: |
-          cd paper
-          make
-
-      - name: Upload artifacts to MinIO
-        if: always()
-        shell: bash
-        env:
-          MINIO_ACCESS_KEY: ${{ secrets.MINIO_ACCESS_KEY }}
-          MC_HOST_minio: https://${{ secrets.MINIO_ACCESS_KEY }}:${{ secrets.MINIO_SECRET_KEY }}@minio.weaselab.dev
-        run: |
-          if [ -z "$MINIO_ACCESS_KEY" ]; then
-            echo "MinIO credentials not configured; skipping upload"
-            exit 0
-          fi
-          dest="minio/jenkins/conflict-set/${{ gitea.run_number }}/release-${{ matrix.arch }}/"
-          zstd build/Testing/*/Test.xml
-          mc cp build/Testing/*/Test.xml.zst "$dest"
-          # This step runs even when a previous step failed, to upload test
-          # results. The packages may never have been built though, so skip
-          # them if they're missing.
-          if compgen -G "build/*.deb" > /dev/null; then
-            mc cp build/*.deb "$dest"
-          fi
-          if compgen -G "build/*.rpm" > /dev/null; then
-            mc cp build/*.rpm "$dest"
-          fi
-          if compgen -G "paper/*.pdf" > /dev/null; then
-            mc cp paper/*.pdf "$dest"
-          fi
-
-      - name: Test summary
-        if: always()
-        run: |
-          python3 ctest_summary.py build/Testing/*/Test.xml \
-            --link "https://minio.weaselab.dev/jenkins/conflict-set/${{ gitea.run_number }}/release-${{ matrix.arch }}/Test.xml.zst" \
-            | tee -a "$GITHUB_STEP_SUMMARY"
-
-  coverage:
-    needs: build-image
-    runs-on: ubuntu-latest-amd64
-    container:
-      image: git.weaselab.dev/weaselab/conflict-set-ci:latest-amd64
-      credentials:
-        username: ${{ secrets.REGISTRY_USER }}
-        password: ${{ secrets.REGISTRY_TOKEN }}
-    steps:
-      - uses: actions/checkout@v4
-
-      - uses: actions/cache@v4
-        with:
-          path: .ccache
-          key: ccache-coverage-${{ gitea.sha }}
-          restore-keys: |
-            ccache-coverage-
-
-      - name: Build
-        run: |
-          export CCACHE_DIR="$GITHUB_WORKSPACE/.ccache"
-          rm -rf build
-          cmake -S . -B build -G Ninja -DCMAKE_CXX_COMPILER_LAUNCHER=ccache \
-            -DCMAKE_C_FLAGS=--coverage -DCMAKE_CXX_FLAGS=--coverage \
-            -DCMAKE_BUILD_TYPE=Debug -DDISABLE_TSAN=ON
-          ninja -C build
-          ccache -s
-
-      - name: Test
-        run: |
-          cd build
-          ctest --no-compress-output --test-output-size-passed 100000 --test-output-size-failed 100000 -T Test -j "$(nproc)" --timeout 90 > /dev/null
-
-      - name: Coverage report
-        shell: bash
-        run: |
-          gcov_args=(-f ConflictSet.cpp -f LongestCommonPrefix.h -f Metrics.h
-                     --gcov-executable "llvm-cov gcov" --exclude-noncode-lines)
-          gcovr "${gcov_args[@]}" --cobertura > build/coverage.xml
-          gcovr "${gcov_args[@]}"
-          mkdir -p build/coverage_html
-          gcovr "${gcov_args[@]}" --html-details build/coverage_html/index.html
-          gcovr "${gcov_args[@]}" --fail-under-line 100 > /dev/null
-
-      - name: Upload results to MinIO
-        if: always()
-        env:
-          MINIO_ACCESS_KEY: ${{ secrets.MINIO_ACCESS_KEY }}
-          MC_HOST_minio: https://${{ secrets.MINIO_ACCESS_KEY }}:${{ secrets.MINIO_SECRET_KEY }}@minio.weaselab.dev
-        run: |
-          if [ -z "$MINIO_ACCESS_KEY" ]; then
-            echo "MinIO credentials not configured; skipping upload"
-            exit 0
-          fi
-          dest="minio/jenkins/conflict-set/${{ gitea.run_number }}/coverage/"
-          zstd build/Testing/*/Test.xml
-          mc cp build/Testing/*/Test.xml.zst "$dest"
-          if [ -e build/coverage.xml ]; then
-            mc cp build/coverage.xml "$dest"
-          fi
-          if [ -d build/coverage_html ]; then
-            mc cp -r build/coverage_html "$dest"
-          fi
-
-      - name: Test summary
-        if: always()
-        run: |
-          python3 ctest_summary.py build/Testing/*/Test.xml \
-            --link "https://minio.weaselab.dev/jenkins/conflict-set/${{ gitea.run_number }}/coverage/Test.xml.zst" \
-            | tee -a "$GITHUB_STEP_SUMMARY"
-          echo "" | tee -a "$GITHUB_STEP_SUMMARY"
-          echo "📊 [Coverage report](https://minio.weaselab.dev/jenkins/conflict-set/${{ gitea.run_number }}/coverage/coverage_html/index.html)" | tee -a "$GITHUB_STEP_SUMMARY"

Bench.cpp

@@ -57,7 +57,6 @@ ConflictSet::ReadRange prefixRange(Arena &arena, TrivialSpan key) {
 
 void benchConflictSet() {
   ankerl::nanobench::Bench bench;
-  bench.minEpochIterations(10000);
   ConflictSet cs{0};
 
   bench.batch(kOpsPerTx);

CMakeLists.txt

@@ -32,12 +32,7 @@ if(NOT CMAKE_BUILD_TYPE AND NOT CMAKE_CONFIGURATION_TYPES)
 endif()
 
 add_compile_options(
-  -Werror=switch-enum
-  -Wswitch-enum
-  -Wunused-variable
-  -fPIC
-  -fdata-sections
-  -ffunction-sections
+  -Werror=switch-enum -Wswitch-enum -fPIC -fdata-sections -ffunction-sections
   -fno-jump-tables # https://github.com/llvm/llvm-project/issues/54247
 )
 
@@ -53,7 +48,6 @@ endif()
 
 if(CMAKE_CXX_COMPILER_ID STREQUAL "GNU")
   add_compile_options("-Wno-maybe-uninitialized")
-  add_compile_options("-Wno-maybe-musttail-local-addr")
 endif()
 
 if(NOT APPLE)
@@ -61,21 +55,16 @@ if(NOT APPLE)
   add_compile_options(-g -fno-omit-frame-pointer)
 endif()
 
-set(relro_flags "LINKER:-z,relro,-z,now,-z,noexecstack")
-set(full_relro_flags "-pie;${relro_flags}")
+set(full_relro_flags "-pie;LINKER:-z,relro,-z,now,-z,noexecstack")
 cmake_push_check_state()
 list(APPEND CMAKE_REQUIRED_LINK_OPTIONS ${full_relro_flags})
 check_cxx_source_compiles("int main(){}" HAS_FULL_RELRO FAIL_REGEX "warning:")
 if(HAS_FULL_RELRO)
-  # -pie only applies to executables; passing it when linking a shared library
-  # makes the driver pull in Scrt1.o, which requires main.
-  add_link_options("$<$<STREQUAL:$<TARGET_PROPERTY:TYPE>,EXECUTABLE>:-pie>"
-                   ${relro_flags})
+  add_link_options(${full_relro_flags})
 endif()
 cmake_pop_check_state()
 
-if(CMAKE_SYSTEM_PROCESSOR STREQUAL aarch64 OR CMAKE_SYSTEM_PROCESSOR STREQUAL
-                                              arm64)
+if(CMAKE_SYSTEM_PROCESSOR STREQUAL aarch64)
   add_compile_options(-mbranch-protection=standard)
 else()
   add_compile_options(-fcf-protection)
@@ -107,17 +96,27 @@ option(DISABLE_TSAN "Disable TSAN" OFF)
 # https://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.clientreq
 include_directories(SYSTEM ${CMAKE_CURRENT_SOURCE_DIR}/third_party/valgrind)
 
+add_compile_options($<$<COMPILE_LANGUAGE:CXX>:-Wno-invalid-offsetof>)
+
 if(APPLE)
   add_link_options(-Wl,-dead_strip)
 else()
   add_link_options(-Wl,--gc-sections)
 endif()
 
-if(USE_SIMD_FALLBACK)
-  add_compile_definitions(USE_SIMD_FALLBACK)
-else()
-  if(CMAKE_SYSTEM_PROCESSOR STREQUAL x86_64)
+if(NOT USE_SIMD_FALLBACK)
+  cmake_push_check_state()
+  list(APPEND CMAKE_REQUIRED_FLAGS -mavx)
+  check_include_file_cxx("immintrin.h" HAS_AVX)
+  if(HAS_AVX)
     add_compile_options(-mavx)
+    add_compile_definitions(HAS_AVX)
+  endif()
+  cmake_pop_check_state()
+
+  check_include_file_cxx("arm_neon.h" HAS_ARM_NEON)
+  if(HAS_ARM_NEON)
+    add_compile_definitions(HAS_ARM_NEON)
   endif()
 endif()
 
@@ -347,7 +346,8 @@ if(CMAKE_SOURCE_DIR STREQUAL CMAKE_CURRENT_SOURCE_DIR AND BUILD_TESTING)
   # c++98
   add_executable(conflict_set_cxx_api_test conflict_set_cxx_api_test.cpp)
   target_compile_options(conflict_set_cxx_api_test PRIVATE ${TEST_FLAGS})
-  target_link_libraries(conflict_set_cxx_api_test PRIVATE ${PROJECT_NAME})
+  target_link_libraries(conflict_set_cxx_api_test
+                        PRIVATE ${PROJECT_NAME}-static)
   set_target_properties(conflict_set_cxx_api_test PROPERTIES CXX_STANDARD 98)
   set_target_properties(conflict_set_cxx_api_test
                         PROPERTIES CXX_STANDARD_REQUIRED ON)
@@ -380,22 +380,11 @@ if(CMAKE_SOURCE_DIR STREQUAL CMAKE_CURRENT_SOURCE_DIR AND BUILD_TESTING)
         ${symbol_imports})
   endif()
 
-  if(NOT CMAKE_CROSSCOMPILING)
   find_program(HARDENING_CHECK hardening-check)
   if(HARDENING_CHECK)
-      # Control flow integrity (CET) is x86-only and branch protection (PAC/BTI)
-      # is arm64-only, so ignore whichever doesn't apply.
-      if(CMAKE_SYSTEM_PROCESSOR STREQUAL aarch64 OR CMAKE_SYSTEM_PROCESSOR
-                                                    STREQUAL arm64)
-        set(hardening_check_arch_flags --nocfprotection)
-      else()
-        set(hardening_check_arch_flags --nobranchprotection)
-      endif()
-      add_test(
-        NAME hardening_check
-        COMMAND ${HARDENING_CHECK} $<TARGET_FILE:${PROJECT_NAME}> --nofortify
-                --nostackprotector ${hardening_check_arch_flags})
-    endif()
+    add_test(NAME hardening_check
+             COMMAND ${HARDENING_CHECK} $<TARGET_FILE:${PROJECT_NAME}>
+                     --nofortify --nostackprotector)
   endif()
 
   # bench

Dockerfile

@@ -8,36 +8,48 @@ RUN chmod -R 777 /tmp
 RUN apt-get update
 RUN apt-get upgrade -y
 RUN TZ=America/Los_Angeles DEBIAN_FRONTEND=noninteractive apt-get install -y \
-    biber \
+    binutils-aarch64-linux-gnu \
     build-essential \
     ccache \
-    clang \
     cmake \
     curl \
     devscripts \
+    g++-aarch64-linux-gnu \
     gcovr \
     git \
-    latexmk \
+    gnupg \
     libc6-dbg \
-    llvm \
+    lsb-release \
     mold \
     ninja-build \
-    nodejs \
     pre-commit \
     python3-requests \
+    qemu-user \
     rpm \
-    texlive-bibtex-extra \
-    texlive-fonts-recommended \
-    texlive-latex-extra \
-    texlive-pictures \
-    valgrind \
+    software-properties-common \
+    texlive-full \
     wget \
     zstd
 
-# MinIO client, for uploading build artifacts
-RUN curl -Ls "https://dl.min.io/client/mc/release/linux-$(dpkg --print-architecture)/mc" \
-    -o /usr/local/bin/mc && chmod +x /usr/local/bin/mc
+# Install recent valgrind from source
+RUN curl -Ls https://sourceware.org/pub/valgrind/valgrind-3.22.0.tar.bz2 -o valgrind.tar.bz2 && \
+    echo "c811db5add2c5f729944caf47c4e7a65dcaabb9461e472b578765dd7bf6d2d4c  valgrind.tar.bz2" > valgrind-sha.txt && \
+    sha256sum --quiet -c valgrind-sha.txt && \
+    mkdir valgrind && \
+    tar --strip-components 1 --no-same-owner --no-same-permissions --directory valgrind -xjf valgrind.tar.bz2 && \
+    cd valgrind && \
+    ./configure --enable-only64bit --enable-lto && \
+    make -j`nproc` && \
+    make install && \
+    cd .. && \
+    rm -rf /tmp/*
 
+# Recent clang
+RUN wget https://apt.llvm.org/llvm.sh && chmod +x ./llvm.sh && ./llvm.sh 20
+
+RUN apt-get -y install clang llvm
+
+# Set after building valgrind, which doesn't build with clang for some reason
 ENV CC=clang
 ENV CXX=clang++
 

Internal.h

@@ -18,6 +18,7 @@ using namespace weaselab;
 #include <span>
 #include <string>
 #include <thread>
+#include <unordered_set>
 #include <utility>
 
 #include <callgrind.h>
@@ -367,6 +368,23 @@ template <class T, class C = std::less<T>> auto set(Arena &arena) {
   return Set<T, C>(ArenaAlloc<T>(&arena));
 }
 
+template <class T> struct MyHash;
+
+template <class T> struct MyHash<T *> {
+  size_t operator()(const T *t) const noexcept {
+    size_t result;
+    memcpy(&result, &t, sizeof(result));
+    return result;
+  }
+};
+
+template <class T>
+using HashSet =
+    std::unordered_set<T, MyHash<T>, std::equal_to<T>, ArenaAlloc<T>>;
+template <class T> auto hashSet(Arena &arena) {
+  return HashSet<T>(ArenaAlloc<T>(&arena));
+}
+
 template <class T, class U>
 bool operator==(const ArenaAlloc<T> &lhs, const ArenaAlloc<U> &rhs) {
   return lhs.arena == rhs.arena;

Jenkinsfile

@@ -0,0 +1,151 @@
+def CleanBuildAndTest(String cmakeArgs) {
+    sh """
+    export CCACHE_DIR=/ccache
+    rm -rf build
+    mkdir build
+    cd build
+    cmake .. -G Ninja -DCMAKE_CXX_COMPILER_LAUNCHER=ccache ${cmakeArgs} 
+    ninja
+    ccache -s
+    """
+    catchError {
+        sh '''
+        cd build
+        ctest --no-compress-output --test-output-size-passed 100000 --test-output-size-failed 100000 -T Test -j `nproc` --timeout 90 > /dev/null
+        zstd Testing/*/Test.xml
+        '''
+    }
+    xunit tools: [CTest(pattern: 'build/Testing/*/Test.xml')], skipPublishingChecks: false
+    minio bucket: 'jenkins', credentialsId: 'jenkins-minio', excludes: '', host: 'minio.weaselab.dev', includes: 'build/Testing/*/Test.xml.zst', targetFolder: '${JOB_NAME}/${BUILD_NUMBER}/${STAGE_NAME}/'
+}
+
+pipeline {
+    agent any
+    stages {
+        stage('Pre-commit') {
+            agent {
+                dockerfile {
+                    args '-v /home/jenkins/ccache:/ccache'
+                    reuseNode true
+                }
+            }
+            steps {
+                script {
+                    env.HOME = env.WORKSPACE
+                }
+                sh 'pre-commit run --all-files --show-diff-on-failure'
+            }
+        }
+        stage('64 bit versions') {
+            agent {
+                dockerfile {
+                    args '-v /home/jenkins/ccache:/ccache'
+                    reuseNode true
+                }
+            }
+            steps {
+                CleanBuildAndTest("-DCMAKE_CXX_FLAGS=-DUSE_64_BIT=1")
+            }
+        }
+        stage('Debug') {
+            agent {
+                dockerfile {
+                    args '-v /home/jenkins/ccache:/ccache'
+                    reuseNode true
+                }
+            }
+            steps {
+                CleanBuildAndTest("-DCMAKE_BUILD_TYPE=Debug")
+            }
+        }
+        stage('SIMD fallback') {
+            agent {
+                dockerfile {
+                    args '-v /home/jenkins/ccache:/ccache'
+                    reuseNode true
+                }
+            }
+            steps {
+                CleanBuildAndTest("-DUSE_SIMD_FALLBACK=ON")
+            }
+        }
+        stage('Release [clang]') {
+            agent {
+                dockerfile {
+                    args '-v /home/jenkins/ccache:/ccache'
+                    reuseNode true
+                }
+            }
+            steps {
+                CleanBuildAndTest("-DCMAKE_CXX_FLAGS=-DNVALGRIND")
+                recordIssues(tools: [clang()])
+                sh '''
+                cd build
+                cpack -G DEB
+                cpack -G RPM
+                '''
+                sh '''
+                cd paper
+                make
+                '''
+                minio bucket: 'jenkins', credentialsId: 'jenkins-minio', excludes: '', host: 'minio.weaselab.dev', includes: 'build/*.deb,build/*.rpm,paper/*.pdf', targetFolder: '${JOB_NAME}/${BUILD_NUMBER}/${STAGE_NAME}/'
+            }
+        }
+        stage('Release [gcc]') {
+            agent {
+                dockerfile {
+                    args '-v /home/jenkins/ccache:/ccache'
+                    reuseNode true
+                }
+            }
+            steps {
+                CleanBuildAndTest("-DCMAKE_C_COMPILER=gcc -DCMAKE_CXX_COMPILER=g++ -DCMAKE_CXX_FLAGS=-DNVALGRIND")
+                recordIssues(tools: [gcc()])
+            }
+        }
+        stage('Release [clang,aarch64]') {
+            agent {
+                dockerfile {
+                    args '-v /home/jenkins/ccache:/ccache'
+                    reuseNode true
+                }
+            }
+            steps {
+                CleanBuildAndTest("-DCMAKE_TOOLCHAIN_FILE=../aarch64-toolchain.cmake -DCMAKE_CXX_FLAGS=-DNVALGRIND")
+                sh '''
+                cd build
+                cpack -G DEB
+                cpack -G RPM
+                '''
+                minio bucket: 'jenkins', credentialsId: 'jenkins-minio', excludes: '', host: 'minio.weaselab.dev', includes: 'build/*.deb,build/*.rpm', targetFolder: '${JOB_NAME}/${BUILD_NUMBER}/${STAGE_NAME}'
+            }
+        }
+        stage('Coverage') {
+            agent {
+                dockerfile {
+                    args '-v /home/jenkins/ccache:/ccache'
+                    reuseNode true
+                }
+            }
+            steps {
+                script {
+                    gcov_args = "-f ConflictSet.cpp -f LongestCommonPrefix.h -f Metrics.h --gcov-executable 'llvm-cov gcov' --exclude-noncode-lines"
+                }
+                CleanBuildAndTest("-DCMAKE_C_FLAGS=--coverage -DCMAKE_CXX_FLAGS=--coverage -DCMAKE_BUILD_TYPE=Debug -DDISABLE_TSAN=ON")
+                sh """
+                gcovr ${gcov_args} --cobertura > build/coverage.xml
+                """
+                recordCoverage qualityGates: [[criticality: 'NOTE', metric: 'MODULE']], tools: [[parser: 'COBERTURA', pattern: 'build/coverage.xml']]
+                sh """
+                gcovr ${gcov_args}
+                gcovr ${gcov_args} --fail-under-line 100 > /dev/null
+                """
+            }
+        }
+    }
+    post {
+        always {
+            emailext mimeType: 'text/html', body: '${SCRIPT, template="groovy-html.template"}', subject: "${env.JOB_NAME} - Build# ${env.BUILD_NUMBER} - ${currentBuild.currentResult}", to: 'andrew@weaselab.dev'
+        }
+    }
+}

README.md

@@ -4,14 +4,7 @@ Intended as an alternative to FoundationDB's skip list.
 
 Hardware for all benchmarks is an AMD Ryzen 9 7900 with (2x32GB) 5600MT/s CL28-34-34-89 1.35V RAM.
 
-```
-$ clang++ --version
-
-Ubuntu clang version 20.0.0 (++20241120082228+86734c857724-1~exp1~20241120202359.554)
-Target: x86_64-pc-linux-gnu
-Thread model: posix
-InstalledDir: /usr/lib/llvm-20/bin
-```
+Compiler is `Ubuntu clang version 20.0.0 (++20241029082144+7544d3af0e28-1~exp1~20241029082307.506)`.
 
 # Microbenchmark
 
@@ -19,45 +12,44 @@ InstalledDir: /usr/lib/llvm-20/bin
 
 |               ns/op |                op/s |    err% |          ins/op |          cyc/op |    IPC |         bra/op |   miss% |     total | benchmark
 |--------------------:|--------------------:|--------:|----------------:|----------------:|-------:|---------------:|--------:|----------:|:----------
-|              161.29 |        6,200,056.17 |    0.1% |        3,014.03 |          831.04 |  3.627 |         504.59 |    0.0% |      1.93 | `point reads`
-|              158.32 |        6,316,160.64 |    0.1% |        2,954.16 |          815.80 |  3.621 |         490.17 |    0.0% |      1.89 | `prefix reads`
-|              237.39 |        4,212,409.50 |    0.2% |        3,592.41 |        1,233.96 |  2.911 |         629.31 |    0.0% |      2.84 | `range reads`
-|              442.11 |        2,261,878.94 |    0.0% |        4,450.57 |        2,314.25 |  1.923 |         707.92 |    2.1% |      5.28 | `point writes`
-|              439.89 |        2,273,308.53 |    0.1% |        4,410.22 |        2,302.29 |  1.916 |         694.74 |    2.1% |      5.25 | `prefix writes`
-|              290.96 |        3,436,936.78 |    0.0% |        2,315.38 |        1,528.68 |  1.515 |         396.69 |    3.3% |      3.49 | `range writes`
-|              476.93 |        2,096,762.02 |    0.6% |        6,999.33 |        2,484.94 |  2.817 |       1,251.73 |    1.3% |      0.06 | `monotonic increasing point writes`
-|          131,736.57 |            7,590.91 |    1.1% |      807,444.50 |      704,941.71 |  1.145 |     144,584.60 |    0.9% |      0.01 | `worst case for radix tree`
-|               45.50 |       21,978,369.95 |    1.1% |          902.00 |          232.36 |  3.882 |         132.00 |    0.0% |      0.01 | `create and destroy`
+|              159.65 |        6,263,576.52 |    1.6% |        2,972.36 |          820.37 |  3.623 |         504.59 |    0.0% |      0.01 | `point reads`
+|              156.32 |        6,397,320.65 |    0.7% |        2,913.62 |          806.87 |  3.611 |         490.19 |    0.0% |      0.01 | `prefix reads`
+|              229.18 |        4,363,293.65 |    1.2% |        3,541.05 |        1,219.75 |  2.903 |         629.33 |    0.0% |      0.01 | `range reads`
+|              363.37 |        2,752,026.30 |    0.3% |        5,273.63 |        1,951.54 |  2.702 |         851.66 |    1.7% |      0.01 | `point writes`
+|              364.99 |        2,739,787.02 |    0.3% |        5,250.92 |        1,958.54 |  2.681 |         839.24 |    1.7% |      0.01 | `prefix writes`
+|              242.26 |        4,127,796.58 |    2.9% |        3,117.33 |        1,304.41 |  2.390 |         541.07 |    2.8% |      0.02 | `range writes`
+|              562.48 |        1,777,855.27 |    0.8% |        7,305.21 |        3,034.34 |  2.408 |       1,329.30 |    1.3% |      0.01 | `monotonic increasing point writes`
+|          122,688.57 |            8,150.72 |    0.7% |      798,766.00 |      666,842.00 |  1.198 |     144,584.50 |    0.1% |      0.01 | `worst case for radix tree`
+|               41.71 |       23,976,459.34 |    1.7% |          885.00 |          219.17 |  4.038 |         132.00 |    0.0% |      0.01 | `create and destroy`
 
 ## Radix tree (this implementation)
 
-
 |               ns/op |                op/s |    err% |          ins/op |          cyc/op |    IPC |         bra/op |   miss% |     total | benchmark
 |--------------------:|--------------------:|--------:|----------------:|----------------:|-------:|---------------:|--------:|----------:|:----------
-|               12.36 |       80,885,626.43 |    0.2% |          243.56 |           63.62 |  3.828 |          31.07 |    0.6% |      0.15 | `point reads`
-|               14.18 |       70,502,196.81 |    0.1% |          297.72 |           73.13 |  4.071 |          40.31 |    0.5% |      0.17 | `prefix reads`
-|               33.44 |       29,901,623.04 |    0.1% |          767.90 |          172.42 |  4.454 |         101.32 |    0.2% |      0.40 | `range reads`
-|               19.48 |       51,342,564.70 |    0.3% |          374.45 |          100.43 |  3.728 |          48.92 |    0.5% |      0.23 | `point writes`
-|               37.46 |       26,694,471.44 |    0.1% |          672.00 |          193.14 |  3.479 |         101.28 |    0.3% |      0.45 | `prefix writes`
-|               38.78 |       25,784,784.34 |    0.0% |          738.26 |          199.93 |  3.693 |         111.59 |    0.1% |      0.47 | `range writes`
-|               76.05 |       13,148,995.74 |    0.7% |        1,450.77 |          397.16 |  3.653 |         275.72 |    0.0% |      0.01 | `monotonic increasing point writes`
-|          286,920.33 |            3,485.29 |    0.4% |    4,117,948.00 |    1,521,352.00 |  2.707 |     714,833.00 |    0.1% |      0.01 | `worst case for radix tree`
-|               95.66 |       10,453,798.72 |    0.5% |        1,986.00 |          495.04 |  4.012 |         315.00 |    0.0% |      0.01 | `create and destroy`
+|               12.63 |       79,186,868.18 |    1.4% |          241.61 |           64.76 |  3.731 |          31.64 |    0.8% |      0.01 | `point reads`
+|               14.48 |       69,078,073.40 |    0.3% |          292.42 |           74.69 |  3.915 |          41.49 |    0.5% |      0.01 | `prefix reads`
+|               34.37 |       29,094,694.11 |    0.2% |          759.53 |          179.77 |  4.225 |         100.38 |    0.2% |      0.01 | `range reads`
+|               19.34 |       51,713,896.36 |    0.7% |          369.70 |          101.81 |  3.631 |          47.88 |    0.6% |      0.01 | `point writes`
+|               39.16 |       25,538,968.61 |    0.2% |          653.16 |          206.77 |  3.159 |          89.62 |    0.8% |      0.01 | `prefix writes`
+|               40.58 |       24,642,681.12 |    4.7% |          718.44 |          216.44 |  3.319 |          99.28 |    0.6% |      0.01 | `range writes`
+|               78.77 |       12,694,520.69 |    3.8% |        1,395.55 |          421.73 |  3.309 |         249.81 |    0.1% |      0.01 | `monotonic increasing point writes`
+|          287,760.50 |            3,475.11 |    0.5% |    3,929,266.50 |    1,550,225.50 |  2.535 |     639,064.00 |    0.0% |      0.01 | `worst case for radix tree`
+|              104.76 |        9,545,250.65 |    3.1% |        2,000.00 |          552.82 |  3.618 |         342.00 |    0.0% |      0.01 | `create and destroy`
 
 # "Real data" test
 
-Point queries only. Gc ratio is the ratio of time spent doing garbage collection to time spent adding writes or doing garbage collection. Lower is better.
+Point queries only, best of three runs. Gc ratio is the ratio of time spent doing garbage collection to time spent adding writes or doing garbage collection. Lower is better.
 
 ## skip list
 
 ```
-Check: 4.53508 seconds, 371.81 MB/s, Add: 3.81222 seconds, 150.919 MB/s, Gc ratio: 33.66%, Peak idle memory: 5.61007e+06
+Check: 4.39702 seconds, 370.83 MB/s, Add: 4.50025 seconds, 124.583 MB/s, Gc ratio: 29.1333%, Peak idle memory: 5.51852e+06
 ```
 
 ## radix tree
 
 ```
-Check: 0.957735 seconds, 1760.6 MB/s, Add: 1.19942 seconds, 479.678 MB/s, Gc ratio: 38.6069%, Peak idle memory: 2.05667e+06
+Check: 0.987757 seconds, 1650.76 MB/s, Add: 1.24815 seconds, 449.186 MB/s, Gc ratio: 41.4675%, Peak idle memory: 2.02872e+06
 ```
 
 ## hash table
@@ -65,6 +57,5 @@ Check: 0.957735 seconds, 1760.6 MB/s, Add: 1.19942 seconds, 479.678 MB/s, Gc rat
 (The hash table implementation doesn't work on range queries, and its purpose is to provide an idea of how fast point queries can be)
 
 ```
-Check: 0.804598 seconds, 2095.69 MB/s, Add: 0.671221 seconds, 857.147 MB/s, Gc ratio: 35.0034%, Peak idle memory: 0
+Check: 0.84256 seconds, 1935.23 MB/s, Add: 0.697204 seconds, 804.146 MB/s, Gc ratio: 35.4091%
 ```
-

RealDataBench.cpp

@@ -5,7 +5,7 @@
 #include <cstdio>
 #include <cstring>
 #include <fcntl.h>
-#include <span>
+#include <string_view>
 #include <sys/mman.h>
 #include <sys/stat.h>
 #include <unistd.h>
@@ -64,7 +64,7 @@ int main(int argc, const char **argv) {
     auto *const mapOriginal = begin;
     const auto sizeOriginal = size;
 
-    using StringView = std::span<const uint8_t>;
+    using StringView = std::basic_string_view<uint8_t>;
 
     StringView write;
     std::vector<StringView> reads;
@@ -78,9 +78,9 @@ int main(int argc, const char **argv) {
       end = (uint8_t *)memchr(begin, '\n', size);
 
       if (line.size() > 0 && line[0] == 'P') {
-        write = line.subspan(2, line.size());
+        write = line.substr(2, line.size());
       } else if (line.size() > 0 && line[0] == 'L') {
-        reads.push_back(line.subspan(2, line.size()));
+        reads.push_back(line.substr(2, line.size()));
       } else if (line.empty()) {
         {
           readRanges.resize(reads.size());
@@ -133,10 +133,10 @@ int main(int argc, const char **argv) {
   int metricsCount;
   cs.getMetricsV1(&metrics, &metricsCount);
   for (int i = 0; i < metricsCount; ++i) {
-    fprintf(stderr, "# HELP %s %s\n", metrics[i].name, metrics[i].help);
-    fprintf(stderr, "# TYPE %s %s\n", metrics[i].name,
+    printf("# HELP %s %s\n", metrics[i].name, metrics[i].help);
+    printf("# TYPE %s %s\n", metrics[i].name,
            metrics[i].type == metrics[i].Counter ? "counter" : "gauge");
-    fprintf(stderr, "%s %g\n", metrics[i].name, metrics[i].getValue());
+    printf("%s %g\n", metrics[i].name, metrics[i].getValue());
   }
 
   printf("Check: %g seconds, %g MB/s, Add: %g seconds, %g MB/s, Gc ratio: "

ServerBench.cpp

@@ -23,97 +23,6 @@
 #include "Internal.h"
 #include "third_party/nadeau.h"
 
-constexpr int kCacheLine = 64; // TODO mac m1 is 128
-
-template <class T> struct TxQueue {
-
-  explicit TxQueue(int lgSlotCount)
-      : slotCount(1 << lgSlotCount), slotCountMask(slotCount - 1),
-        slots(new T[slotCount]) {
-    // Otherwise we can't tell the difference between full and empty.
-    assert(!(slotCountMask & 0x80000000));
-  }
-
-  /// Call from producer thread, after ensuring consumer is no longer accessing
-  /// it somehow
-  ~TxQueue() { delete[] slots; }
-
-  /// Must be called from the producer thread
-  void push(T t) {
-    if (wouldBlock()) {
-      // Wait for pops to change and try again
-      consumer.pops.wait(producer.lastPopRead, std::memory_order_relaxed);
-      producer.lastPopRead = consumer.pops.load(std::memory_order_acquire);
-    }
-    slots[producer.pushesNonAtomic++ & slotCountMask] = std::move(t);
-    // seq_cst so that the notify can't be ordered before the store
-    producer.pushes.store(producer.pushesNonAtomic, std::memory_order_seq_cst);
-    // We have to notify every time, since we don't know if this is the last
-    // push ever
-    producer.pushes.notify_one();
-  }
-
-  /// Must be called from the producer thread
-  uint32_t outstanding() {
-    return producer.pushesNonAtomic -
-           consumer.pops.load(std::memory_order_relaxed);
-  }
-
-  /// Returns true if a call to push might block. Must be called from the
-  /// producer thread.
-  bool wouldBlock() {
-    // See if we can determine that overflow won't happen entirely from state
-    // local to the producer
-    if (producer.pushesNonAtomic - producer.lastPopRead == slotCount - 1) {
-      // Re-read pops with memory order
-      producer.lastPopRead = consumer.pops.load(std::memory_order_acquire);
-      return producer.pushesNonAtomic - producer.lastPopRead == slotCount - 1;
-    }
-    return false;
-  }
-
-  /// Valid until the next pop, or until this queue is destroyed.
-  T *pop() {
-    // See if we can determine that there's an entry we can pop entirely from
-    // state local to the consumer
-    if (consumer.lastPushRead - consumer.popsNonAtomic == 0) {
-      // Re-read pushes with memory order and try again
-      consumer.lastPushRead = producer.pushes.load(std::memory_order_acquire);
-      if (consumer.lastPushRead - consumer.popsNonAtomic == 0) {
-        // Wait for pushes to change and try again
-        producer.pushes.wait(consumer.lastPushRead, std::memory_order_relaxed);
-        consumer.lastPushRead = producer.pushes.load(std::memory_order_acquire);
-      }
-    }
-    auto result = &slots[consumer.popsNonAtomic++ & slotCountMask];
-    // We only have to write pops with memory order if we've run out of items.
-    // We know that we'll eventually run out.
-    if (consumer.lastPushRead - consumer.popsNonAtomic == 0) {
-      // seq_cst so that the notify can't be ordered before the store
-      consumer.pops.store(consumer.popsNonAtomic, std::memory_order_seq_cst);
-      consumer.pops.notify_one();
-    }
-    return result;
-  }
-
-private:
-  const uint32_t slotCount;
-  const uint32_t slotCountMask;
-  T *slots;
-  struct alignas(kCacheLine) ProducerState {
-    std::atomic<uint32_t> pushes{0};
-    uint32_t pushesNonAtomic{0};
-    uint32_t lastPopRead{0};
-  };
-  struct alignas(kCacheLine) ConsumerState {
-    std::atomic<uint32_t> pops{0};
-    uint32_t popsNonAtomic{0};
-    uint32_t lastPushRead{0};
-  };
-  ProducerState producer;
-  ConsumerState consumer;
-};
-
 std::atomic<int64_t> transactions;
 
 int64_t safeUnaryMinus(int64_t x) {
@@ -138,7 +47,6 @@ void tupleAppend(std::string &output, int64_t value) {
 
 void tupleAppend(std::string &output, std::string_view value) {
   output.push_back('\x02');
-  if (memchr(value.data(), '\x00', value.size()) != nullptr) {
   for (auto c : value) {
     if (c == '\x00') {
       output.push_back('\x00');
@@ -147,9 +55,6 @@ void tupleAppend(std::string &output, std::string_view value) {
       output.push_back(c);
     }
   }
-  } else {
-    output.insert(output.end(), value.begin(), value.end());
-  }
   output.push_back('\x00');
 }
 
@@ -159,71 +64,35 @@ template <class... Ts> std::string tupleKey(const Ts &...ts) {
   return result;
 }
 
-constexpr int kTotalKeyRange = 1'000'000'000;
-constexpr int kWindowSize = 1'000'000;
-constexpr int kNumReadKeysPerTx = 5;
-constexpr int kNumWriteKeysPerTx = 10;
+constexpr int kWindowSize = 300000;
 
-struct Transaction {
-  std::vector<std::string> keys;
-  std::vector<weaselab::ConflictSet::ReadRange> reads;
-  std::vector<weaselab::ConflictSet::WriteRange> writes;
-  int64_t version;
-  int64_t oldestVersion;
-  Transaction() = default;
-  explicit Transaction(int64_t version)
-      : version(version), oldestVersion(version - kWindowSize) {
+void workload(weaselab::ConflictSet *cs) {
+  int64_t version = kWindowSize;
+  constexpr int kNumWrites = 16;
+  for (;; transactions.fetch_add(1, std::memory_order_relaxed)) {
     std::vector<int64_t> keyIndices;
-    for (int i = 0; i < std::max(kNumReadKeysPerTx, kNumWriteKeysPerTx); ++i) {
-      keyIndices.push_back(rand() % kTotalKeyRange);
+    for (int i = 0; i < kNumWrites; ++i) {
+      keyIndices.push_back(rand() % 100'000'000);
     }
     std::sort(keyIndices.begin(), keyIndices.end());
-    constexpr std::string_view fullString =
-        "this is a string, where a prefix of it is used as an element of the "
-        "tuple forming the key";
-    for (int i = 0; i < int(keyIndices.size()); ++i) {
-      keys.push_back(
-          tupleKey(0x100, keyIndices[i] / fullString.size(),
-                   fullString.substr(0, keyIndices[i] % fullString.size())));
+    std::vector<std::string> keys;
+    std::vector<weaselab::ConflictSet::WriteRange> writes;
+    constexpr std::string_view suffix = "this is a suffix";
+    for (int i = 0; i < kNumWrites; ++i) {
+      keys.push_back(tupleKey(0x100, i, keyIndices[i],
+                              suffix.substr(0, rand() % suffix.size()),
+                              rand()));
       // printf("%s\n", printable(keys.back()).c_str());
     }
-    for (int i = 0; i < kNumWriteKeysPerTx; ++i) {
+    for (int i = 0; i < kNumWrites; ++i) {
       writes.push_back({{(const uint8_t *)keys[i].data(), int(keys[i].size())},
                         {nullptr, 0}});
     }
-    reads.push_back({{(const uint8_t *)keys[0].data(), int(keys[0].size())},
-                     {(const uint8_t *)keys[1].data(), int(keys[1].size())},
-                     version - std::min(10, kWindowSize)});
-    static_assert(kNumReadKeysPerTx >= 3);
-    for (int i = 2; i < kNumReadKeysPerTx; ++i) {
-      reads.push_back({{(const uint8_t *)keys[i].data(), int(keys[i].size())},
-                       {nullptr, 0},
-                       version - kWindowSize});
+    cs->addWrites(writes.data(), writes.size(), version);
+    cs->setOldestVersion(version - kWindowSize);
+    ++version;
   }
-  }
-
-  Transaction(Transaction &&) = default;
-  Transaction &operator=(Transaction &&) = default;
-  Transaction(Transaction const &) = delete;
-  Transaction const &operator=(Transaction const &) = delete;
-};
-
-struct Resolver {
-
-  void resolve(const weaselab::ConflictSet::ReadRange *reads, int readCount,
-               const weaselab::ConflictSet::WriteRange *writes, int writeCount,
-               int64_t newVersion, int64_t newOldestVersion) {
-    results.resize(readCount);
-    cs.check(reads, results.data(), readCount);
-    cs.addWrites(writes, writeCount, newVersion);
-    cs.setOldestVersion(newOldestVersion);
-  }
-
-  ConflictSet cs{0};
-
-private:
-  std::vector<weaselab::ConflictSet::Result> results;
-};
+}
 
 // Adapted from getaddrinfo man page
 int getListenFd(const char *node, const char *service) {
@@ -366,8 +235,7 @@ int main(int argc, char **argv) {
   {
     int listenFd = getListenFd(argv[1], argv[2]);
 
-    Resolver resolver;
-    auto &cs = resolver.cs;
+    weaselab::ConflictSet cs{0};
     weaselab::ConflictSet::MetricsV1 *metrics;
     int metricsCount;
     cs.getMetricsV1(&metrics, &metricsCount);
@@ -416,22 +284,7 @@ int main(int argc, char **argv) {
     }
 #endif
 
-    TxQueue<Transaction> queue{10};
-
-    auto workloadThread = std::thread{[&]() {
-      for (int64_t version = kWindowSize;;
-           ++version, transactions.fetch_add(1, std::memory_order_relaxed)) {
-        queue.push(Transaction(version));
-      }
-    }};
-
-    auto resolverThread = std::thread{[&]() {
-      for (;;) {
-        auto tx = queue.pop();
-        resolver.resolve(tx->reads.data(), tx->reads.size(), tx->writes.data(),
-                         tx->writes.size(), tx->version, tx->oldestVersion);
-      }
-    }};
+    auto w = std::thread{workload, &cs};
 
     for (;;) {
       struct sockaddr_storage peer_addr = {};

SkipList.cpp

@@ -767,9 +767,7 @@ struct __attribute__((visibility("hidden"))) ConflictSet::Impl {
           false, true);
     }
 
-    if (!std::is_sorted(points.begin(), points.end())) {
     sortPoints(points);
-    }
 
     int activeWriteCount = 0;
     std::vector<std::pair<StringRef, StringRef>> combinedWriteConflictRanges;
@@ -796,6 +794,7 @@ struct __attribute__((visibility("hidden"))) ConflictSet::Impl {
     int temp[stripeSize];
     int stripes = (stringCount + stripeSize - 1) / stripeSize;
     StringRef values[stripeSize];
+    int64_t writeVersions[stripeSize / 2];
     int ss = stringCount - (stripes - 1) * stripeSize;
     int64_t entryDelta = 0;
     for (int s = stripes - 1; s >= 0; s--) {

apple-symbol-imports.txt

@@ -1,4 +1,3 @@
-___chkstk_darwin
 ___stack_chk_fail
 ___stack_chk_guard
 __tlv_bootstrap
@@ -6,7 +5,6 @@ _abort
 _bzero
 _free
 _malloc
-_memcmp
 _memcpy
 _memmove
 dyld_stub_binder