Revert "Fix Windows build issue in trie_index_factory.h"

This reverts commit c62e9a0d19.

.github/actions/install-jdk8-on-macos/action.yml

@@ -1,9 +0,0 @@
-name: install-jdk8-on-macos
-runs:
-  using: composite
-  steps:
-  - name: Install JDK 8 on macos
-    run: |-
-      HOMEBREW_NO_AUTO_UPDATE=1 brew tap bell-sw/liberica
-      HOMEBREW_NO_AUTO_UPDATE=1 brew install --cask liberica-jdk8
-    shell: bash

.github/actions/setup-jdk-on-macos/action.yml

@@ -0,0 +1,17 @@
+name: setup-jdk-on-macos
+runs:
+  using: composite
+  steps:
+  - name: Set up JDK on macos
+    run: |-
+      set -euo pipefail
+
+      JAVA_HOME="$(/usr/libexec/java_home -v 21)"
+
+      echo "JAVA_HOME=${JAVA_HOME}" >> "$GITHUB_ENV"
+      echo "JAVAC_ARGS=--release 8" >> "$GITHUB_ENV"
+      echo "${JAVA_HOME}/bin" >> "$GITHUB_PATH"
+      echo "Using JDK at ${JAVA_HOME}"
+      "${JAVA_HOME}/bin/java" -version
+      "${JAVA_HOME}/bin/javac" -version
+    shell: bash

.github/actions/windows-build-steps/action.yml

@@ -11,8 +11,49 @@ inputs:
 runs:
   using: composite
   steps:
+  - name: Detect Visual Studio generator
+    id: detect_vs
+    shell: pwsh
+    run: |
+      $ErrorActionPreference = "Stop"
+
+      $vswhere = Join-Path ([Environment]::GetFolderPath("ProgramFilesX86")) "Microsoft Visual Studio\Installer\vswhere.exe"
+      if (!(Test-Path $vswhere)) {
+        throw "vswhere.exe was not found at $vswhere"
+      }
+
+      $vsInstallationsJson = & $vswhere -latest -products * -requires Microsoft.VisualStudio.Component.VC.Tools.x86.x64 -format json -utf8
+      $vsInstallations = @($vsInstallationsJson | ConvertFrom-Json)
+      if ($vsInstallations.Count -eq 0) {
+        throw "No Visual Studio instance with C++ tools was found."
+      }
+
+      $vsInstallation = $vsInstallations[0]
+      $vsMajor = [int]$vsInstallation.catalog.productLineVersion
+      if ($vsMajor -eq 0) {
+        $vsMajor = [int]($vsInstallation.installationVersion.Split(".")[0])
+      }
+
+      $generatorLine = cmake --help | Where-Object {
+        $_ -match "^\*?\s*Visual Studio $vsMajor\s+\d{4}\s*="
+      } | Select-Object -First 1
+      if (!$generatorLine) {
+        throw "CMake does not support a Visual Studio generator for VS major version $vsMajor."
+      }
+
+      $cmakeGenerator = ($generatorLine -replace "^\*?\s*", "" -replace "\s*=.*$", "").Trim()
+      $vsVersionRange = "[{0}.0,{1}.0)" -f $vsMajor, ($vsMajor + 1)
+
+      echo "Detected Visual Studio at $($vsInstallation.installationPath)"
+      echo "Using CMake generator: $cmakeGenerator"
+      echo "Using setup-msbuild vs-version: $vsVersionRange"
+
+      "CMAKE_GENERATOR=$cmakeGenerator" >> $env:GITHUB_ENV
+      "vs_version=$vsVersionRange" >> $env:GITHUB_OUTPUT
   - name: Add msbuild to PATH
-    uses: microsoft/setup-msbuild@v1.3.1
+    uses: microsoft/setup-msbuild@v2
+    with:
+      vs-version: ${{ steps.detect_vs.outputs.vs_version }}
   - name: ccache
     uses: hendrikmuhs/ccache-action@v1.2
     with:
@@ -32,15 +73,23 @@ runs:
       CMAKE_HOME: C:/Program Files/CMake
       CMAKE_BIN: C:/Program Files/CMake/bin/cmake.exe
       CTEST_BIN: C:/Program Files/CMake/bin/ctest.exe
-      JAVA_HOME: C:/Program Files/BellSoft/LibericaJDK-8
       SNAPPY_HOME: ${{ github.workspace }}/thirdparty/snappy-1.2.2
       SNAPPY_INCLUDE: ${{ github.workspace }}/thirdparty/snappy-1.2.2;${{ github.workspace }}/thirdparty/snappy-1.2.2/build
       SNAPPY_LIB_DEBUG: ${{ github.workspace }}/thirdparty/snappy-1.2.2/build/Debug/snappy.lib
     run: |-
       # NOTE: if ... Exit $LASTEXITCODE lines needed to exit and report failure
       echo ===================== Install Dependencies =====================
-      choco install liberica8jdk -y
-      if(!$?) { Exit $LASTEXITCODE }
+      if (!$env:JAVA_HOME -or !(Test-Path (Join-Path $env:JAVA_HOME "bin\javac.exe"))) {
+        $javac = Get-Command javac -ErrorAction SilentlyContinue
+        if (!$javac) {
+          Write-Error "javac was not found on PATH and JAVA_HOME is not set."
+          Exit 1
+        }
+        $env:JAVA_HOME = Split-Path (Split-Path $javac.Source -Parent) -Parent
+      }
+      echo "JAVA_HOME=$env:JAVA_HOME"
+      & "$env:JAVA_HOME\bin\java.exe" -version
+      & "$env:JAVA_HOME\bin\javac.exe" -version
       mkdir $Env:THIRDPARTY_HOME
       cd $Env:THIRDPARTY_HOME
       echo "Building Snappy dependency..."
@@ -53,11 +102,11 @@ runs:
       cd build
       & cmake -G "$Env:CMAKE_GENERATOR" .. -DSNAPPY_BUILD_TESTS=OFF -DSNAPPY_BUILD_BENCHMARKS=OFF
       if(!$?) { Exit $LASTEXITCODE }
-      msbuild Snappy.sln -maxCpuCount -property:Configuration=Debug -property:Platform=x64
+      cmake --build . --config Debug --parallel 32 -- /p:Platform=x64
       if(!$?) { Exit $LASTEXITCODE }
       echo ======================== Build RocksDB =========================
       cd ${{ github.workspace }}
-      $env:Path = $env:JAVA_HOME + ";" + $env:Path
+      $env:Path = "$env:JAVA_HOME\bin;" + $env:Path
       mkdir build
       cd build
       & cmake -G "$Env:CMAKE_GENERATOR" -DCMAKE_BUILD_TYPE=Debug -DWIN_CI=1 -DPORTABLE="$Env:CMAKE_PORTABLE" -DSNAPPY=1 -DXPRESS=1 -DJNI=1 ..
@@ -65,7 +114,7 @@ runs:
       cd ..
       echo "Building with VS version: $Env:CMAKE_GENERATOR"
       # use more parallel processes than the number of processes available, as most of the compile command would be cache hit
-      msbuild build/rocksdb.sln /m:32 /p:LinkIncremental=false -property:Configuration=Debug -property:Platform=x64
+      cmake --build build --config Debug --parallel 32 -- /p:LinkIncremental=false /p:Platform=x64
       if(!$?) { Exit $LASTEXITCODE }
       echo ========================= Test RocksDB =========================
       $suiteRun = "${{ inputs.suite-run }}"

.github/workflows/ai-review-analysis.yml

@@ -714,8 +714,6 @@ jobs:
           CODEX_HOME: /tmp/codex-home
           CODEX_MODEL: ${{ inputs.default_model }}
           CODEX_REASONING_EFFORT: ${{ steps.budget_codex.outputs.effort }}
-          BASE_SHA: ${{ steps.pr_info.outputs.base_sha }}
-          PR_TITLE: ${{ steps.pr_info.outputs.pr_title }}
         run: |
           if [ -z "${OPENAI_API_KEY}" ]; then
             echo "OPENAI_API_KEY not configured, skipping Codex review"
@@ -723,9 +721,7 @@ jobs:
           fi
           mkdir -p "${CODEX_HOME}"
           set +e
-          codex exec review \
-            --base "${BASE_SHA}" \
-            --title "${PR_TITLE}" \
+          codex exec \
             -m "${CODEX_MODEL}" \
             -c model_reasoning_effort="${CODEX_REASONING_EFFORT}" \
             --dangerously-bypass-approvals-and-sandbox \
@@ -1340,8 +1336,6 @@ jobs:
           CODEX_HOME: /tmp/codex-home
           CODEX_MODEL: ${{ inputs.selected_model != '' && inputs.selected_model || inputs.default_model }}
           CODEX_REASONING_EFFORT: ${{ steps.manual_budget_codex.outputs.effort }}
-          BASE_SHA: ${{ steps.pr_info.outputs.base_sha }}
-          PR_TITLE: ${{ steps.pr_info.outputs.pr_title }}
           COMMAND_TYPE: ${{ steps.request.outputs.type }}
         run: |
           mkdir -p "${CODEX_HOME}"
@@ -1354,9 +1348,7 @@ jobs:
               --output-last-message codex-review-output.md \
               - < /tmp/prompt.txt > codex-review-execution.log 2>&1
           else
-            codex exec review \
-              --base "${BASE_SHA}" \
-              --title "${PR_TITLE}" \
+            codex exec \
               -m "${CODEX_MODEL}" \
               -c model_reasoning_effort="${CODEX_REASONING_EFFORT}" \
               --dangerously-bypass-approvals-and-sandbox \

.github/workflows/ai-review-comment.yml

@@ -40,6 +40,7 @@ jobs:
           name: ${{ inputs.result_artifact_name }}
           run-id: ${{ github.event.workflow_run.id }}
           github-token: ${{ secrets.GITHUB_TOKEN }}
+          path: _artifact
         continue-on-error: true
 
       - name: Post or update PR comment
@@ -52,8 +53,18 @@ jobs:
         with:
           script: |
             const fs = require('fs');
-            if (!fs.existsSync(process.env.COMMENT_FILE) ||
-                !fs.existsSync('pr_number.txt')) {
+            const path = require('path');
+            const artifactDir = '_artifact';
+            const commentFile = path.join(
+              artifactDir,
+              process.env.COMMENT_FILE
+            );
+            const prNumberFile = path.join(artifactDir, 'pr_number.txt');
+            const triggerTypeFile = path.join(artifactDir, 'trigger_type.txt');
+            const headShaFile = path.join(artifactDir, 'head_sha.txt');
+
+            if (!fs.existsSync(commentFile) ||
+                !fs.existsSync(prNumberFile)) {
               core.info('No review results found; skipping.');
               return;
             }
@@ -61,16 +72,16 @@ jobs:
             const post = require('./.github/scripts/post-pr-comment.js');
             const provider = process.env.PROVIDER;
             const providerTitle = provider === 'codex' ? 'Codex' : 'Claude';
-            const body = fs.readFileSync(process.env.COMMENT_FILE, 'utf8');
+            const body = fs.readFileSync(commentFile, 'utf8');
             const prNumber = parseInt(
-              fs.readFileSync('pr_number.txt', 'utf8').trim(),
+              fs.readFileSync(prNumberFile, 'utf8').trim(),
               10
             );
-            const trigger = fs.existsSync('trigger_type.txt') ?
-              fs.readFileSync('trigger_type.txt', 'utf8').trim() :
+            const trigger = fs.existsSync(triggerTypeFile) ?
+              fs.readFileSync(triggerTypeFile, 'utf8').trim() :
               'auto';
-            const headSha = fs.existsSync('head_sha.txt') ?
-              fs.readFileSync('head_sha.txt', 'utf8').trim() :
+            const headSha = fs.existsSync(headShaFile) ?
+              fs.readFileSync(headShaFile, 'utf8').trim() :
               '';
             const shortSha = headSha ? headSha.substring(0, 7) : '';
             const runId = process.env.RUN_ID;
@@ -119,12 +130,16 @@ jobs:
         with:
           script: |
             const fs = require('fs');
-            if (!fs.existsSync('trigger_type.txt')) return;
-            const trigger = fs.readFileSync('trigger_type.txt', 'utf8').trim();
+            const path = require('path');
+            const artifactDir = '_artifact';
+            const triggerTypeFile = path.join(artifactDir, 'trigger_type.txt');
+            const commentIdFile = path.join(artifactDir, 'comment_id.txt');
+            if (!fs.existsSync(triggerTypeFile)) return;
+            const trigger = fs.readFileSync(triggerTypeFile, 'utf8').trim();
             if (trigger !== 'manual') return;
-            if (!fs.existsSync('comment_id.txt')) return;
+            if (!fs.existsSync(commentIdFile)) return;
             const commentId = parseInt(
-              fs.readFileSync('comment_id.txt', 'utf8').trim(),
+              fs.readFileSync(commentIdFile, 'utf8').trim(),
               10
             );
             if (!commentId) return;
@@ -147,6 +162,7 @@ jobs:
           name: ${{ inputs.result_artifact_name }}
           run-id: ${{ github.event.workflow_run.id }}
           github-token: ${{ secrets.GITHUB_TOKEN }}
+          path: _artifact
         continue-on-error: true
 
       - name: React with failure
@@ -155,9 +171,11 @@ jobs:
         with:
           script: |
             const fs = require('fs');
-            if (!fs.existsSync('comment_id.txt')) return;
+            const path = require('path');
+            const commentIdFile = path.join('_artifact', 'comment_id.txt');
+            if (!fs.existsSync(commentIdFile)) return;
             const commentId = parseInt(
-              fs.readFileSync('comment_id.txt', 'utf8').trim(),
+              fs.readFileSync(commentIdFile, 'utf8').trim(),
               10
             );
             if (!commentId) return;

.github/workflows/clang-tidy-comment.yml

@@ -25,6 +25,7 @@ jobs:
         name: clang-tidy-result
         run-id: ${{ github.event.workflow_run.id }}
         github-token: ${{ secrets.GITHUB_TOKEN }}
+        path: _artifact
       continue-on-error: true
 
     - name: Post or update PR comment
@@ -33,12 +34,19 @@ jobs:
       with:
         script: |
           const fs = require('fs');
-          if (!fs.existsSync('clang-tidy-comment.md') || !fs.existsSync('pr_number.txt')) {
+          if (!fs.existsSync('_artifact/clang-tidy-comment.md') ||
+              !fs.existsSync('_artifact/pr_number.txt')) {
             core.info('No clang-tidy results found; skipping.');
             return;
           }
-          const body = fs.readFileSync('clang-tidy-comment.md', 'utf8');
-          const prNumber = parseInt(fs.readFileSync('pr_number.txt', 'utf8').trim());
+          const body = fs.readFileSync(
+            '_artifact/clang-tidy-comment.md',
+            'utf8'
+          );
+          const prNumber = parseInt(
+            fs.readFileSync('_artifact/pr_number.txt', 'utf8').trim(),
+            10
+          );
           const post = require('./.github/scripts/post-pr-comment.js');
 
           await post({

.github/workflows/claude-review-comment.yml

@@ -22,4 +22,3 @@ jobs:
       provider: claude
       result_artifact_name: claude-review-result
       comment_file: claude-review-comment.md
-    secrets: inherit

.github/workflows/codex-review-comment.yml

@@ -22,4 +22,3 @@ jobs:
       provider: codex
       result_artifact_name: codex-review-result
       comment_file: codex-review-comment.md
-    secrets: inherit

.github/workflows/codex-review.yml

@@ -7,14 +7,9 @@
 name: Codex Code Review
 
 on:
-  workflow_run:
-    workflows: ["facebook/rocksdb/pr-jobs"]
-    types: [completed]
-
-  # The early pull_request_target path is limited to same-repo PRs by the job
-  # condition below. Fork PRs skip this path and rely on workflow_run instead.
-  pull_request_target:
-    types: [opened, reopened, synchronize]
+  # Automatic PR reviews are temporarily disabled while Codex authentication is
+  # repaired. Manual review requests remain available through issue comments
+  # and workflow_dispatch.
 
   issue_comment:
     types: [created]

.github/workflows/nightly.yml

@@ -113,11 +113,10 @@ jobs:
     - run: "USE_FOLLY=1 LIB_MODE=static DEBUG_LEVEL=0 V=1 make -j20 release"
     - run: "(mkdir build && cd build && cmake -DUSE_FOLLY=1 -DWITH_GFLAGS=1 -DROCKSDB_BUILD_SHARED=0 -DCMAKE_BUILD_TYPE=Release .. && make VERBOSE=1 -j20 && ctest -j20)"
     - uses: "./.github/actions/post-steps"
-  build-windows-vs2022-avx2:
+  build-windows-msvc-avx2:
     if: ${{ github.repository_owner == 'facebook' }}
-    runs-on: windows-2022
+    runs-on: windows-8-core
     env:
-      CMAKE_GENERATOR: Visual Studio 17 2022
       CMAKE_PORTABLE: AVX2
     steps:
     - uses: actions/checkout@v4.1.0

.github/workflows/pr-jobs.yml

@@ -314,6 +314,23 @@ jobs:
     - run: CC=clang-21 CXX=clang++-21 USE_CLANG=1 make -j32 all microbench
     - run: make clean
     - run: CC=clang-21 CXX=clang++-21 USE_CLANG=1 DEBUG_LEVEL=0 make -j32 release
+    - name: Install clang-format-21 for C API gen checks
+      run: apt-get update && apt-get install -y clang-format-21
+    - name: Check C API codegen (link-complete, backward-compatible, up to date)
+      # Run the make target in STRICT mode so this core CI job HARD-FAILS rather
+      # than silently skipping if a prerequisite (clang++, clang-format, or the
+      # compat baseline ref) is ever missing. See `make check-c-api-gen`.
+      run: |
+        # The container checkout is owned by a different user, so git refuses to
+        # operate on it ("dubious ownership") -- mark it safe for the `git fetch`
+        # and the `git show` the compat checker runs.
+        git config --global --add safe.directory "$GITHUB_WORKSPACE"
+        git config --global --add safe.directory '*'
+        git fetch --no-tags --depth=1 origin main
+        CXX=clang++-21 make check-c-api-gen \
+          CHECK_C_API_GEN_STRICT=1 \
+          API_COMPAT_REF=FETCH_HEAD \
+          CLANG_FORMAT_BINARY=clang-format-21
     - uses: "./.github/actions/teardown-ccache"
       if: always()
     - uses: "./.github/actions/post-steps"
@@ -525,8 +542,8 @@ jobs:
     - uses: "./.github/actions/post-steps"
   # ======================== Windows with Tests ======================= #
   # NOTE: some windows jobs are in "nightly" to save resources
-  build-windows-vs2022:
-    if: needs.config.outputs.only_job == 'build-windows-vs2022' || (needs.config.outputs.only_job == '' && github.repository_owner == 'facebook')
+  build-windows-msvc:
+    if: needs.config.outputs.only_job == 'build-windows-msvc' || (needs.config.outputs.only_job == '' && github.repository_owner == 'facebook')
     needs: config
     runs-on: windows-8-core
     strategy:
@@ -543,7 +560,6 @@ jobs:
           suite_run: ""
           run_java: "true"
     env:
-      CMAKE_GENERATOR: Visual Studio 17 2022
       CMAKE_PORTABLE: 1
     steps:
     - uses: actions/checkout@v4.1.0
@@ -603,7 +619,6 @@ jobs:
     needs: config
     runs-on: macos-15-xlarge
     env:
-      JAVA_HOME: "/Library/Java/JavaVirtualMachines/liberica-jdk-8.jdk/Contents/Home"
       ROCKSDB_DISABLE_JEMALLOC: 1
     steps:
     - uses: actions/checkout@v4.1.0
@@ -615,7 +630,7 @@ jobs:
         cache-key-prefix: macos-java
     - uses: "./.github/actions/increase-max-open-files-on-macos"
     - uses: "./.github/actions/install-gflags-on-macos"
-    - uses: "./.github/actions/install-jdk8-on-macos"
+    - uses: "./.github/actions/setup-jdk-on-macos"
     - uses: "./.github/actions/pre-steps-macos"
     - name: Set Java Environment
       run: |-
@@ -631,8 +646,6 @@ jobs:
     if: needs.config.outputs.only_job == 'build-macos-java-static' || (needs.config.outputs.only_job == '' && github.repository_owner == 'facebook')
     needs: config
     runs-on: macos-15-xlarge
-    env:
-      JAVA_HOME: "/Library/Java/JavaVirtualMachines/liberica-jdk-8.jdk/Contents/Home"
     steps:
     - uses: actions/checkout@v4.1.0
     - uses: maxim-lobanov/setup-xcode@v1.6.0
@@ -643,7 +656,7 @@ jobs:
         cache-key-prefix: macos-java-static
     - uses: "./.github/actions/increase-max-open-files-on-macos"
     - uses: "./.github/actions/install-gflags-on-macos"
-    - uses: "./.github/actions/install-jdk8-on-macos"
+    - uses: "./.github/actions/setup-jdk-on-macos"
     - uses: "./.github/actions/pre-steps-macos"
     - name: Set Java Environment
       run: |-
@@ -659,8 +672,6 @@ jobs:
     if: needs.config.outputs.only_job == 'build-macos-java-static-universal' || (needs.config.outputs.only_job == '' && github.repository_owner == 'facebook')
     needs: config
     runs-on: macos-15-xlarge
-    env:
-      JAVA_HOME: "/Library/Java/JavaVirtualMachines/liberica-jdk-8.jdk/Contents/Home"
     steps:
     - uses: actions/checkout@v4.1.0
     - uses: maxim-lobanov/setup-xcode@v1.6.0
@@ -671,7 +682,7 @@ jobs:
         cache-key-prefix: macos-java-static-universal
     - uses: "./.github/actions/increase-max-open-files-on-macos"
     - uses: "./.github/actions/install-gflags-on-macos"
-    - uses: "./.github/actions/install-jdk8-on-macos"
+    - uses: "./.github/actions/setup-jdk-on-macos"
     - uses: "./.github/actions/pre-steps-macos"
     - name: Set Java Environment
       run: |-

.gitignore

@@ -1,5 +1,7 @@
 make_config.mk
 rocksdb.pc
+.build_signature
+build_tools/__pycache__/
 
 *.a
 *.arc
@@ -88,6 +90,8 @@ fbcode/
 fbcode
 buckifier/*.pyc
 buckifier/__pycache__
+tools/__pycache__/
+tools/c_api_gen/__pycache__/
 .arcconfig
 
 compile_commands.json

BUCK

@@ -35,6 +35,7 @@ cpp_library_wrapper(name="rocksdb_lib", srcs=[
         "db/blob/blob_file_partition_manager.cc",
         "db/blob/blob_file_reader.cc",
         "db/blob/blob_garbage_meter.cc",
+        "db/blob/blob_gen2_format.cc",
         "db/blob/blob_log_format.cc",
         "db/blob/blob_log_sequential_reader.cc",
         "db/blob/blob_log_writer.cc",
@@ -267,6 +268,7 @@ cpp_library_wrapper(name="rocksdb_lib", srcs=[
         "util/compression.cc",
         "util/compression_context_cache.cc",
         "util/concurrent_task_limiter_impl.cc",
+        "util/coro_stats_util.cc",
         "util/crc32c.cc",
         "util/crc32c_arm64.cc",
         "util/data_structure.cc",
@@ -305,6 +307,7 @@ cpp_library_wrapper(name="rocksdb_lib", srcs=[
         "utilities/compaction_filters.cc",
         "utilities/compaction_filters/remove_emptyvalue_compactionfilter.cc",
         "utilities/convenience/info_log_finder.cc",
+        "utilities/copy_engine/copy_engine.cc",
         "utilities/counted_fs.cc",
         "utilities/debug.cc",
         "utilities/env_mirror.cc",
@@ -378,11 +381,21 @@ cpp_library_wrapper(name="rocksdb_lib", srcs=[
         "utilities/write_batch_with_index/write_batch_with_index_internal.cc",
     ], deps=[
         "//folly/container:f14_hash",
+        "//folly/coro:baton",
         "//folly/coro:blocking_wait",
         "//folly/coro:collect",
         "//folly/coro:coroutine",
+        "//folly/coro:nothrow",
         "//folly/coro:task",
+        "//folly/executors/thread_factory:named_thread_factory",
+        "//folly/executors:io_executor",
+        "//folly/executors:io_thread_pool_executor",
+        "//folly/io/async:async_base",
+        "//folly/io/async:event_base_manager",
+        "//folly/io/async:io_uring_backend",
+        "//folly/io/async:request_context",
         "//folly/synchronization:distributed_mutex",
+        "//folly:executor",
     ], headers=glob(["**/*.h"]), link_whole=False, extra_test_libs=False)
 
 cpp_library_wrapper(name="rocksdb_whole_archive_lib", srcs=[], deps=[":rocksdb_lib"], headers=[], link_whole=True, extra_test_libs=False)
@@ -4742,6 +4755,12 @@ cpp_unittest_wrapper(name="configurable_test",
             extra_compiler_flags=[])
 
 
+cpp_unittest_wrapper(name="copy_engine_test",
+            srcs=["utilities/copy_engine/copy_engine_test.cc"],
+            deps=[":rocksdb_test_lib"],
+            extra_compiler_flags=[])
+
+
 cpp_unittest_wrapper(name="corruption_test",
             srcs=["db/corruption_test.cc"],
             deps=[":rocksdb_test_lib"],
@@ -5795,3 +5814,5 @@ cpp_unittest_wrapper(name="write_unprepared_transaction_test",
 
 
 export_file(name = "tools/db_crashtest.py")
+
+export_file(name = "tools/fault_injection_log_parser.py")

CLAUDE.md

@@ -8,7 +8,7 @@ This document provides guidance for generating and reviewing code in the RocksDB
 
 ### Code Quality and Maintainability
 
-**Clarity and Readability:** Write clear, self-documenting code. Use meaningful variable names, add comments for complex logic, and structure code to minimize cognitive load. Avoid clever tricks that sacrifice readability for marginal performance gains unless absolutely necessary.
+**Clarity and Readability:** Write clear, self-documenting code. Use meaningful variable names, add comments for complex logic, and structure code to minimize cognitive load. Avoid clever tricks that sacrifice readability for marginal performance gains unless absolutely necessary. Avoid static_cast, reinterpret_cast, and C-style casts; static_cast_with_check, up_cast, and lossless_cast from cast_util.h are preferred.
 
 **Consistent Style:** Follow existing code style conventions. RocksDB uses `.clang-format` for formatting, specific naming conventions, and structural patterns. Deviations from these patterns are frequently flagged in reviews.
 
@@ -38,7 +38,7 @@ This document provides guidance for generating and reviewing code in the RocksDB
 
 **SIMD and Vectorization:** Leverage SIMD instructions (SSE, AVX) for data-parallel operations when appropriate. Structure data to enable auto-vectorization by the compiler. Consider explicit SIMD intrinsics for critical hot paths like checksum computation, encoding/decoding, and bulk data processing.
 
-**Branch Prediction:** Minimize unpredictable branches in hot paths. Use `LIKELY`/`UNLIKELY` macros to hint branch prediction. Consider branchless alternatives for simple conditionals. Order switch cases and if-else chains by frequency.
+**Branch Prediction:** Minimize unpredictable branches in hot paths. Use `LIKELY`/`UNLIKELY` macros to hint branch prediction, e.g. for error cases and other rare or otherwise costly cases, but NOT for predicting popular configurations. Consider branchless alternatives for simple conditionals. Order switch cases and if-else chains by frequency.
 
 **Memory and Resource Management:** Be mindful of memory allocations, especially in hot paths. Use RAII patterns, smart pointers, and RocksDB's memory management utilities appropriately.
 
@@ -149,6 +149,26 @@ Cache management is critical for RocksDB's performance.
 
 When reviewing RocksDB code (or preparing code for review), use this checklist:
 
+### Contract Boundaries
+- [ ] Is each behavior owned by the right layer? High-level policy (for example,
+  "compaction wants this I/O mode") should live at the caller/policy layer, while
+  lower layers should expose generic mechanisms (for example, "open a fresh
+  reader", "skip shared cache insertion", or "use these FileOptions").
+- [ ] Do comments and names describe local contracts rather than leaking a
+  specific caller's rationale into reusable APIs? Generic code should not need to
+  know about one current use case unless the API itself is intentionally
+  use-case-specific.
+- [ ] Does each flag or parameter control one coherent behavior? If one boolean
+  starts implying ownership, cache policy, I/O mode, prefetching, and caller
+  identity, split it into explicit flags or an options struct.
+- [ ] Could a future caller use this lower-level API without accidentally
+  inheriting assumptions from compaction, backup, user reads, or a particular
+  table format? If not, tighten the contract with assertions, clearer names, or
+  a narrower API.
+- [ ] Are implementation details not being used as policy signals? Prefer an
+  explicit contract over inferring behavior from incidental fields such as file
+  options, cache handles, or current table-reader state.
+
 ### Correctness
 - [ ] Does the change preserve database semantics (e.g., snapshot isolation, key ordering)?
 - [ ] Are all error cases handled appropriately?
@@ -204,6 +224,14 @@ The following patterns emerged as frequent sources of review feedback:
 
 10. **Platform Compatibility:** Ensure changes work correctly on all supported platforms (Linux, Windows, macOS) and with all supported compilers (GCC, Clang, MSVC).
 
+11. **Contract Boundary Leaks:** When a change plumbs a new option or use-case
+specific behavior through multiple subsystems, review the call chain for
+contract leaks. Caller-specific rationale belongs at the call site or public API
+documentation; reusable layers should expose precise, layer-local capabilities.
+Watch especially for comments mentioning one caller in generic code, booleans
+that silently bundle several behaviors, and downstream code inferring policy
+from an implementation detail instead of an explicit option.
+
 ---
 
 ## Important tips
@@ -215,27 +243,21 @@ The following patterns emerged as frequent sources of review feedback:
 * Don't manually edit BUCK file, after updating src.mk, run
     /usr/local/bin/python3 buckifier/buckify_rocksdb.py to update it
 * For -j in make command, use the number of CPU cores to decide it.
+* When searching for references to something (a symbol, library, etc.), do not
+  restrict or truncate your search based on presumed relevance or scope. It is
+  important and time-saving to keep the repo reasonably consistent across
+  different build systems, programming languages, and even between
+  documentation and implementation.
 
-### When to run `make clean` (avoid mixing build modes)
+### Avoiding mixed build modes with Make (use `AUTO_CLEAN=1`)
 
-The Makefile does **not** track build mode, so object files from a prior
-build are silently reused even when compiled with different flags, leading
-to confusing linker errors, sanitizer false negatives, ODR violations, or
-"phantom" bugs.
-
-Run `make clean` before switching any of these:
-* **`ASSERT_STATUS_CHECKED=1` ↔ unset** — changes the `Status` class layout (ABI break).
-* **Sanitizer builds** — toggling any of `COMPILE_WITH_ASAN=1`,
-    `COMPILE_WITH_UBSAN=1`, `COMPILE_WITH_TSAN=1` on/off.
-* `DEBUG_LEVEL=0` (release) ↔ `DEBUG_LEVEL=1` (debug, default for `make dbg`).
-* Different compilers, `OPT` levels, or other flags affecting codegen/ABI.
-
-**Notable exception:** `DEBUG_LEVEL=2` can be safely mixed with
-`DEBUG_LEVEL=1` — rebuild a subset of files with `DEBUG_LEVEL=2` to get
-extra/more accurate runtime checks for those files without a full clean.
-
-When in doubt, `make clean` is cheap insurance compared to chasing a
-phantom bug.
+Object files are written to the same paths regardless of build flags, so
+reusing objects from a prior build with different flags causes confusing
+linker errors, etc. This problem is essentially avoidable by ALWAYS using
+`AUTO_CLEAN=1 make -j<n> <something>` for manual make invocations. This
+will automatically clean object files if the build parameters/flavor have
+changed. The `build_tools/rockstest.sh` / `rocksptest.sh` helpers described
+below set `AUTO_CLEAN=1` for you.
 
 ### Source checks
 * Run `make check-sources` before committing. This catches non-ASCII
@@ -244,6 +266,22 @@ phantom bug.
     smart quotes, etc.) in comments or strings -- use ASCII equivalents
     (`--` instead of em dash, `'` instead of smart quote, etc.).
 
+### License headers
+* Every new source file needs a license header. For a file that does **not**
+    carry an outside/third-party copyright, use the standard Meta dual-licensed
+    header (the dual-license designation is required -- a bare
+    "All Rights Reserved" copyright is not an acceptable open-source header):
+    ```
+    //  Copyright (c) Meta Platforms, Inc. and affiliates.
+    //  This source code is licensed under both the GPLv2 (found in the
+    //  COPYING file in the root directory) and Apache 2.0 License
+    //  (found in the LICENSE.Apache file in the root directory).
+    ```
+    Use a `#` comment prefix instead of `//` for shell, Python, and Makefile
+    fragments.
+* Files derived from an external source (e.g. LevelDB) keep their original
+    upstream copyright line in addition to the header above.
+
 ### RTTI and dynamic_cast
 * Production code and `db_stress` must build in **release mode
     (`-fno-rtti`)**. Do not use `dynamic_cast` anywhere except unit tests.
@@ -251,6 +289,36 @@ phantom bug.
     `dynamic_cast` in debug builds, plain `static_cast` in release).
 * Unit tests (`*_test.cc`) are built in debug mode with RTTI enabled.
 
+### Cross-platform / portability
+Local `make` only exercises Linux with GCC/Clang, but CI
+(`.github/workflows/pr-jobs.yml` and `nightly.yml`) gates on a much wider
+matrix, so portability breaks are invisible locally until CI fails. Code must
+build (and where noted, run tests) across:
+
+| Axis | Must support |
+|------|--------------|
+| OS | Linux (x86_64 + ARM), macOS, Windows |
+| Architecture | Little-endian and big-endian targets, including Linux on s390x |
+| Compiler | GCC, Clang (libstdc++ **and** libc++), AppleClang, **MSVC (VS2022)**, MinGW (Linux cross-compile, build-only, no gflags) |
+| Build system | Make, CMake, and BUCK (internal) -- keep all in sync (see "Build system" above) |
+| Config | release (`-fno-rtti`), `ASSERT_STATUS_CHECKED`, ASAN/UBSAN/TSAN, folly, unity build, JNI/Java |
+
+Treat these as constraints to satisfy and infer the specifics from them before
+adding any system header, libc call, or compiler-specific construct. The most
+common trap: anything that compiles under GCC/Clang on Linux but not under
+**MSVC/MinGW** -- e.g. unguarded POSIX-only headers/functions (`<unistd.h>`,
+`<sys/*.h>`, `getpid`, `_exit`, ...) or GCC/Clang extensions
+(`__attribute__`, `__builtin_*`, VLAs, `alloca`). Prefer the `port::`/`Env`
+abstractions; otherwise guard with `#ifdef OS_WIN` (POSIX `<unistd.h>` ->
+Windows `<process.h>`). Because libc++ is also tested, include what you use
+rather than relying on libstdc++ transitive includes.
+
+Do not use `static_assert` to reject a supported architecture for an optional
+feature. Check an existing `port::` capability such as `port::kLittleEndian`
+and disable only the unsupported feature so the rest of RocksDB still builds.
+Use a portable representation when the feature itself must work across
+architectures.
+
 ### Unit Test
 * After all of the unit tests are added, review them and try to extract common
     reusable utility functions to reduce code duplication due to copy past between
@@ -258,14 +326,29 @@ phantom bug.
 * Don't use sleep to wait for certain events to happen. This will cause test to
     be flaky. Instead, use sync point to synchronize thread progress.
 * Cap unit test execution with 60 seconds timeout.
-* When there are multiple unit tests need to be executed, try to use
-    gtest_parallel.py if available. E.g.
-    python3 ${GTEST_PARALLEL}/gtest_parallel.py ./table_test
-* After writing a test, stress-test for flakiness:
+* To build and run unit tests locally, prefer these helper scripts:
+    * `build_tools/rocksptest.sh <test_binary> [more_binaries...] [args...]`
+        builds the binary(ies) with parallel make and `AUTO_CLEAN=1` and runs
+        them under gtest-parallel, sharding the test cases across CPUs. Prefer
+        this whenever running more than a couple of test cases, e.g.
+        `build_tools/rocksptest.sh table_test` or
+        `build_tools/rocksptest.sh db_test env_test --gtest_filter=*Foo*`.
+    * `build_tools/rockstest.sh <test_binary> [args...]` builds with parallel
+        make and `AUTO_CLEAN=1` and runs the binary directly (serially).
+        Use it only for a very small number of test cases, e.g.
+        `build_tools/rockstest.sh db_test --gtest_filter=*MixedSlowdown*`.
+* After writing a test, stress-test for flakiness (AUTO_CLEAN handles the
+    rebuild needed by the `COERCE_CONTEXT_SWITCH=1` flag change):
     ```bash
-    COERCE_CONTEXT_SWITCH=1 make {test_binary}
-    ./{test_binary} --gtest_filter="*YourTestName*" --gtest_repeat=5
+    COERCE_CONTEXT_SWITCH=1 build_tools/rockstest.sh {test_binary} -r100 \
+        --gtest_filter="*YourTestName*"
     ```
+* For CI-style flaky tests that do not reproduce with `gtest_parallel.py`,
+    `--gtest_repeat`, or normal coerce-mode runs, inspect
+    `tools/gtest_parallel_repro.py --help`.
+* Each unit test file has overheads, so avoid creating new unit test files
+  for random minor features. Consider adding to slice_test, db_etc3_test, or
+  others.
 
 ### Unit test dedup guidelines
 * Extract helper functions for repeated patterns such as object
@@ -313,17 +396,20 @@ phantom bug.
 
 ### Adding release note
 * Release note should be kept short at high level for external user consumption.
+* If more than single markdown line, consider how their formatting will be
+    integrated into HISTORY.md.
 
 ### Blog posts (docs/_posts)
 * Blog post authors must be defined in `docs/_data/authors.yml` to be displayed
 
 ### Final verification of the change
-* Execute make clean to clean all of the changes.
-* Execute make check to build all of the changes and execute all of the tests.
-    Note that executing all of the tests could take multiple minutes.
-* Run `ASSERT_STATUS_CHECKED=1 make check` to verify all Status objects are
-    properly checked. This catches missing error handling that can lead to
-    silent data corruption.
+* Execute `AUTO_CLEAN=1 make check` to build all of the changes and execute all
+    of the tests. `AUTO_CLEAN=1` ensures a clean rebuild if your previous build
+    used different parameters. Note that executing all of the tests could take
+    multiple minutes.
+* Run `AUTO_CLEAN=1 ASSERT_STATUS_CHECKED=1 make check` to verify all Status
+    objects are properly checked. This catches missing error handling that can
+    lead to silent data corruption.
 
 ### Monitoring make check progress
 * Use `make check-progress` to get machine-parseable JSON progress while
@@ -331,7 +417,7 @@ phantom bug.
     builds without timeout issues.
 * Run `make check` in background, then poll progress:
     ```bash
-    make check &
+    AUTO_CLEAN=1 make check &
     # Poll periodically:
     make check-progress
     ```
@@ -356,9 +442,10 @@ phantom bug.
 
 ### Executing benchmark using db_bench
 * Since the goal is to measure performance, we need to build a release binary
-    using `make clean && DEBUG_LEVEL=0 make db_bench`. If there is an engine
-    crash due to bug, we need to switch back to debug build. Make sure to run
-    `make clean` before running `make dbg`.
+    using `AUTO_CLEAN=1 DEBUG_LEVEL=0 make db_bench`. If there is an engine
+    crash due to a bug, switch back to a debug build with
+    `AUTO_CLEAN=1 make dbg`; `AUTO_CLEAN=1` handles the release<->debug rebuild
+    automatically.
 
 ### Formatting code
 * After making change, use `make format-auto` to auto-apply formatting without

CMakeLists.txt

@@ -420,13 +420,6 @@ if(WITH_NUMA)
   list(APPEND THIRDPARTY_LIBS NUMA::NUMA)
 endif()
 
-option(WITH_TBB "build with Threading Building Blocks (TBB)" OFF)
-if(WITH_TBB)
-  find_package(TBB REQUIRED)
-  add_definitions(-DTBB)
-  list(APPEND THIRDPARTY_LIBS TBB::TBB)
-endif()
-
 # Stall notifications eat some performance from inserts
 option(DISABLE_STALL_NOTIF "Build with stall notifications" OFF)
 if(DISABLE_STALL_NOTIF)
@@ -705,10 +698,19 @@ if(USE_FOLLY)
 
     include(${FOLLY_INST_PATH}/lib/cmake/folly/folly-config.cmake)
 
-    # Fix gflags library name for debug builds
+    # Fix gflags library name for debug builds. The getdeps gflags shared
+    # library is versioned (e.g. libgflags_debug.so.2.3), so resolve the
+    # actual file by glob rather than hard-coding a version.
     if(CMAKE_BUILD_TYPE STREQUAL "Debug")
+      file(GLOB GFLAGS_DEBUG_SHARED_LIBS
+        "${GFLAGS_INST_PATH}/lib/libgflags_debug.so.*")
+      if(NOT GFLAGS_DEBUG_SHARED_LIBS)
+        message(FATAL_ERROR
+          "Could not find getdeps libgflags_debug.so under ${GFLAGS_INST_PATH}/lib")
+      endif()
+      list(GET GFLAGS_DEBUG_SHARED_LIBS 0 GFLAGS_DEBUG_SHARED_LIB)
       set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -Wl,-rpath=${GFLAGS_INST_PATH}/lib")
-      set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} ${GFLAGS_INST_PATH}/lib/libgflags_debug.so.2.2")
+      set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} ${GFLAGS_DEBUG_SHARED_LIB}")
     endif()
   endif()
 
@@ -775,6 +777,7 @@ set(SOURCES
         db/blob/blob_file_meta.cc
         db/blob/blob_file_reader.cc
         db/blob/blob_garbage_meter.cc
+        db/blob/blob_gen2_format.cc
         db/blob/blob_log_format.cc
         db/blob/blob_log_sequential_reader.cc
         db/blob/blob_log_writer.cc
@@ -999,6 +1002,7 @@ set(SOURCES
         util/simple_mixed_compressor.cc
         util/compression_context_cache.cc
         util/concurrent_task_limiter_impl.cc
+        util/coro_stats_util.cc
         util/crc32c.cc
         util/data_structure.cc
         util/dynamic_bloom.cc
@@ -1020,6 +1024,7 @@ set(SOURCES
         util/xxhash.cc
         utilities/agg_merge/agg_merge.cc
         utilities/backup/backup_engine.cc
+        utilities/copy_engine/copy_engine.cc
         utilities/blob_db/blob_compaction_filter.cc
         utilities/blob_db/blob_db.cc
         utilities/blob_db/blob_db_impl.cc
@@ -1637,6 +1642,7 @@ if(WITH_TESTS)
         utilities/cassandra/cassandra_row_merge_test.cc
         utilities/cassandra/cassandra_serialize_test.cc
         utilities/checkpoint/checkpoint_test.cc
+        utilities/copy_engine/copy_engine_test.cc
         utilities/sorted_run_builder/sorted_run_builder_test.cc
         utilities/env_timed_test.cc
         utilities/fault_injection_fs_test.cc
@@ -1704,7 +1710,21 @@ if(WITH_TESTS)
       )
       target_link_libraries(${exename}${ARTIFACT_SUFFIX} testutillib${ARTIFACT_SUFFIX} testharness gtest ${THIRDPARTY_LIBS} ${ROCKSDB_LIB})
       if(NOT "${exename}" MATCHES "db_sanity_test")
-        gtest_discover_tests(${exename} DISCOVERY_TIMEOUT 120)
+        if(CMAKE_VERSION VERSION_GREATER_EQUAL 3.18)
+          # Discover tests at ctest time instead of during the parallel build.
+          # POST_BUILD discovery (the default) runs each freshly-built test
+          # binary while the -j build is still compiling; on resource-constrained
+          # CI (notably the arm64 macOS runners) that discovery run can yield
+          # empty output, and CMake's JSON parser then aborts the whole build
+          # with "Syntax error: value, object or array expected". PRE_TEST moves
+          # discovery out of the build phase so a slow/empty enumeration no
+          # longer fails the build.
+          gtest_discover_tests(${exename}
+                               DISCOVERY_TIMEOUT 120
+                               DISCOVERY_MODE PRE_TEST)
+        else()
+          gtest_discover_tests(${exename} DISCOVERY_TIMEOUT 120)
+        endif()
         add_dependencies(rocksdb_check ${exename}${ARTIFACT_SUFFIX})
       endif()
   endforeach(sourcefile ${TESTS})

HISTORY.md

@@ -1,6 +1,98 @@
 # Rocksdb Change Log
 > NOTE: Entries for next release do not go here. Follow instructions in `unreleased_history/README.txt`
 
+## 11.8.1 (07/28/2026)
+### Public API Changes
+* Add `DBOptions::read_io_executor_threads` to increase the maximum thread count of the shared filesystem read I/O executor used for asynchronous reads when opening a DB. Opening a DB never reduces the shared executor thread count. Change the experimental `FSRandomAccessFile::SubmitReadAsync()` API to return whether the asynchronous path was used, and add the `FILE_SUBMIT_ASYNC_READ_FALLBACK` ticker to count calls that fall back to synchronous reads.
+
+### Performance Improvements
+* When `DBOptions::avoid_unnecessary_blocking_io` is true, obsolete `OPTIONS-*` files found during DB open are deleted by background purge instead of synchronously on the opening thread.
+
+## 11.8.0 (07/28/2026)
+### Public API Changes
+* Add callback-based asynchronous read APIs, `DB::GetAsync()` and `DB::MultiGetAsync()`. The idea is that when IO is required, RocksDB can suspend its internal coroutine read path, allowing the read executor thread to do other work. When the IO is complete, RocksDB invokes the user callback. This requires filesystem support for full performance benefits. A new filesystem API `FSRandomAccessFile::SubmitReadAsync()` is introduced for this. Unlike `ReadAsync`, the filesystem is responsible for eventually calling the callback. `FileSystem::GetReadExecutor()` returns the IO executor whose EventBases run coroutine read processing. In the Posix filesystem, an IO-uring based event loop is used to complete the IO.
+* Added `FlushOptions::listener_wait` (default `false`). When set together with `FlushOptions::wait == true`, `DB::Flush()` will not return until the registered `EventListener::OnFlushCompleted` callbacks for the flushed memtables have finished running. By default (`false`), `Flush(wait=true)` may return as soon as the flush result is committed, which can be before (or while) the `OnFlushCompleted` callbacks execute on the background flush thread. Also added the corresponding C API `rocksdb_flushoptions_set_listener_wait()` / `rocksdb_flushoptions_get_listener_wait()`.
+* Deprecated `PinnableWideColumns::serialized_size()` in favor of the new `PinnableWideColumns::payload_size()`, which returns the total size of the columns' names and values (for a plain value this equals the value size). Relatedly, wide-column read accounting changed subtly: read statistics for entities (`BYTES_READ`, `BYTES_PER_READ`, and the per-key byte counts for `GetEntity`/`MultiGetEntity`) and the `ReadOptions::value_size_soft_limit` threshold now measure entities by this payload size instead of the serialized entity size. This drops serialization framing bytes from those measurements (so reported/limited sizes for entity reads are slightly smaller than before) and makes plain-value and wide-column accounting consistent.
+
+### Behavior Changes
+* `ReadOptions::value_size_soft_limit` now also bounds the blob values resolved from pre-flush blob direct write references during `MultiGet`/`MultiGetEntity` (previously these were resolved after the limit check and read without bound). Enforcement is "always make progress": at least one key is read even if its value alone exceeds the limit, and only once the returned size exceeds the limit do subsequent keys get `Status::Aborted` -- so a caller retrying the aborted keys cannot loop forever on a single value that by itself exceeds the limit.
+
+### Performance Improvements
+* Reduced copying when reading wide-column entities that have blob-referenced columns, whether stored in separate blob files or embedded (same-file) as blob records in the same SST. On the point-lookup (`GetEntity`/`MultiGetEntity`) and blob-backed memtable/direct-write read paths, resolving these columns no longer re-serializes the whole entity into a fresh buffer: inline columns are referenced in place in the pinned entity and each resolved blob value is referenced in place from its fetched buffer. This is enabled by an internal change to `PinnableWideColumns` that lets its columns reference multiple backing buffers.
+
+## 11.7.0 (07/16/2026)
+### Public API Changes
+* Added `RangeLockManagerHandle::SetIsKilledCallback()`, a `std::function<bool()>` predicate polled during range-lock waits. When it returns true, a blocked `GetRangeLock()` wait returns promptly with `Status::Aborted()` instead of waiting for the lock timeout. Range-lock users that install no callback keep the original wait-until-grant-or-timeout behavior.
+* Added `MultiScanArgs::reverse` to support reverse MultiScan reads over bounded scan ranges.
+
+### Behavior Changes
+* When using `LockWAL()`, live-file capture APIs such as checkpoint and backup now flush WAL-disabled unpersisted data; they can block until `UnlockWAL()` or return `Aborted` when called from the same thread that holds `LockWAL()` to avoid deadlock. `UnlockWAL()` must now also be called from the same thread that called `LockWAL()`.
+
+### Performance Improvements
+* Fixed slow DB open with `best_efforts_recovery`, and slow catch-up of secondary and follower (read-only) instances, in which time and memory grew quadratically with the number of files in the DB (so the slowdown was worst for large DBs with long history).
+
+## 11.6.0 (07/02/2026)
+### New Features
+* Added EXPERIMENTAL embedded blob SST support through `SstFileWriter::OpenWithEmbeddedBlobs()`, storing eligible large values as same-file blob records in block-based SST files and resolving them transparently for reads. This niche feature currently supports uncompressed embedded blobs only; compression options are placeholders and compression support is deferred to follow-up work.
+
+### Public API Changes
+* Expanded the C API (`include/rocksdb/c.h`) with a large set of new `rocksdb_*` functions, mostly option getters/setters plus table-properties, job/event-listener, and metadata accessors, a WAL filter, a ReadOptions table filter, and a backup exclude-files callback. Many are now produced by a new semi-automated generator (`tools/c_api_gen/`) from the C++ headers; `include/rocksdb/c.h` remains a single self-contained header and the signatures of pre-existing functions are unchanged.
+
+### Bug Fixes
+* Reverted PR14831 that made range_lock_manager aware of reverse-order CF
+* Fixed a bug in `RandomAccessFileReader::ReadAsync` where an already-aligned direct-IO read request with a null `scratch` and a caller-provided `aligned_buf` would take the "already aligned" fast path and submit the null buffer to the underlying async read (e.g. a null iovec base to io_uring, failing with EFAULT). This could surface as spurious iterator failures during async prefetch (MultiScan with async IO) on direct-IO databases. The async path now allocates a backing buffer in this case, matching the synchronous `Read` path.
+* Fixed a bug where closing a read-only DB instance could delete live SST files created by a concurrent read-write DB sharing the same directory.
+
+## 11.5.0 (06/16/2026)
+### New Features
+* External table readers that open files through `ExternalTableOptions::fs` now update RocksDB SST/file-read statistics and file IO listener callbacks, making external file IO activity visible in existing read metrics.
+* Added `DB::PrepareFileIngestion()`, a two-phase form of `IngestExternalFile`/`IngestExternalFiles`. `PrepareFileIngestion()` performs all of the work that does not require the DB mutex (validating the arguments, reading each external file's metadata, and linking/copying the files into the DB) and returns an opaque `FileIngestionHandle`. `DB::CommitFileIngestionHandle()` (or `DB::CommitFileIngestionHandles()` for several handles at once) then makes the prepared files visible; multiple handles are committed atomically in a single MANIFEST write, or none are. `FileIngestionHandle::Abort()` (or simply destroying the handle) cancels a prepared ingestion and rolls it back. This gives flexibility to the application to Prepare a file separately from when its committed, and may shorten an applications critical path on `IngestExternalFile`.
+* Added two stats histograms reporting the per-call latency (in microseconds) of each successful `IngestExternalFile`/`IngestExternalFiles` call, split by phase: `rocksdb.ingest.external.file.prepare.micros` (argument validation and reading/validating/linking the external files, which does not block live writes) and `rocksdb.ingest.external.file.run.micros` (the ingestion performed under the DB mutex while live writes are blocked). Failed ingestions are not recorded.
+* Added `DBOptions::use_direct_io_for_compaction_reads` (default false). When enabled, compaction-input SST reads use `O_DIRECT` while user reads remain buffered, avoiding page-cache eviction of hot user-read data by sequential compaction scans. Pair with `use_direct_io_for_flush_and_compaction = true` on write-heavy workloads for direct I/O on both compaction inputs and outputs. Rejected at Open when combined with `allow_mmap_reads = true`. No-op when `use_direct_reads = true` is already set (since all reads are already direct).
+
+### Public API Changes
+* `DB::GetPreparedFileInfoForExternalSstIngestion()` can now prepare metadata for a live DB-generated SST file so it can be passed to `IngestExternalFileArg::file_infos`. The input path must exactly match a live table file owned by the source DB, and the returned metadata handle must outlive the ingestion.
+* `ExternalSstFileInfo` gained a `prepared_file_info` member produced by `SstFileWriter::Finish`, and `IngestExternalFileArg` gained a `file_infos` field: a vector of borrowed `const PreparedFileInfo*` pointers, parallel to `external_files`. The owning `ExternalSstFileInfo::prepared_file_info` handle must outlive the ingestion. When set, `IngestExternalFiles()` reuses the supplied per-file metadata instead of re-opening and scanning each file to recompute it, avoiding that extra I/O.
+* Corrected the public C++ option spelling from `memtable_veirfy_per_key_checksum_on_seek` to `memtable_verify_per_key_checksum_on_seek`. RocksDB continues to accept the old misspelled OPTIONS-file key for compatibility, while newly serialized OPTIONS files use the corrected key.
+* Added experimental read-scoped block buffer provider API, configured through `ReadOptions::read_scoped_block_buffer_provider`, for supported block-based table iterator scans and MultiScan reads to use caller-provided read-scoped storage for final data-block contents. When configured, supported provider-backed scan data-block reads bypass the data-block cache. The provider is ignored when mmap reads are enabled. RocksDB may still use ordinary temporary scratch for serialized block bytes, such as when a block may be compressed. Get and MultiGet do not currently provide API guarantees for this provider.
+* Added `FileSystem::SyncFile()` and `Env::SyncFile()` public APIs for syncing or fsyncing a file by name without requiring callers to reopen it as writable.
+
+### Behavior Changes
+* The default `compression` for column families is now `kLZ4Compression` rather than `kSnappyCompression`. This affects only column families that do not explicitly set `compression`, and only newly-written SST files. The change is fully compatible: existing data remains readable with no migration needed, as RocksDB selects the decompressor per block. LZ4 offers slightly better compression ratios and decompression CPU efficiency vs. Snappy and similar compression CPU, tested across various server CPUs. When support is not compiled in, the fallback path for default compression is LZ4 -> Snappy -> NoCompression.
+* Disable parallel compression (ignore `CompressionOptions::parallel_threads`) for fast built-in compressors: Snappy, LZ4 (accelerated, not LZ4HC), and accelerated levels of ZSTD (level < 0). These do not generally benefit from parallel compression. This behavior can be overridden with a custom Compressor from a custom CompressionManager.
+* `PosixFileSystem::OptimizeForCompactionTableRead` now delegates to the base `FileSystem::OptimizeForCompactionTableRead` implementation before applying its Linux-only compaction-readahead clamp (added for #12038). The returned `FileOptions::use_direct_reads` therefore reflects `DBOptions::use_direct_reads`, consistent with the base class and with other `FileSystem` implementations. Previously the override ignored the base implementation and keyed both the returned options and the Linux readahead clamp off the incoming `FileOptions` rather than `DBOptions`. At the in-tree call site the incoming `FileOptions::use_direct_reads` already matches `DBOptions::use_direct_reads`, so this is effectively a no-op for existing configurations; it removes a latent inconsistency where a caller passing `FileOptions` whose `use_direct_reads` disagreed with `DBOptions` would have had the global flag silently ignored for compaction-input reads on Linux. (Note: the `#12038` readahead clamp still keys off the global `use_direct_reads`; it is not skipped for the compaction-only `use_direct_io_for_compaction_reads` path, since that flag enables O_DIRECT after this hook runs.)
+* Brought more unity to `kLZ4Compression` and `kLZ4HCCompression` by giving each access to the other's compression levels. Negative (fast) values previously only available to LZ4 and positive (slow) values previously only available to LZ4HC are now available to both, so configuring a non-default `compression_opts.level` now selects the LZ4 compressor variant. This is a behavior change for previously tolerated but dubious configurations such as positive compression level with `kLZ4Compression`. `kLZ4Compression` and `kLZ4HCCompression` keep their effective default levels, equivalent to -1 and 9 respectively.
+* Removed a discontinuity in `kZSTD` compression levels at `compression_opts.level == 0` by mapping that setting to `level = -1` rather than to `level = 3` as the ZSTD library does internally. This improves the compression auto-tuning landscape.
+
+### Bug Fixes
+* Fixed a bug where the range lock manager would crash with an assertion failure when using a reverse comparator column family.
+
+### Performance Improvements
+* Reduced commit latency when ingesting many external files by allowing `IngestExternalFileOptions::file_opening_threads` to open table readers for committed ingested files using multiple threads.
+* Reduced commit latency for large external file ingestions into the last level by adding `IngestExternalFileOptions::prefetch_lmax_index_and_filter_blocks`, which can skip commit-time index and filter block prefetching for cache-backed table metadata.
+
+## 11.4.0 (06/02/2026)
+### Public API Changes
+* Added `rocksdb_options_set_memtable_batch_lookup_optimization()` and `rocksdb_options_get_memtable_batch_lookup_optimization()` to the C API, exposing the existing `AdvancedColumnFamilyOptions::memtable_batch_lookup_optimization` field. This allows C API users (and downstream language bindings) to enable the skip-list memtable's batch-lookup optimization for `MultiGet`, which caches the search path between consecutive keys and reduces per-key cost from O(log N) to O(log d) where d is the distance between consecutive keys.
+* Added `rocksdb_readoptions_set_optimize_multiget_for_io()` and `rocksdb_readoptions_get_optimize_multiget_for_io()` to the C API, exposing the existing `ReadOptions::optimize_multiget_for_io` field. This allows C API users (and downstream language bindings) to opt out of the multi-level parallel MultiGet path, which only takes effect when the library is built with `USE_COROUTINES`.
+* Added `rocksdb_block_based_table_index_block_search_type_auto` enum constant and the `rocksdb_block_based_options_set_uniform_cv_threshold()` setter to the C API. The constant exposes `BlockSearchType::kAuto`, and the setter exposes `BlockBasedTableOptions::uniform_cv_threshold`. Both are required for `kAuto` index-block search to take effect from the C API: without setting `uniform_cv_threshold >= 0`, the per-block `is_uniform` footer bit is never written, and `kAuto` falls back to binary search at read time.
+* Added `EventListener::OnDBShutdownBegin`, a callback that fires once when a DB begins shutdown, including when RocksDB cleans up a failed `DB::Open()` attempt.
+* Added a new PerfContext counter `blob_cache_read_byte` for blob cache bytes read.
+
+### Behavior Changes
+* Remote compaction now falls back to local compaction when the primary cannot deserialize a successful `CompactionServiceResult` before installing any remote output files.
+* StringToMap (rocksdb/convenience.h) now preserves the outer braces of nested values so each map entry is in self-contained form and can be embedded directly into another `key=value;` string (e.g. SetOptions) without losing inner ';' delimiters. A new symmetric MapToString utility is provided.
+
+### Bug Fixes
+* Fixed a bug where `AbortAllCompactions()` could leave automatic compaction work unscheduled when aborting before the background worker picked it, causing compaction and write stalls until DB restart.
+* Fix a bug where db had a false positive compaction corruption error due to remote compaction service result with `has_accurate_num_input_records=false` not being serialized.
+* Fixed WritePrepared TransactionDB cleanup after retryable commit or rollback write failures so prepared transactions remain rollbackable instead of leaving unresolved prepared state.
+* Fix a rare corruption bug for tiered compaction that incorrectly moved last level range tombstones into proximal level. This corruption error is surfaced when force_consistency_checks is enabled.
+
+### Performance Improvements
+* Reduce the likelihood of user-facing metadata operations blocking on MANIFEST rotation when file creation is slow, such as on remote storage. MANIFEST write batches containing foreground edits from `CreateColumnFamily()`, `DropColumnFamily()`, `CreateColumnFamilyWithImport()`, `IngestExternalFile()`, and `DeleteFilesInRanges()` now get a relaxed file size threshold for triggering MANIFEST rotation; background-only MANIFEST write batches, such as compaction and flush, continue to use the normal threshold. This change should make auto-tuning manifest file size more attractive (see `max_manifest_file_size` and `max_manifest_space_amp_pct` options).
+
+
 ## 11.3.0 (05/15/2026)
 ### New Features
 * Add experimental DB option `async_wal_precreate` to precreate the next WAL file in a background thread and reduce foreground WAL rotation latency. The option is sanitized to false when WAL recycling is enabled.

Makefile

@@ -288,7 +288,7 @@ endif
 endif
 
 export JAVAC_ARGS
-CLEAN_FILES += make_config.mk rocksdb.pc
+CLEAN_FILES += rocksdb.pc
 
 ifeq ($(V), 1)
 $(info $(shell uname -a))
@@ -764,7 +764,6 @@ util/build_version.cc: $(filter-out $(OBJ_DIR)/util/build_version.o, $(LIB_OBJEC
 	$(AM_V_at)$(gen_build_version) > $@
 endif
 CLEAN_FILES += util/build_version.cc
-
 default: all
 
 #-----------------------------------------------
@@ -809,7 +808,8 @@ endif  # PLATFORM_SHARED_EXT
 .PHONY: check clean coverage ldb_tests package dbg gen-pc build_size \
 	release tags tags0 valgrind_check format static_lib shared_lib all \
 	rocksdbjavastatic rocksdbjava install install-static install-shared \
-	uninstall analyze tools tools_lib check-headers checkout_folly clang-tidy
+	uninstall analyze tools tools_lib check-headers checkout_folly clang-tidy \
+	check-c-api-c_test
 
 # Auto-configure git hooks on first build so developers do not need to run
 # "make install-hooks" manually. This is a no-op if already set.
@@ -1127,8 +1127,102 @@ ifndef SKIP_FORMAT_BUCK_CHECKS
 	$(MAKE) check-buck-targets
 	$(MAKE) check-sources
 	$(MAKE) check-workflow-yaml
+	$(MAKE) check-c-api-gen
 endif
 
+# Check that the auto-generated C API files are up to date. It regenerates the
+# fragments and the inlined c.h/c.cc and compares them against a snapshot of the
+# checked-in copies (no net change when everything is up to date). It requires
+# clang++ (libclang, used to parse the C++ headers) and clang-format. When those
+# are unavailable the staleness/compat sub-checks are SKIPPED with a message so
+# `make check` still works without the codegen toolchain; the link-completeness
+# sub-check always runs (it needs no toolchain).
+#
+# Pin the formatter to match CI by setting CLANG_FORMAT_BINARY, e.g.:
+#   make check-c-api-gen CLANG_FORMAT_BINARY=clang-format-21
+# This target is part of `make check` and is skipped by SKIP_FORMAT_BUCK_CHECKS.
+#
+# Set CHECK_C_API_GEN_STRICT=1 to turn every "skip" below into a hard error, so a
+# core CI job that runs this target cannot silently regress to a no-op if a
+# prerequisite (clang++, or the compat baseline ref) goes missing. The dedicated
+# build-linux-clang-21-no_test_run CI job runs this target with the flag set.
+# Any non-empty value other than 0/no/false enables strict mode.
+CLANG_FORMAT_BINARY ?=
+# Backward-compatibility baseline for the C API (signature-level) check. CI
+# overrides this with the PR's merge target; locally it falls back to main /
+# origin/main and is skipped if neither resolves.
+API_COMPAT_REF ?= main
+CHECK_C_API_GEN_STRICT ?=
+check-c-api-gen:
+	# Link-completeness is a property of the checked-in c.h/c.cc and needs no
+	# clang toolchain, so it always runs: every declared public C API function
+	# must have exactly one definition (guards against dropped wrappers that
+	# would break downstream language bindings at link time).
+	$(PYTHON) tools/c_api_gen/check_api_completeness.py
+	# Backward-compatibility: no public C function may be removed or have its
+	# signature changed vs the baseline. Skipped if the baseline ref is not
+	# resolvable locally (CI passes an explicit ref), unless CHECK_C_API_GEN_STRICT.
+	@strict=""; case "$(CHECK_C_API_GEN_STRICT)" in ""|0|no|NO|false|FALSE) ;; *) strict=1 ;; esac; \
+	ref=""; \
+	if git rev-parse --verify --quiet "$(API_COMPAT_REF)^{commit}" >/dev/null; then ref="$(API_COMPAT_REF)"; \
+	elif git rev-parse --verify --quiet "origin/$(API_COMPAT_REF)^{commit}" >/dev/null; then ref="origin/$(API_COMPAT_REF)"; fi; \
+	if [ -n "$$ref" ]; then \
+	  $(PYTHON) tools/c_api_gen/check_api_compatibility.py --ref "$$ref"; \
+	elif [ -n "$$strict" ]; then \
+	  echo "ERROR: C API compat baseline '$(API_COMPAT_REF)' not resolvable and CHECK_C_API_GEN_STRICT is set" >&2; exit 1; \
+	else \
+	  echo "Skipping C API backward-compatibility check ($(API_COMPAT_REF) not found; set API_COMPAT_REF)"; \
+	fi
+	# Staleness: regenerate and confirm the checked-in output is current. Needs a
+	# clang++ (the generator parses C++ ASTs). Use CLANG_CXX, which
+	# build_tools/build_detect_platform resolves once (the fbcode clang++ in
+	# fbcode builds, else a versioned clang++ on PATH) and passes explicitly, so
+	# the check and the generator agree on the same binary without heuristics.
+	@strict=""; case "$(CHECK_C_API_GEN_STRICT)" in ""|0|no|NO|false|FALSE) ;; *) strict=1 ;; esac; \
+	cf_arg=""; \
+	if [ -n "$(CLANG_FORMAT_BINARY)" ]; then cf_arg="--clang-format $(CLANG_FORMAT_BINARY)"; fi; \
+	if [ -n "$(CLANG_CXX)" ] && command -v "$(CLANG_CXX)" >/dev/null 2>&1; then \
+	  $(PYTHON) tools/c_api_gen/verify_generated_up_to_date.py --clang "$(CLANG_CXX)" $$cf_arg; \
+	elif [ -n "$$strict" ]; then \
+	  echo "ERROR: no clang++ found (CLANG_CXX unset) and CHECK_C_API_GEN_STRICT is set; cannot run the C API staleness check" >&2; exit 1; \
+	else \
+	  echo "Skipping C API codegen staleness check (no clang++ found; CLANG_CXX unset, install clang++ to enable)"; \
+	fi
+
+# Regenerate the checked-in C API generated files (fragments under c_api_gen/,
+# the inlined include/rocksdb/c.h and db/c.cc, and the round-trip tests). Uses
+# CLANG_CXX (resolved by build_tools/build_detect_platform) so it works in a gcc
+# build too. Aborts up front if no usable clang++ is available, so a bad
+# toolchain never leaves the tree in a partially regenerated state. Pin the
+# formatter to match CI with CLANG_FORMAT_BINARY, e.g.:
+#   make regen-c-api CLANG_FORMAT_BINARY=clang-format-21
+.PHONY: regen-c-api
+regen-c-api:
+	@if [ -z "$(CLANG_CXX)" ] || ! command -v "$(CLANG_CXX)" >/dev/null 2>&1; then \
+	  echo "ERROR: no clang++ found (CLANG_CXX unset); build on a host with clang++ or the fbcode toolchain to regenerate the C API" >&2; exit 1; \
+	fi
+	@cf_arg=""; \
+	if [ -n "$(CLANG_FORMAT_BINARY)" ]; then cf_arg="--clang-format $(CLANG_FORMAT_BINARY)"; fi; \
+	$(PYTHON) tools/c_api_gen/regen_all.py --clang "$(CLANG_CXX)" $$cf_arg
+
+# Quick local validation for C API generation plus the focused C API test.
+# This verifies the checked-in generated fragments as well as the inlined
+# include/rocksdb/c.h and db/c.cc outputs, then runs c_test in an isolated
+# TEST_TMPDIR to avoid stale-state failures.
+check-c-api-c_test:
+	$(PYTHON) tools/c_api_gen/verify_generated_up_to_date.py --clang "$(CLANG_CXX)"
+	$(MAKE) c_test
+	@tmpdir=$$(mktemp -d); \
+	  trap 'rm -rf "$$tmpdir"' EXIT; \
+	  echo "===== Running c_test with TEST_TMPDIR=$$tmpdir"; \
+	  if command -v timeout >/dev/null 2>&1; then \
+	    TEST_TMPDIR="$$tmpdir" timeout 60 ./c_test; \
+	  elif command -v gtimeout >/dev/null 2>&1; then \
+	    TEST_TMPDIR="$$tmpdir" gtimeout 60 ./c_test; \
+	  else \
+	    TEST_TMPDIR="$$tmpdir" ./c_test; \
+	  fi
+
 # TODO add ldb_tests
 check_some: $(ROCKSDBTESTS_SUBSET)
 	for t in $(ROCKSDBTESTS_SUBSET); do echo "===== Running $$t (`date`)"; ./$$t || exit 1; done
@@ -1278,8 +1372,13 @@ rocksdb.h rocksdb.cc: build_tools/amalgamate.py Makefile $(LIB_SOURCES) unity.cc
 	build_tools/amalgamate.py -I. -i./include unity.cc -x include/rocksdb/c.h -H rocksdb.h -o rocksdb.cc
 
 clean: clean-ext-libraries-all clean-rocks clean-rocksjava
+# Removed here rather than in clean-rocks (via CLEAN_FILES) so the build-parameter
+# auto-clean, which runs clean-rocks, doesn't delete the make_config.mk already
+# included by the in-progress make.
+	@rm -f make_config.mk
 
 clean-not-downloaded: clean-ext-libraries-bin clean-rocks clean-not-downloaded-rocksjava
+	@rm -f make_config.mk
 
 clean-rocks:
 # Not practical to exactly match all versions/variants in naming (e.g. debug or not)
@@ -1294,6 +1393,11 @@ clean-rocks:
 	else \
 		$(FIND) . -name "*.[oda]" -exec rm -f {} \; ; \
 	fi
+# Remove the build signature(s) LAST. Done after the object files are gone so
+# that a Ctrl+C partway through clean leaves the old signature in place: it
+# still matches the leftover objects, so a later build with different
+# parameters is still detected (signature usefulness is preserved).
+	@rm -f $(BUILD_SIG_FILE) jl/.build_signature jls/.build_signature
 
 clean-rocksjava: clean-rocks
 	rm -rf jl jls
@@ -1725,6 +1829,9 @@ backup_engine_test: $(OBJ_DIR)/utilities/backup/backup_engine_test.o $(TEST_LIBR
 checkpoint_test: $(OBJ_DIR)/utilities/checkpoint/checkpoint_test.o $(TEST_LIBRARY) $(LIBRARY)
 	$(AM_LINK)
 
+copy_engine_test: $(OBJ_DIR)/utilities/copy_engine/copy_engine_test.o $(TEST_LIBRARY) $(LIBRARY)
+	$(AM_LINK)
+
 sorted_run_builder_test: $(OBJ_DIR)/utilities/sorted_run_builder/sorted_run_builder_test.o $(TEST_LIBRARY) $(LIBRARY)
 	$(AM_LINK)
 
@@ -2327,6 +2434,101 @@ ifeq ($(PLATFORM), OS_OPENBSD)
 endif
 export SHA256_CMD
 
+# ----------------------------------------------------------------------------
+# Build parameter change detection
+# ----------------------------------------------------------------------------
+# RocksDB writes object files to the same $(OBJ_DIR) paths regardless of
+# DEBUG_LEVEL, sanitizers (ASAN/TSAN/UBSAN), ASSERT_STATUS_CHECKED, RTTI, LTO,
+# COERCE_CONTEXT_SWITCH, etc.  Most of those are applied in this Makefile after
+# `include make_config.mk` and are therefore NOT reflected in make_config.mk,
+# so switching them and rebuilding silently mixes incompatible object files.
+# To guard against that, we hash the fully-resolved compile/link parameters
+# (which already embed make_config.mk via PLATFORM_*) and compare against the
+# value recorded from the previous build.  On mismatch the build stops so the
+# user can `make clean`.  Knobs:
+#   AUTO_CLEAN=1
+#     run `make clean-rocks` automatically on mismatch
+#   ALLOW_BUILD_PARAMETER_CHANGE=1
+#     skip the check entirely (e.g. intentionally mixing DEBUG_LEVEL=1 and
+#     DEBUG_LEVEL=2)
+# The signature is stored per-$(OBJ_DIR), so Java (jl/jls) builds are tracked
+# independently from the default build.
+BUILD_SIG_FILE := $(OBJ_DIR)/.build_signature
+# NOTE: deliberately NOT added to CLEAN_FILES. The signature is removed as the
+# very last step of clean-rocks so that an interrupted clean keeps it (see
+# clean-rocks), preserving change detection across a Ctrl+C'd clean.
+
+# Goals that do not compile object files into the tree (pure utilities,
+# informational, source/config checks): the change check is irrelevant for them.
+BUILD_SIG_NONBUILD_GOALS := \
+	clean clean-rocks clean-not-downloaded clean-rocksjava \
+	clean-not-downloaded-rocksjava clean-ext-libraries-all \
+	clean-ext-libraries-bin \
+	format format-auto check-format check-buck-targets check-headers \
+	check-sources check-workflow-yaml check-progress clang-tidy \
+	tags tags0 package jclean checkout_folly build_folly \
+	watch-log dump-log suggest-slow-tests list_all_tests gen-pc \
+	gen_parallel_tests check-c-api-gen \
+	setup-hooks install-hooks uninstall-hooks uninstall db_crashtest_tests
+# Goals that clean before building (depend on or invoke `clean`): they manage
+# their own freshness, so the check must not block them (it would error before
+# their built-in clean runs).
+BUILD_SIG_NONBUILD_GOALS += \
+	release coverage build_size analyze analyze_incremental \
+	asan_check asan_crash_test asan_crash_test_with_atomic_flush \
+	asan_crash_test_with_txn asan_crash_test_with_best_efforts_recovery \
+	whitebox_asan_crash_test blackbox_asan_crash_test \
+	ubsan_check ubsan_crash_test ubsan_crash_test_with_atomic_flush \
+	ubsan_crash_test_with_txn ubsan_crash_test_with_best_efforts_recovery \
+	whitebox_ubsan_crash_test blackbox_ubsan_crash_test
+
+# Goals that explicitly clean; never trip the check when one is requested.
+BUILD_SIG_CLEAN_GOALS := \
+	clean clean-rocks clean-not-downloaded clean-rocksjava \
+	clean-not-downloaded-rocksjava clean-ext-libraries-all \
+	clean-ext-libraries-bin
+
+BUILD_SIG_GOALS := $(if $(MAKECMDGOALS),$(MAKECMDGOALS),all)
+BUILD_SIG_DO_BUILD := $(filter-out $(BUILD_SIG_NONBUILD_GOALS),$(BUILD_SIG_GOALS))
+BUILD_SIG_CLEANING := $(filter $(BUILD_SIG_CLEAN_GOALS),$(MAKECMDGOALS))
+# Dry runs (-n/--just-print) must not write the stamp, clean, or error; the
+# parse-time $(shell) below would otherwise run even under -n. Note that
+# actual options are canonicalized and shorted as possible such that all
+# short options are in the first word of MAKEFLAGS.
+BUILD_SIG_DRYRUN := $(findstring n,$(firstword -$(MAKEFLAGS)))
+
+# Only enforce at the top level. Sub-makes (e.g. `check` invokes
+# $(MAKE) gen_parallel_tests / check-c-api-gen / check_0) inherit the parent's
+# build flags, so the top-level check already covers them; re-checking inside a
+# sub-make only causes spurious failures.
+ifneq ($(ALLOW_BUILD_PARAMETER_CHANGE),1)
+ifeq ($(MAKELEVEL),0)
+ifeq ($(BUILD_SIG_DRYRUN),)
+ifeq ($(BUILD_SIG_CLEANING),)
+ifneq ($(BUILD_SIG_DO_BUILD),)
+  BUILD_SIG := $(shell printf '%s' '$(CC)|$(CXX)|$(CFLAGS)|$(CXXFLAGS)|$(LDFLAGS)|$(EXEC_LDFLAGS)' | $(SHA256_CMD) | cut -d ' ' -f 1)
+  BUILD_SIG_OLD := $(shell cat $(BUILD_SIG_FILE) 2>/dev/null)
+  ifneq ($(BUILD_SIG_OLD),)
+  ifneq ($(BUILD_SIG_OLD),$(BUILD_SIG))
+  ifeq ($(AUTO_CLEAN),1)
+    $(info *** Build parameters changed since last build (OBJ_DIR=$(OBJ_DIR)); running 'make clean-rocks' because AUTO_CLEAN=1)
+    BUILD_SIG_CLEAN_OUTPUT := $(shell $(MAKE) clean-rocks 1>&2)
+    ifneq ($(.SHELLSTATUS),0)
+      $(error AUTO_CLEAN: 'make clean-rocks' failed (exit $(.SHELLSTATUS)); not building against a partially-cleaned tree)
+    endif
+  else
+    $(error Build parameters changed since the last build (OBJ_DIR=$(OBJ_DIR)). Existing object files are stale and must be removed. Run 'make clean', or set AUTO_CLEAN=1 to clean automatically, or ALLOW_BUILD_PARAMETER_CHANGE=1 to build anyway)
+  endif
+  endif
+  endif
+  # Record the current signature for the next build.
+  BUILD_SIG_WRITE := $(shell mkdir -p $(OBJ_DIR) 2>/dev/null; printf '%s\n' '$(BUILD_SIG)' > $(BUILD_SIG_FILE))
+endif
+endif
+endif
+endif
+endif
+
 zlib-$(ZLIB_VER).tar.gz:
 	curl --fail --output zlib-$(ZLIB_VER).tar.gz --location ${ZLIB_DOWNLOAD_BASE}/zlib-$(ZLIB_VER).tar.gz
 	ZLIB_SHA256_ACTUAL=`$(SHA256_CMD) zlib-$(ZLIB_VER).tar.gz | cut -d ' ' -f 1`; \

buckifier/buckify_rocksdb.py

@@ -145,11 +145,21 @@ def generate_buck(repo_path, deps_map):
         # always add range_tree, it's only excluded on ppc64, which we don't use internally
         src_mk["RANGE_TREE_SOURCES"] + src_mk["TOOL_LIB_SOURCES"],
         deps=[
+            "//folly:executor",
             "//folly/container:f14_hash",
+            "//folly/coro:baton",
             "//folly/coro:blocking_wait",
             "//folly/coro:collect",
             "//folly/coro:coroutine",
+            "//folly/coro:nothrow",
             "//folly/coro:task",
+            "//folly/executors:io_executor",
+            "//folly/executors:io_thread_pool_executor",
+            "//folly/executors/thread_factory:named_thread_factory",
+            "//folly/io/async:async_base",
+            "//folly/io/async:event_base_manager",
+            "//folly/io/async:io_uring_backend",
+            "//folly/io/async:request_context",
             "//folly/synchronization:distributed_mutex",
         ],
         headers=LiteralValue("glob([\"**/*.h\"])")
@@ -360,6 +370,7 @@ def generate_buck(repo_path, deps_map):
                         extra_compiler_flags=json.dumps(deps["extra_compiler_flags"]),
                     )
     BUCK.export_file("tools/db_crashtest.py")
+    BUCK.export_file("tools/fault_injection_log_parser.py")
 
     print(ColorString.info("Generated BUCK Summary:"))
     print(ColorString.info("- %d libs" % BUCK.total_lib))

buckifier/targets_builder.py

@@ -118,6 +118,12 @@ class TARGETSBuilder:
         self.total_bin = self.total_bin + 1
 
     def add_c_test(self):
+        # The actual c_test_bin target is defined by add_c_test_wrapper in the
+        # internal //rocks/buckifier:defs.bzl (not in this OSS repo). db/c_test.c
+        # #includes the generated c_api_gen/*.inc fragments, so under Buck's
+        # hermetic sandbox that wrapper must expose them as headers, e.g.
+        #   headers = native.glob(["c_api_gen/**/*.inc"])
+        # (Make/CMake resolve the include via -I. / PROJECT_SOURCE_DIR.)
         with open(self.path, "ab") as targets_file:
             targets_file.write(
                 b"""

build_tools/build_detect_platform

@@ -27,7 +27,6 @@
 #       -DLZ4                       if the LZ4 library is present
 #       -DZSTD                      if the ZSTD library is present
 #       -DNUMA                      if the NUMA library is present
-#       -DTBB                       if the TBB library is present
 #       -DMEMKIND                   if the memkind library is present
 #
 # Using gflags in rocksdb:
@@ -90,13 +89,36 @@ if test -z "$CC"; then
     fi
 fi
 
+# CLANG_CXX is the single clang++ detection point for the whole script: it is
+# used for CXX when building with clang, for CLANG_ANALYZER (scan-build), and by
+# tooling that needs clang even in a gcc build (e.g. the C API codegen, which
+# parses C++ ASTs). The fbcode config above already sets CLANG_CXX; otherwise
+# resolve it here.
+if test -z "$CLANG_CXX"; then
+    if [ -n "$CXX" ] && echo | $CXX -dM -E -x c++ - 2>/dev/null | grep -q '\b__clang__\b'; then
+        # Honor an already-configured clang compiler (e.g. CI sets
+        # CXX=clang++-21) so codegen parses headers with the same clang the
+        # build uses, not an unrelated clang++ that merely happens to come first
+        # on PATH (an image may ship both a bare clang++ and clang++-NN).
+        CLANG_CXX="$CXX"
+    else
+        # Otherwise take the first clang++ on PATH. Candidates cover a bare
+        # clang++, versioned names, and clang++.par (a Meta-internal wrapper
+        # often present on dev machines).
+        CLANG_CXX=$(command -v clang++ || command -v clang++-21 || command -v clang++-20 \
+            || command -v clang++-19 || command -v clang++-18 || command -v clang++-17 \
+            || command -v clang++-16 || command -v clang++-15 || command -v clang++-14 \
+            || command -v clang++-13 || command -v clang++.par || true)
+    fi
+fi
+
 if test -z "$CXX"; then
-    if [ "$USE_CLANG" -a -x "$(command -v clang++)" ]; then
-        CXX=clang++
+    if [ "$USE_CLANG" ] && [ -n "$CLANG_CXX" ]; then
+        CXX="$CLANG_CXX"
     elif [ -x "$(command -v g++)" ]; then
         CXX=g++
-    elif [ -x "$(command -v clang++)" ]; then
-        CXX=clang++
+    elif [ -n "$CLANG_CXX" ]; then
+        CXX="$CLANG_CXX"
     else
         CXX=g++
     fi
@@ -114,6 +136,17 @@ if test -z "$AR"; then
     fi
 fi
 
+# If the selected C++ compiler is actually clang (e.g. clang was the only
+# compiler available), reflect that as USE_CLANG so clang-specific logic in the
+# Makefile applies consistently. Detect it via the compiler's own __clang__
+# predefined macro rather than matching the binary path, which can false-positive
+# (e.g. a home directory containing "clang").
+if test -z "$USE_CLANG"; then
+    if echo | $CXX -dM -E -x c++ - 2>/dev/null | grep -q '\b__clang__\b'; then
+        USE_CLANG=1
+    fi
+fi
+
 # Detect OS
 if test -z "$TARGET_OS"; then
     TARGET_OS=`uname -s`
@@ -128,7 +161,7 @@ if test -z "$CLANG_SCAN_BUILD"; then
 fi
 
 if test -z "$CLANG_ANALYZER"; then
-    CLANG_ANALYZER=$(command -v clang++ 2> /dev/null)
+    CLANG_ANALYZER="$CLANG_CXX"
 fi
 
 if test -z "$FIND"; then
@@ -424,19 +457,6 @@ EOF
         fi
     fi
 
-    if ! test $ROCKSDB_DISABLE_TBB; then
-        # Test whether tbb is available
-        $CXX $PLATFORM_CXXFLAGS $LDFLAGS -x c++ - -o test.o -ltbb 2>/dev/null  <<EOF
-          #include <tbb/tbb.h>
-          int main() {}
-EOF
-        if [ "$?" = 0 ]; then
-            COMMON_FLAGS="$COMMON_FLAGS -DTBB"
-            PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -ltbb"
-            JAVA_LDFLAGS="$JAVA_LDFLAGS -ltbb"
-        fi
-    fi
-
     if ! test $ROCKSDB_DISABLE_JEMALLOC; then
         # Test whether jemalloc is available
         if echo 'int main() {}' | $CXX $PLATFORM_CXXFLAGS $LDFLAGS -x c++ - -o test.o -ljemalloc \
@@ -832,6 +852,10 @@ echo "ROCKSDB_MINOR=$ROCKSDB_MINOR" >> "$OUTPUT"
 echo "ROCKSDB_PATCH=$ROCKSDB_PATCH" >> "$OUTPUT"
 echo "CLANG_SCAN_BUILD=$CLANG_SCAN_BUILD" >> "$OUTPUT"
 echo "CLANG_ANALYZER=$CLANG_ANALYZER" >> "$OUTPUT"
+echo "CLANG_CXX=$CLANG_CXX" >> "$OUTPUT"
+if test -n "$USE_CLANG"; then
+  echo "USE_CLANG=$USE_CLANG" >> "$OUTPUT"
+fi
 echo "PROFILING_FLAGS=$PROFILING_FLAGS" >> "$OUTPUT"
 echo "FIND=$FIND" >> "$OUTPUT"
 echo "WATCH=$WATCH" >> "$OUTPUT"

build_tools/check-public-header.sh

@@ -1,5 +1,8 @@
 #!/usr/bin/env bash
-# Copyright (c) Meta Platforms, Inc. and affiliates. All Rights Reserved.
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# This source code is licensed under both the GPLv2 (found in the
+# COPYING file in the root directory) and Apache 2.0 License
+# (found in the LICENSE.Apache file in the root directory).
 #
 # Check for some simple mistakes in public headers (on the command line)
 # that should prevent commit or push

build_tools/dependencies_platform010.sh

@@ -1,21 +1,20 @@
 # Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
 # The file is generated using update_dependencies.sh.
-GCC_BASE=/mnt/gvfs/third-party2/gcc/62de5a92e5f23c661c3d4b9f322e04eb14e7a5bd/11.x/centos8-native/886b5eb
-CLANG_BASE=/mnt/gvfs/third-party2/llvm-fb/1f6edd1ff15c99c861afc8f3cd69054cd974dd64/15/platform010/72a2ff8
-LIBGCC_BASE=/mnt/gvfs/third-party2/libgcc/d1129753c8361ac8e9453c0f4291337a4507ebe6/11.x/platform010/5684a5a
-GLIBC_BASE=/mnt/gvfs/third-party2/glibc/fed6e93d87571fb162734c86636119d45a398963/2.34/platform010/f259413
-SNAPPY_BASE=/mnt/gvfs/third-party2/snappy/31a346126a1f3b64812c362511cb04cc1bd40855/1.1.8/platform010/76ebdda
-ZLIB_BASE=/mnt/gvfs/third-party2/zlib/0c65c05468b5a38cef1a106a1f526463e120c8dd/1.2.8/platform010/76ebdda
+GCC_BASE=/mnt/gvfs/third-party2/gcc/e9cb2d4550b5a95024125084420342d256b349cd/11.x/centos9-native/3bed279
+CLANG_BASE=/mnt/gvfs/third-party2/llvm-fb/4e46e6967e15412702e51960c69790a78933856f/21/platform010/72a2ff8
+LIBGCC_BASE=/mnt/gvfs/third-party2/libgcc/63c5179edc90fec473e53c1b2f92269c65c39e07/11.x/platform010/5684a5a
+GLIBC_BASE=/mnt/gvfs/third-party2/glibc/b4a6248a76a163bcccb1e4751e247712c535846e/2.34/platform010/f259413
+SNAPPY_BASE=/mnt/gvfs/third-party2/snappy/ef895164499fdfedb9d9b9c85520de10aa8ff9ca/1.1.8/platform010/76ebdda
+ZLIB_BASE=/mnt/gvfs/third-party2/zlib/643dbf0136f869562485cfcb2953f5cab60447ef/1.2.8/platform010/76ebdda
 BZIP2_BASE=/mnt/gvfs/third-party2/bzip2/09703139cfc376bd8a82642385a0e97726b28287/1.0.6/platform010/76ebdda
-LZ4_BASE=/mnt/gvfs/third-party2/lz4/ff23d17b932725cc1734a14896a8b67c518ba169/1.9.4/platform010/76ebdda
-ZSTD_BASE=/mnt/gvfs/third-party2/zstd/576397d8b1d9cea7306ad1e454d5e55caaa2ff1c/1.4.x/platform010/64091f4
-GFLAGS_BASE=/mnt/gvfs/third-party2/gflags/fecac07861cb829f5e60dbeff0503d3272db73c0/2.2.0/platform010/76ebdda
-JEMALLOC_BASE=/mnt/gvfs/third-party2/jemalloc/0bb3f5756788ce26e2e16a1cb2f2af2c59b51abe/master/platform010/f57cc4a
+LZ4_BASE=/mnt/gvfs/third-party2/lz4/823bf8e000654abef784311296ffc8166d9d3986/1.9.4/platform010/76ebdda
+ZSTD_BASE=/mnt/gvfs/third-party2/zstd/0cda78647b5712bec7ec6d9ba2f0dcdabfb6f44f/1.4.x/platform010/64091f4
+GFLAGS_BASE=/mnt/gvfs/third-party2/gflags/a02e0de00945729ad553c187c64e86b57f45a33b/2.2.0/platform010/76ebdda
+JEMALLOC_BASE=/mnt/gvfs/third-party2/jemalloc/a1a50742bec7195bfafb594e174ea45dc197a420/master/platform010/779a252
 NUMA_BASE=/mnt/gvfs/third-party2/numa/5b602edd46fda54cdd7ea45f77dbe4061206e174/2.0.11/platform010/76ebdda
-LIBUNWIND_BASE=/mnt/gvfs/third-party2/libunwind/97cac22a149c2e202917e05d44e87e516b68216f/1.4/platform010/5074a48
-TBB_BASE=/mnt/gvfs/third-party2/tbb/53953ebc4e3eda85ad6fc3e429ba146035e97b90/2018_U5/platform010/76ebdda
+LIBUNWIND_BASE=/mnt/gvfs/third-party2/libunwind/f3cce2e1a79ac8d64e102886603a6997d6e88640/1.8/platform010/76ebdda
 LIBURING_BASE=/mnt/gvfs/third-party2/liburing/a98e2d137007e3ebf7f33bd6f99c2c56bdaf8488/20210212/platform010/76ebdda
 BENCHMARK_BASE=/mnt/gvfs/third-party2/benchmark/780c7a0f9cf0967961e69ad08e61cddd85d61821/trunk/platform010/76ebdda
-KERNEL_HEADERS_BASE=/mnt/gvfs/third-party2/kernel-headers/624a2f8f6c93c3c1df8aa4a6255d8202631a6c80/fb/platform010/da39a3e
-BINUTILS_BASE=/mnt/gvfs/third-party2/binutils/39579e8603b48b3540f8b0633f43adf29acccb8b/2.37/centos8-native/da39a3e
-VALGRIND_BASE=/mnt/gvfs/third-party2/valgrind/cd9cc656d49ecb53797ce4d055e49fde29fd57ff/3.19.0/platform010/76ebdda
+KERNEL_HEADERS_BASE=/mnt/gvfs/third-party2/kernel-headers/0883cccda758e68b47e4c04e4fa01142e1f60f32/fb/platform010/da39a3e
+BINUTILS_BASE=/mnt/gvfs/third-party2/binutils/1d5d19135a406a941d3a441eb2bf506bb9d19210/2.43/centos9-native/be7139e
+VALGRIND_BASE=/mnt/gvfs/third-party2/valgrind/3e2b7ccc315c9a0f2b678ce6fd774f4970f9bf3b/3.22.0/platform010/76ebdda

build_tools/fbcode_config.sh

@@ -87,15 +87,6 @@ if test -z $PIC_BUILD; then
   LIBUNWIND="$LIBUNWIND_BASE/lib/libunwind.a"
 fi
 
-# location of TBB
-TBB_INCLUDE=" -isystem $TBB_BASE/include/"
-if test -z $PIC_BUILD; then
-  TBB_LIBS="$TBB_BASE/lib/libtbb.a"
-else
-  TBB_LIBS="$TBB_BASE/lib/libtbb_pic.a"
-fi
-CFLAGS+=" -DTBB"
-
 test "$USE_SSE" || USE_SSE=1
 export USE_SSE
 test "$PORTABLE" || PORTABLE=1
@@ -104,7 +95,7 @@ export PORTABLE
 BINUTILS="$BINUTILS_BASE/bin"
 AR="$BINUTILS/ar"
 
-DEPS_INCLUDE="$SNAPPY_INCLUDE $ZLIB_INCLUDE $BZIP_INCLUDE $LZ4_INCLUDE $ZSTD_INCLUDE $GFLAGS_INCLUDE $NUMA_INCLUDE $TBB_INCLUDE"
+DEPS_INCLUDE="$SNAPPY_INCLUDE $ZLIB_INCLUDE $BZIP_INCLUDE $LZ4_INCLUDE $ZSTD_INCLUDE $GFLAGS_INCLUDE $NUMA_INCLUDE"
 
 STDLIBS="-L $GCC_BASE/lib64"
 
@@ -150,17 +141,17 @@ CFLAGS+=" $DEPS_INCLUDE"
 CFLAGS+=" -DROCKSDB_PLATFORM_POSIX -DROCKSDB_LIB_IO_POSIX -DROCKSDB_FALLOCATE_PRESENT -DROCKSDB_MALLOC_USABLE_SIZE -DROCKSDB_RANGESYNC_PRESENT -DROCKSDB_SCHED_GETCPU_PRESENT"
 CXXFLAGS+=" $CFLAGS"
 
-EXEC_LDFLAGS=" $SNAPPY_LIBS $ZLIB_LIBS $BZIP_LIBS $LZ4_LIBS $ZSTD_LIBS $GFLAGS_LIBS $NUMA_LIB $TBB_LIBS"
+EXEC_LDFLAGS=" $SNAPPY_LIBS $ZLIB_LIBS $BZIP_LIBS $LZ4_LIBS $ZSTD_LIBS $GFLAGS_LIBS $NUMA_LIB"
 EXEC_LDFLAGS+=" -B$BINUTILS/gold"
 EXEC_LDFLAGS+=" -Wl,--dynamic-linker,/usr/local/fbcode/gcc-5-glibc-2.23/lib/ld.so"
 EXEC_LDFLAGS+=" $LIBUNWIND"
 EXEC_LDFLAGS+=" -Wl,-rpath=/usr/local/fbcode/gcc-5-glibc-2.23/lib"
-# required by libtbb
+# dynamic loading
 EXEC_LDFLAGS+=" -ldl"
 
 PLATFORM_LDFLAGS="$LIBGCC_LIBS $GLIBC_LIBS $STDLIBS -lgcc -lstdc++"
 
-EXEC_LDFLAGS_SHARED="$SNAPPY_LIBS $ZLIB_LIBS $BZIP_LIBS $LZ4_LIBS $ZSTD_LIBS $GFLAGS_LIBS $TBB_LIBS"
+EXEC_LDFLAGS_SHARED="$SNAPPY_LIBS $ZLIB_LIBS $BZIP_LIBS $LZ4_LIBS $ZSTD_LIBS $GFLAGS_LIBS"
 
 VALGRIND_VER="$VALGRIND_BASE/bin/"
 

build_tools/fbcode_config_platform010.sh

@@ -82,11 +82,6 @@ CFLAGS+=" -DNUMA"
 # location of libunwind
 LIBUNWIND="$LIBUNWIND_BASE/lib/libunwind${MAYBE_PIC}.a"
 
-# location of TBB
-TBB_INCLUDE=" -isystem $TBB_BASE/include/"
-TBB_LIBS="$TBB_BASE/lib/libtbb${MAYBE_PIC}.a"
-CFLAGS+=" -DTBB"
-
 # location of LIBURING
 LIBURING_INCLUDE=" -isystem $LIBURING_BASE/include/"
 LIBURING_LIBS="$LIBURING_BASE/lib/liburing${MAYBE_PIC}.a"
@@ -101,7 +96,7 @@ BINUTILS="$BINUTILS_BASE/bin"
 AR="$BINUTILS/ar"
 AS="$BINUTILS/as"
 
-DEPS_INCLUDE="$SNAPPY_INCLUDE $ZLIB_INCLUDE $BZIP_INCLUDE $LZ4_INCLUDE $ZSTD_INCLUDE $GFLAGS_INCLUDE $NUMA_INCLUDE $TBB_INCLUDE $LIBURING_INCLUDE $BENCHMARK_INCLUDE"
+DEPS_INCLUDE="$SNAPPY_INCLUDE $ZLIB_INCLUDE $BZIP_INCLUDE $LZ4_INCLUDE $ZSTD_INCLUDE $GFLAGS_INCLUDE $NUMA_INCLUDE $LIBURING_INCLUDE $BENCHMARK_INCLUDE"
 
 STDLIBS="-L $GCC_BASE/lib64"
 
@@ -109,6 +104,10 @@ CLANG_BIN="$CLANG_BASE/bin"
 CLANG_LIB="$CLANG_BASE/lib"
 CLANG_SRC="$CLANG_BASE/../../src"
 
+# A clang++ is always available in the fbcode toolchain regardless of whether
+# the build itself uses gcc or clang. CLANG_CXX exposes it for tooling (e.g. the
+# C API codegen, which parses C++ ASTs) that needs clang even in a gcc build.
+CLANG_CXX="$CLANG_BIN/clang++"
 CLANG_ANALYZER="$CLANG_BIN/clang++"
 CLANG_SCAN_BUILD="$CLANG_BIN/scan-build"
 
@@ -157,19 +156,19 @@ CFLAGS+=" $DEPS_INCLUDE"
 CFLAGS+=" -DROCKSDB_PLATFORM_POSIX -DROCKSDB_LIB_IO_POSIX -DROCKSDB_FALLOCATE_PRESENT -DROCKSDB_MALLOC_USABLE_SIZE -DROCKSDB_RANGESYNC_PRESENT -DROCKSDB_SCHED_GETCPU_PRESENT -DROCKSDB_IOURING_PRESENT"
 CXXFLAGS+=" $CFLAGS"
 
-EXEC_LDFLAGS=" $SNAPPY_LIBS $ZLIB_LIBS $BZIP_LIBS $LZ4_LIBS $ZSTD_LIBS $GFLAGS_LIBS $NUMA_LIB $TBB_LIBS $LIBURING_LIBS $BENCHMARK_LIBS"
+EXEC_LDFLAGS=" $SNAPPY_LIBS $ZLIB_LIBS $BZIP_LIBS $LZ4_LIBS $ZSTD_LIBS $GFLAGS_LIBS $NUMA_LIB $LIBURING_LIBS $BENCHMARK_LIBS"
 EXEC_LDFLAGS+=" -Wl,--dynamic-linker,/usr/local/fbcode/platform010/lib/ld.so"
 EXEC_LDFLAGS+=" $LIBUNWIND"
 EXEC_LDFLAGS+=" -Wl,-rpath=/usr/local/fbcode/platform010/lib"
 EXEC_LDFLAGS+=" -Wl,-rpath=$GCC_BASE/lib64"
-# required by libtbb
+# dynamic loading
 EXEC_LDFLAGS+=" -ldl"
 
 PLATFORM_LDFLAGS="$LIBGCC_LIBS $GLIBC_LIBS $STDLIBS -lgcc -lstdc++"
 PLATFORM_LDFLAGS+=" -B$BINUTILS"
 
-EXEC_LDFLAGS_SHARED="$SNAPPY_LIBS $ZLIB_LIBS $BZIP_LIBS $LZ4_LIBS $ZSTD_LIBS $GFLAGS_LIBS $TBB_LIBS $LIBURING_LIBS $BENCHMARK_LIBS"
+EXEC_LDFLAGS_SHARED="$SNAPPY_LIBS $ZLIB_LIBS $BZIP_LIBS $LZ4_LIBS $ZSTD_LIBS $GFLAGS_LIBS $LIBURING_LIBS $BENCHMARK_LIBS"
 
 VALGRIND_VER="$VALGRIND_BASE/bin/"
 
-export CC CXX AR AS CFLAGS CXXFLAGS EXEC_LDFLAGS EXEC_LDFLAGS_SHARED VALGRIND_VER JEMALLOC_LIB JEMALLOC_INCLUDE CLANG_ANALYZER CLANG_SCAN_BUILD
+export CC CXX AR AS CFLAGS CXXFLAGS EXEC_LDFLAGS EXEC_LDFLAGS_SHARED VALGRIND_VER JEMALLOC_LIB JEMALLOC_INCLUDE CLANG_CXX CLANG_ANALYZER CLANG_SCAN_BUILD

build_tools/gtest-parallel

@@ -0,0 +1,18 @@
+#!/usr/bin/env python3
+# Copyright 2017 Google Inc. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+import gtest_parallel
+import sys
+
+sys.exit(gtest_parallel.main())

build_tools/gtest_parallel.py

@@ -0,0 +1,959 @@
+#  Copyright (c) Meta Platforms, Inc. and affiliates.
+#  This source code is licensed under both the GPLv2 (found in the COPYING file in the root directory)
+#  and the Apache 2.0 License (found in the LICENSE.Apache file in the root directory).
+
+# Copyright 2013 Google Inc. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+import errno
+from functools import total_ordering
+import gzip
+import io
+import json
+import multiprocessing
+import optparse
+import os
+import re
+import shutil
+import signal
+import subprocess
+import sys
+import tempfile
+import threading
+import time
+
+if sys.version_info.major >= 3:
+  long = int
+  import _pickle as cPickle
+  import _thread as thread
+else:
+  import cPickle
+  import thread
+
+from pickle import HIGHEST_PROTOCOL as PICKLE_HIGHEST_PROTOCOL
+
+if sys.platform == 'win32':
+  import msvcrt
+else:
+  import fcntl
+
+
+# An object that catches SIGINT sent to the Python process and notices
+# if processes passed to wait() die by SIGINT (we need to look for
+# both of those cases, because pressing Ctrl+C can result in either
+# the main process or one of the subprocesses getting the signal).
+#
+# Before a SIGINT is seen, wait(p) will simply call p.wait() and
+# return the result. Once a SIGINT has been seen (in the main process
+# or a subprocess, including the one the current call is waiting for),
+# wait(p) will call p.terminate() and raise ProcessWasInterrupted.
+class SigintHandler(object):
+  class ProcessWasInterrupted(Exception):
+    pass
+
+  sigint_returncodes = {
+      -signal.SIGINT,  # Unix
+      -1073741510,  # Windows
+  }
+
+  def __init__(self):
+    self.__lock = threading.Lock()
+    self.__processes = set()
+    self.__got_sigint = False
+    signal.signal(signal.SIGINT, lambda signal_num, frame: self.interrupt())
+
+  def __on_sigint(self):
+    self.__got_sigint = True
+    while self.__processes:
+      try:
+        self.__processes.pop().terminate()
+      except OSError:
+        pass
+
+  def interrupt(self):
+    with self.__lock:
+      self.__on_sigint()
+
+  def got_sigint(self):
+    with self.__lock:
+      return self.__got_sigint
+
+  def wait(self, p, timeout_per_test):
+    with self.__lock:
+      if self.__got_sigint:
+        p.terminate()
+      self.__processes.add(p)
+    try:
+      code = p.wait(timeout_per_test)
+    except subprocess.TimeoutExpired :
+      p.terminate()
+      self.__processes.remove(p)
+      code = -errno.ETIME
+    with self.__lock:
+      self.__processes.discard(p)
+      if code in self.sigint_returncodes:
+        self.__on_sigint()
+      if self.__got_sigint:
+        raise self.ProcessWasInterrupted
+    return code
+
+
+sigint_handler = SigintHandler()
+
+
+# Return the width of the terminal, or None if it couldn't be
+# determined (e.g. because we're not being run interactively).
+def term_width(out):
+  if not out.isatty():
+    return None
+  try:
+    p = subprocess.Popen(["stty", "size"],
+                         stdout=subprocess.PIPE,
+                         stderr=subprocess.PIPE)
+    (out, err) = p.communicate()
+    if p.returncode != 0 or err:
+      return None
+    return int(out.split()[1])
+  except (IndexError, OSError, ValueError):
+    return None
+
+
+# Output transient and permanent lines of text. If several transient
+# lines are written in sequence, the new will overwrite the old. We
+# use this to ensure that lots of unimportant info (tests passing)
+# won't drown out important info (tests failing).
+class Outputter(object):
+  def __init__(self, out_file):
+    self.__out_file = out_file
+    self.__previous_line_was_transient = False
+    self.__width = term_width(out_file)  # Line width, or None if not a tty.
+
+  def transient_line(self, msg):
+    if self.__width is None:
+      self.__out_file.write(msg + "\n")
+      self.__out_file.flush()
+    else:
+      self.__out_file.write("\r" + msg[:self.__width].ljust(self.__width))
+      self.__previous_line_was_transient = True
+
+  def flush_transient_output(self):
+    if self.__previous_line_was_transient:
+      self.__out_file.write("\n")
+      self.__previous_line_was_transient = False
+
+  def permanent_line(self, msg):
+    self.flush_transient_output()
+    self.__out_file.write(msg + "\n")
+    if self.__width is None:
+      self.__out_file.flush()
+
+
+def get_save_file_path():
+  """Return path to file for saving transient data."""
+  if sys.platform == 'win32':
+    default_cache_path = os.path.join(os.path.expanduser('~'), 'AppData',
+                                      'Local')
+    cache_path = os.environ.get('LOCALAPPDATA', default_cache_path)
+  else:
+    # We don't use xdg module since it's not a standard.
+    default_cache_path = os.path.join(os.path.expanduser('~'), '.cache')
+    cache_path = os.environ.get('XDG_CACHE_HOME', default_cache_path)
+
+  if os.path.isdir(cache_path):
+    return os.path.join(cache_path, 'gtest-parallel')
+  else:
+    sys.stderr.write('Directory {} does not exist'.format(cache_path))
+    return os.path.join(os.path.expanduser('~'), '.gtest-parallel-times')
+
+
+@total_ordering
+class Task(object):
+  """Stores information about a task (single execution of a test).
+
+  This class stores information about the test to be executed (gtest binary and
+  test name), and its result (log file, exit code and runtime).
+  Each task is uniquely identified by the gtest binary, the test name and an
+  execution number that increases each time the test is executed.
+  Additionaly we store the last execution time, so that next time the test is
+  executed, the slowest tests are run first.
+  """
+
+  def __init__(self, test_binary, test_name, test_command, execution_number,
+               last_execution_time, output_dir):
+    self.test_name = test_name
+    self.output_dir = output_dir
+    self.test_binary = test_binary
+    self.test_command = test_command
+    self.execution_number = execution_number
+    self.last_execution_time = last_execution_time
+
+    self.exit_code = None
+    self.runtime_ms = None
+
+    self.test_id = (test_binary, test_name)
+    self.task_id = (test_binary, test_name, self.execution_number)
+
+    self.log_file = Task._logname(self.output_dir, self.test_binary, test_name,
+                                  self.execution_number)
+
+  def __sorting_key(self):
+    # Unseen or failing tests (both missing execution time) take precedence over
+    # execution time. Tests are greater (seen as slower) when missing times so
+    # that they are executed first.
+    return (1 if self.last_execution_time is None else 0,
+            self.last_execution_time)
+
+  def __eq__(self, other):
+    return self.__sorting_key() == other.__sorting_key()
+
+  def __ne__(self, other):
+    return not (self == other)
+
+  def __lt__(self, other):
+    return self.__sorting_key() < other.__sorting_key()
+
+  @staticmethod
+  def _normalize(string):
+    return re.sub('[^A-Za-z0-9]', '_', string)
+
+  @staticmethod
+  def _logname(output_dir, test_binary, test_name, execution_number):
+    # Store logs to temporary files if there is no output_dir.
+    if output_dir is None:
+      (log_handle, log_name) = tempfile.mkstemp(prefix='gtest_parallel_',
+                                                suffix=".log")
+      os.close(log_handle)
+      return log_name
+
+    log_name = '%s-%s-%d.log' % (Task._normalize(os.path.basename(test_binary)),
+                                 Task._normalize(test_name), execution_number)
+
+    return os.path.join(output_dir, log_name)
+
+  def run(self, timeout_per_test):
+    begin = time.time()
+    with open(self.log_file, 'w') as log:
+      task = subprocess.Popen(self.test_command, stdout=log, stderr=log)
+      try:
+        self.exit_code = sigint_handler.wait(task, timeout_per_test)
+      except sigint_handler.ProcessWasInterrupted:
+        thread.exit()
+    self.runtime_ms = int(1000 * (time.time() - begin))
+    self.last_execution_time = None if self.exit_code else self.runtime_ms
+
+
+class TaskManager(object):
+  """Executes the tasks and stores the passed, failed and interrupted tasks.
+
+  When a task is run, this class keeps track if it passed, failed or was
+  interrupted. After a task finishes it calls the relevant functions of the
+  Logger, TestResults and TestTimes classes, and in case of failure, retries the
+  test as specified by the --retry_failed flag.
+  """
+
+  def __init__(self, times, logger, test_results, task_factory, times_to_retry,
+               initial_execution_number):
+    self.times = times
+    self.logger = logger
+    self.test_results = test_results
+    self.task_factory = task_factory
+    self.times_to_retry = times_to_retry
+    self.initial_execution_number = initial_execution_number
+
+    self.global_exit_code = 0
+
+    self.passed = []
+    self.failed = []
+    self.started = {}
+    self.timed_out = []
+    self.execution_number = {}
+
+    self.lock = threading.Lock()
+
+  def __get_next_execution_number(self, test_id):
+    with self.lock:
+      next_execution_number = self.execution_number.setdefault(
+          test_id, self.initial_execution_number)
+      self.execution_number[test_id] += 1
+    return next_execution_number
+
+  def __register_start(self, task):
+    with self.lock:
+      self.started[task.task_id] = task
+
+  def register_exit(self, task):
+    self.logger.log_exit(task)
+    self.times.record_test_time(task.test_binary, task.test_name,
+                                task.last_execution_time)
+    if self.test_results:
+      self.test_results.log(task.test_name, task.runtime_ms / 1000.0,
+                            task.exit_code)
+
+    with self.lock:
+      self.started.pop(task.task_id)
+      if task.exit_code == 0:
+        self.passed.append(task)
+      elif task.exit_code == -errno.ETIME:
+        self.timed_out.append(task)
+      else:
+        self.failed.append(task)
+
+  def run_task(self, task, timeout_per_test):
+    for try_number in range(self.times_to_retry + 1):
+      self.__register_start(task)
+      task.run(timeout_per_test)
+      self.register_exit(task)
+
+      if task.exit_code == 0:
+        break
+
+      if try_number < self.times_to_retry:
+        execution_number = self.__get_next_execution_number(task.test_id)
+        # We need create a new Task instance. Each task represents a single test
+        # execution, with its own runtime, exit code and log file.
+        task = self.task_factory(task.test_binary, task.test_name,
+                                 task.test_command, execution_number,
+                                 task.last_execution_time, task.output_dir)
+
+    with self.lock:
+      if task.exit_code != 0:
+        self.global_exit_code = task.exit_code
+
+
+class FilterFormat(object):
+  def __init__(self, output_dir):
+    if sys.stdout.isatty():
+      # stdout needs to be unbuffered since the output is interactive.
+      if isinstance(sys.stdout, io.TextIOWrapper):
+        # workaround for https://bugs.python.org/issue17404
+        sys.stdout = io.TextIOWrapper(sys.stdout.detach(),
+                                      line_buffering=True,
+                                      write_through=True,
+                                      newline='\n')
+      else:
+        sys.stdout = os.fdopen(sys.stdout.fileno(), 'w', 0)
+
+    self.output_dir = output_dir
+
+    self.total_tasks = 0
+    self.finished_tasks = 0
+    self.out = Outputter(sys.stdout)
+    self.stdout_lock = threading.Lock()
+
+  def move_to(self, destination_dir, tasks):
+    if self.output_dir is None:
+      return
+
+    destination_dir = os.path.join(self.output_dir, destination_dir)
+    os.makedirs(destination_dir)
+    for task in tasks:
+      shutil.move(task.log_file, destination_dir)
+
+  def print_tests(self, message, tasks, print_try_number, print_test_command):
+    self.out.permanent_line("%s (%s/%s):" %
+                            (message, len(tasks), self.total_tasks))
+    for task in sorted(tasks):
+      runtime_ms = 'Interrupted'
+      if task.runtime_ms is not None:
+        runtime_ms = '%d ms' % task.runtime_ms
+      if print_test_command:
+        try:
+          cmd_str = " ".join(task.test_command)
+        except TypeError:
+          cmd_str = task.test_command
+        self.out.permanent_line(
+            "%11s: %s%s" %
+            (runtime_ms, cmd_str,
+            (" (try #%d)" % task.execution_number) if print_try_number else ""))
+      else:
+        self.out.permanent_line(
+            "%11s: %s %s%s" %
+            (runtime_ms, task.test_binary, task.test_name,
+            (" (try #%d)" % task.execution_number) if print_try_number else ""))
+
+  def log_exit(self, task):
+    with self.stdout_lock:
+      self.finished_tasks += 1
+      self.out.transient_line("[%d/%d] %s (%d ms)" %
+                              (self.finished_tasks, self.total_tasks,
+                               task.test_name, task.runtime_ms))
+      if task.exit_code != 0:
+        signal_name = None
+        if task.exit_code < 0:
+          try:
+            signal_name = signal.Signals(-task.exit_code).name
+          except ValueError:
+            pass
+
+        with open(task.log_file) as f:
+          for line in f.readlines():
+            self.out.permanent_line(line.rstrip())
+        if task.exit_code is None:
+          self.out.permanent_line("[%d/%d] %s aborted after %d ms" %
+                                  (self.finished_tasks, self.total_tasks,
+                                   task.test_name, task.runtime_ms))
+        elif task.exit_code == -errno.ETIME:
+          self.out.permanent_line(
+            "\033[31m[  TIMEOUT ]\033[0m %s timed out after %d s"
+            % (task.test_name, task.runtime_ms/1000))
+        elif signal_name is not None:
+          self.out.permanent_line(
+              "[%d/%d] %s killed by signal %s (%d ms)" %
+              (self.finished_tasks, self.total_tasks, task.test_name,
+               signal_name, task.runtime_ms))
+        else:
+          self.out.permanent_line(
+              "[%d/%d] %s returned with exit code %d (%d ms)" %
+              (self.finished_tasks, self.total_tasks, task.test_name,
+               task.exit_code, task.runtime_ms))
+
+    if self.output_dir is None:
+      # Try to remove the file 100 times (sleeping for 0.1 second in between).
+      # This is a workaround for a process handle seemingly holding on to the
+      # file for too long inside os.subprocess. This workaround is in place
+      # until we figure out a minimal repro to report upstream (or a better
+      # suspect) to prevent os.remove exceptions.
+      num_tries = 100
+      for i in range(num_tries):
+        try:
+          os.remove(task.log_file)
+        except OSError as e:
+          if e.errno is not errno.ENOENT:
+            if i is num_tries - 1:
+              self.out.permanent_line('Could not remove temporary log file: ' +
+                                      str(e))
+            else:
+              time.sleep(0.1)
+            continue
+        break
+
+  def log_tasks(self, total_tasks):
+    self.total_tasks += total_tasks
+    self.out.transient_line("[0/%d] Running tests..." % self.total_tasks)
+
+  def summarize(self, passed_tasks, failed_tasks, interrupted_tasks):
+    stats = {}
+
+    def add_stats(stats, task, idx):
+      task_key = (task.test_binary, task.test_name)
+      if not task_key in stats:
+        # (passed, failed, interrupted) task_key is added as tie breaker to get
+        # alphabetic sorting on equally-stable tests
+        stats[task_key] = [0, 0, 0, task_key]
+      stats[task_key][idx] += 1
+
+    for task in passed_tasks:
+      add_stats(stats, task, 0)
+    for task in failed_tasks:
+      add_stats(stats, task, 1)
+    for task in interrupted_tasks:
+      add_stats(stats, task, 2)
+
+    self.out.permanent_line("SUMMARY:")
+    for task_key in sorted(stats, key=stats.__getitem__):
+      (num_passed, num_failed, num_interrupted, _) = stats[task_key]
+      (test_binary, task_name) = task_key
+      total_runs = num_passed + num_failed + num_interrupted
+      if num_passed == total_runs:
+        continue
+      self.out.permanent_line("  %s %s passed %d / %d times%s." %
+                              (test_binary, task_name, num_passed, total_runs,
+                               "" if num_interrupted == 0 else
+                               (" (%d interrupted)" % num_interrupted)))
+
+  def flush(self):
+    self.out.flush_transient_output()
+
+
+class CollectTestResults(object):
+  def __init__(self, json_dump_filepath):
+    self.test_results_lock = threading.Lock()
+    self.json_dump_file = open(json_dump_filepath, 'w')
+    self.test_results = {
+        "interrupted": False,
+        "path_delimiter": ".",
+        # Third version of the file format. See the link in the flag description
+        # for details.
+        "version": 3,
+        "seconds_since_epoch": int(time.time()),
+        "num_failures_by_type": {
+            "PASS": 0,
+            "FAIL": 0,
+            "TIMEOUT": 0,
+        },
+        "tests": {},
+    }
+
+  def log(self, test, runtime_seconds, exit_code):
+    if exit_code is None:
+      actual_result = "TIMEOUT"
+    elif exit_code == 0:
+      actual_result = "PASS"
+    else:
+      actual_result = "FAIL"
+    with self.test_results_lock:
+      self.test_results['num_failures_by_type'][actual_result] += 1
+      results = self.test_results['tests']
+      for name in test.split('.'):
+        results = results.setdefault(name, {})
+
+      if results:
+        results['actual'] += ' ' + actual_result
+        results['times'].append(runtime_seconds)
+      else:  # This is the first invocation of the test
+        results['actual'] = actual_result
+        results['times'] = [runtime_seconds]
+        results['time'] = runtime_seconds
+        results['expected'] = 'PASS'
+
+  def dump_to_file_and_close(self):
+    json.dump(self.test_results, self.json_dump_file)
+    self.json_dump_file.close()
+
+
+# Record of test runtimes. Has built-in locking.
+class TestTimes(object):
+  class LockedFile(object):
+    def __init__(self, filename, mode):
+      self._filename = filename
+      self._mode = mode
+      self._fo = None
+
+    def __enter__(self):
+      self._fo = open(self._filename, self._mode)
+
+      # Regardless of opening mode we always seek to the beginning of file.
+      # This simplifies code working with LockedFile and also ensures that
+      # we lock (and unlock below) always the same region in file on win32.
+      self._fo.seek(0)
+
+      try:
+        if sys.platform == 'win32':
+          # We are locking here fixed location in file to use it as
+          # an exclusive lock on entire file.
+          msvcrt.locking(self._fo.fileno(), msvcrt.LK_LOCK, 1)
+        else:
+          fcntl.flock(self._fo.fileno(), fcntl.LOCK_EX)
+      except IOError:
+        self._fo.close()
+        raise
+
+      return self._fo
+
+    def __exit__(self, exc_type, exc_value, traceback):
+      # Flush any buffered data to disk. This is needed to prevent race
+      # condition which happens from the moment of releasing file lock
+      # till closing the file.
+      self._fo.flush()
+
+      try:
+        if sys.platform == 'win32':
+          self._fo.seek(0)
+          msvcrt.locking(self._fo.fileno(), msvcrt.LK_UNLCK, 1)
+        else:
+          fcntl.flock(self._fo.fileno(), fcntl.LOCK_UN)
+      finally:
+        self._fo.close()
+
+      return exc_value is None
+
+  def __init__(self, save_file):
+    "Create new object seeded with saved test times from the given file."
+    self.__times = {}  # (test binary, test name) -> runtime in ms
+
+    # Protects calls to record_test_time(); other calls are not
+    # expected to be made concurrently.
+    self.__lock = threading.Lock()
+
+    try:
+      with TestTimes.LockedFile(save_file, 'rb') as fd:
+        times = TestTimes.__read_test_times_file(fd)
+    except IOError:
+      # We couldn't obtain the lock.
+      return
+
+    # Discard saved times if the format isn't right.
+    if type(times) is not dict:
+      return
+    for ((test_binary, test_name), runtime) in times.items():
+      if (type(test_binary) is not str or type(test_name) is not str
+          or type(runtime) not in {int, long, type(None)}):
+        return
+
+    self.__times = times
+
+  def get_test_time(self, binary, testname):
+    """Return the last duration for the given test as an integer number of
+    milliseconds, or None if the test failed or if there's no record for it."""
+    return self.__times.get((binary, testname), None)
+
+  def record_test_time(self, binary, testname, runtime_ms):
+    """Record that the given test ran in the specified number of
+    milliseconds. If the test failed, runtime_ms should be None."""
+    with self.__lock:
+      self.__times[(binary, testname)] = runtime_ms
+
+  def write_to_file(self, save_file):
+    "Write all the times to file."
+    try:
+      with TestTimes.LockedFile(save_file, 'a+b') as fd:
+        times = TestTimes.__read_test_times_file(fd)
+
+        if times is None:
+          times = self.__times
+        else:
+          times.update(self.__times)
+
+        # We erase data from file while still holding a lock to it. This
+        # way reading old test times and appending new ones are atomic
+        # for external viewer.
+        fd.seek(0)
+        fd.truncate()
+        with gzip.GzipFile(fileobj=fd, mode='wb') as gzf:
+          cPickle.dump(times, gzf, PICKLE_HIGHEST_PROTOCOL)
+    except IOError:
+      pass  # ignore errors---saving the times isn't that important
+
+  @staticmethod
+  def __read_test_times_file(fd):
+    try:
+      with gzip.GzipFile(fileobj=fd, mode='rb') as gzf:
+        times = cPickle.load(gzf)
+    except Exception:
+      # File doesn't exist, isn't readable, is malformed---whatever.
+      # Just ignore it.
+      return None
+    else:
+      return times
+
+
+def find_tests(binaries, additional_args, options, times):
+  test_count = 0
+  tasks = []
+  for test_binary in binaries:
+    command = [test_binary] + additional_args
+    if options.gtest_also_run_disabled_tests:
+      command += ['--gtest_also_run_disabled_tests']
+
+    list_command = command + ['--gtest_list_tests']
+    if options.gtest_filter != '':
+      list_command += ['--gtest_filter=' + options.gtest_filter]
+
+    try:
+      test_list = subprocess.check_output(list_command,
+                                          stderr=subprocess.STDOUT)
+    except subprocess.CalledProcessError as e:
+      sys.exit("%s: %s\n%s" % (test_binary, str(e), e.output))
+
+    try:
+      test_list = test_list.split('\n')
+    except TypeError:
+      # subprocess.check_output() returns bytes in python3
+      test_list = test_list.decode(sys.stdout.encoding).split('\n')
+
+    command += ['--gtest_color=' + options.gtest_color]
+
+    test_group = ''
+    for line in test_list:
+      if not line.strip():
+        continue
+      if line[0] != " ":
+        # Remove comments for typed tests and strip whitespace.
+        test_group = line.split('#')[0].strip()
+        continue
+      # Remove comments for parameterized tests and strip whitespace.
+      line = line.split('#')[0].strip()
+      if not line:
+        continue
+
+      test_name = test_group + line
+      if not options.gtest_also_run_disabled_tests and 'DISABLED_' in test_name:
+        continue
+
+      # Skip PRE_ tests which are used by Chromium.
+      if '.PRE_' in test_name:
+        continue
+
+      last_execution_time = times.get_test_time(test_binary, test_name)
+      if options.failed and last_execution_time is not None:
+        continue
+
+      test_command = command + ['--gtest_filter=' + test_name]
+      if (test_count - options.shard_index) % options.shard_count == 0:
+        for execution_number in range(options.repeat):
+          tasks.append(
+              Task(test_binary, test_name, test_command, execution_number + 1,
+                   last_execution_time, options.output_dir))
+
+      test_count += 1
+
+  # Sort the tasks to run the slowest tests first, so that faster ones can be
+  # finished in parallel.
+  return sorted(tasks, reverse=True)
+
+
+def execute_tasks(tasks, pool_size, task_manager, timeout_seconds,
+                  timeout_per_test, serialize_test_cases):
+  class WorkerFn(object):
+    def __init__(self, tasks, running_groups, timeout_per_test):
+      self.tasks = tasks
+      self.running_groups = running_groups
+      self.timeout_per_test = timeout_per_test
+      self.task_lock = threading.Lock()
+
+    def __call__(self):
+      while True:
+        with self.task_lock:
+          for task_id in range(len(self.tasks)):
+            task = self.tasks[task_id]
+
+            if self.running_groups is not None:
+              test_group = task.test_name.split('.')[0]
+              if test_group in self.running_groups:
+                # Try to find other non-running test group.
+                continue
+              else:
+                self.running_groups.add(test_group)
+
+            del self.tasks[task_id]
+            break
+          else:
+            # Either there is no tasks left or number or remaining test
+            # cases (groups) is less than number or running threads.
+            return
+
+        task_manager.run_task(task, self.timeout_per_test)
+
+        if self.running_groups is not None:
+          with self.task_lock:
+            self.running_groups.remove(test_group)
+
+  def start_daemon(func):
+    t = threading.Thread(target=func)
+    t.daemon = True
+    t.start()
+    return t
+
+  timeout = None
+  try:
+    if timeout_seconds:
+      timeout = threading.Timer(timeout_seconds, sigint_handler.interrupt)
+      timeout.start()
+    running_groups = set() if serialize_test_cases else None
+    worker_fn = WorkerFn(tasks, running_groups, timeout_per_test)
+    workers = [start_daemon(worker_fn) for _ in range(pool_size)]
+    for worker in workers:
+      worker.join()
+  finally:
+    if timeout:
+      timeout.cancel()
+      for task in list(task_manager.started.values()):
+        task.runtime_ms = timeout_seconds * 1000
+        task_manager.register_exit(task)
+
+
+def default_options_parser():
+  parser = optparse.OptionParser(
+      usage='usage: %prog [options] binary [binary ...] -- [additional args]')
+
+  parser.add_option('-d',
+                    '--output_dir',
+                    type='string',
+                    default=None,
+                    help='Output directory for test logs. Logs will be '
+                    'available under gtest-parallel-logs/, so '
+                    '--output_dir=/tmp will results in all logs being '
+                    'available under /tmp/gtest-parallel-logs/.')
+  parser.add_option('-r',
+                    '--repeat',
+                    type='int',
+                    default=1,
+                    help='Number of times to execute all the tests.')
+  parser.add_option('--retry_failed',
+                    type='int',
+                    default=0,
+                    help='Number of times to repeat failed tests.')
+  parser.add_option('--failed',
+                    action='store_true',
+                    default=False,
+                    help='run only failed and new tests')
+  parser.add_option('-w',
+                    '--workers',
+                    type='int',
+                    default=multiprocessing.cpu_count(),
+                    help='number of workers to spawn')
+  parser.add_option('--gtest_color',
+                    type='string',
+                    default='yes',
+                    help='color output')
+  parser.add_option('--gtest_filter',
+                    type='string',
+                    default='',
+                    help='test filter')
+  parser.add_option('--gtest_also_run_disabled_tests',
+                    action='store_true',
+                    default=False,
+                    help='run disabled tests too')
+  parser.add_option(
+      '--print_test_times',
+      action='store_true',
+      default=False,
+      help='list the run time of each test at the end of execution')
+  parser.add_option(
+      '--print_test_command',
+      action='store_true',
+      default=False,
+      help='Print full test command instead of name')
+  parser.add_option('--shard_count',
+                    type='int',
+                    default=1,
+                    help='total number of shards (for sharding test execution '
+                    'between multiple machines)')
+  parser.add_option('--shard_index',
+                    type='int',
+                    default=0,
+                    help='zero-indexed number identifying this shard (for '
+                    'sharding test execution between multiple machines)')
+  parser.add_option(
+      '--dump_json_test_results',
+      type='string',
+      default=None,
+      help='Saves the results of the tests as a JSON machine-'
+      'readable file. The format of the file is specified at '
+      'https://www.chromium.org/developers/the-json-test-results-format')
+  parser.add_option('--timeout',
+                    type='int',
+                    default=None,
+                    help='Interrupt all remaining processes after the given '
+                    'time (in seconds).')
+  parser.add_option('--timeout_per_test',
+                    type='int',
+                    default=None,
+                    help='Interrupt single processes after the given '
+                    'time (in seconds).')
+  parser.add_option('--serialize_test_cases',
+                    action='store_true',
+                    default=False,
+                    help='Do not run tests from the same test '
+                    'case in parallel.')
+  return parser
+
+
+def main():
+  # Remove additional arguments (anything after --).
+  additional_args = []
+
+  for i in range(len(sys.argv)):
+    if sys.argv[i] == '--':
+      additional_args = sys.argv[i + 1:]
+      sys.argv = sys.argv[:i]
+      break
+
+  parser = default_options_parser()
+  (options, binaries) = parser.parse_args()
+
+  if (options.output_dir is not None and not os.path.isdir(options.output_dir)):
+    parser.error('--output_dir value must be an existing directory, '
+                 'current value is "%s"' % options.output_dir)
+
+  # Append gtest-parallel-logs to log output, this is to avoid deleting user
+  # data if an user passes a directory where files are already present. If a
+  # user specifies --output_dir=Docs/, we'll create Docs/gtest-parallel-logs
+  # and clean that directory out on startup, instead of nuking Docs/.
+  if options.output_dir:
+    options.output_dir = os.path.join(options.output_dir, 'gtest-parallel-logs')
+
+  if binaries == []:
+    parser.print_usage()
+    sys.exit(1)
+
+  if options.shard_count < 1:
+    parser.error("Invalid number of shards: %d. Must be at least 1." %
+                 options.shard_count)
+  if not (0 <= options.shard_index < options.shard_count):
+    parser.error("Invalid shard index: %d. Must be between 0 and %d "
+                 "(less than the number of shards)." %
+                 (options.shard_index, options.shard_count - 1))
+
+  # Check that all test binaries have an unique basename. That way we can ensure
+  # the logs are saved to unique files even when two different binaries have
+  # common tests.
+  unique_binaries = set(os.path.basename(binary) for binary in binaries)
+  assert len(unique_binaries) == len(binaries), (
+      "All test binaries must have an unique basename.")
+
+  if options.output_dir:
+    # Remove files from old test runs.
+    if os.path.isdir(options.output_dir):
+      shutil.rmtree(options.output_dir)
+    # Create directory for test log output.
+    try:
+      os.makedirs(options.output_dir)
+    except OSError as e:
+      # Ignore errors if this directory already exists.
+      if e.errno != errno.EEXIST or not os.path.isdir(options.output_dir):
+        raise e
+
+  test_results = None
+  if options.dump_json_test_results is not None:
+    test_results = CollectTestResults(options.dump_json_test_results)
+
+  save_file = get_save_file_path()
+
+  times = TestTimes(save_file)
+  logger = FilterFormat(options.output_dir)
+
+  task_manager = TaskManager(times, logger, test_results, Task,
+                             options.retry_failed, options.repeat + 1)
+
+  tasks = find_tests(binaries, additional_args, options, times)
+  logger.log_tasks(len(tasks))
+  execute_tasks(tasks, options.workers, task_manager, options.timeout,
+                options.timeout_per_test, options.serialize_test_cases)
+
+  print_try_number = options.retry_failed > 0 or options.repeat > 1
+  if task_manager.passed:
+    logger.move_to('passed', task_manager.passed)
+    if options.print_test_times:
+      logger.print_tests('PASSED TESTS', task_manager.passed, print_try_number, options.print_test_command)
+
+  if task_manager.failed:
+    logger.print_tests('FAILED TESTS', task_manager.failed, print_try_number, options.print_test_command)
+    logger.move_to('failed', task_manager.failed)
+
+  if task_manager.timed_out:
+    logger.print_tests('TIMED OUT TESTS', task_manager.timed_out, print_try_number, options.print_test_command)
+    logger.move_to('timed_out', task_manager.timed_out)
+
+  if task_manager.started:
+    logger.print_tests('INTERRUPTED TESTS', task_manager.started.values(),
+                       print_try_number, options.print_test_command)
+    logger.move_to('interrupted', task_manager.started.values())
+
+  if options.repeat > 1 and (task_manager.failed or task_manager.started):
+    logger.summarize(task_manager.passed, task_manager.failed,
+                     task_manager.started.values())
+
+  logger.flush()
+  times.write_to_file(save_file)
+  if test_results:
+    test_results.dump_to_file_and_close()
+
+  if sigint_handler.got_sigint():
+    return -signal.SIGINT
+
+  return task_manager.global_exit_code
+
+
+if __name__ == "__main__":
+  sys.exit(main())

build_tools/rocksptest.sh

@@ -0,0 +1,67 @@
+#!/usr/bin/env bash
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# This source code is licensed under both the GPLv2 (found in the
+# COPYING file in the root directory) and Apache 2.0 License
+# (found in the LICENSE.Apache file in the root directory).
+#
+# Build RocksDB unit test binaries and run them under gtest-parallel,
+# which shards the test cases across CPUs for faster execution.
+#
+# Hardened version of a simple `make <bin> && gtest-parallel ./<bin>` helper:
+#  * AUTO_CLEAN=1 so the Makefile automatically runs a clean when the build
+#    parameters (DEBUG_LEVEL, sanitizers, ASSERT_STATUS_CHECKED, RTTI, etc.)
+#    have changed since the last build, preventing stale/mixed object files.
+#  * Parallel build with -j<NCORES>, computed the same way the Makefile does.
+#  * Uses the gtest-parallel checked in alongside this script (build_tools/),
+#    so it works regardless of PATH.
+#  * `set -euo pipefail` so any failure stops the script.
+#
+# Run from the repository root.
+#
+# Accepts one or more test binaries (gtest-parallel pools all their test cases
+# into one shared worker queue). The leading non-option arguments are treated as
+# binaries; everything from the first option onward is forwarded to
+# gtest-parallel / the test binaries.
+#
+# Example usage:
+#   build_tools/rocksptest.sh db_test
+#   build_tools/rocksptest.sh db_test -r1000 --gtest_filter=*MixedSlowdown*
+#   build_tools/rocksptest.sh db_test env_test db_basic_test
+#   build_tools/rocksptest.sh db_test env_test --gtest_filter=*Foo*
+
+set -euo pipefail
+
+if [ "$#" -lt 1 ]; then
+  echo "usage: $0 <test_binary> [more_test_binaries...] [gtest-parallel/test args...]" >&2
+  echo "example: $0 db_test env_test -r1000 --gtest_filter=*MixedSlowdown*" >&2
+  exit 1
+fi
+
+# First argument must be a test binary, not an option (catches a forgotten name).
+if [ "${1#-}" != "$1" ]; then
+  echo "$0: first argument must be a test binary name, not an option ('$1')" >&2
+  exit 1
+fi
+
+# Collect the leading non-option arguments as test binaries; everything from the
+# first option onward is forwarded to gtest-parallel / the test binaries.
+BINS=()
+while [ "$#" -gt 0 ] && [ "${1#-}" = "$1" ]; do
+  BINS+=("$1")
+  shift
+done
+
+# Paths as gtest-parallel expects them (relative to the current directory).
+BIN_PATHS=()
+for b in "${BINS[@]}"; do
+  BIN_PATHS+=("./$b")
+done
+
+# Directory of this script, so we use the gtest-parallel checked in next to it.
+SCRIPT_DIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
+
+# Compute parallelism the same way the Makefile computes NCORES.
+NCORES=$(nproc 2>/dev/null || sysctl -n hw.ncpu 2>/dev/null || echo 4)
+
+make AUTO_CLEAN=1 -j"$NCORES" "${BINS[@]}"
+"$SCRIPT_DIR/gtest-parallel" "${BIN_PATHS[@]}" "$@"

build_tools/rockstest.sh

@@ -0,0 +1,76 @@
+#!/usr/bin/env bash
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# This source code is licensed under both the GPLv2 (found in the
+# COPYING file in the root directory) and Apache 2.0 License
+# (found in the LICENSE.Apache file in the root directory).
+#
+# Build a single RocksDB unit test binary and run it directly (serially).
+# Only recommended with --gtest_filter=... because of speed. See also
+# build_tools/rocksptest.sh
+#
+# Hardened version of a simple `make <bin> && ./<bin>` helper:
+#  * AUTO_CLEAN=1 so the Makefile automatically runs a clean when the build
+#    parameters (DEBUG_LEVEL, sanitizers, ASSERT_STATUS_CHECKED, RTTI, etc.)
+#    have changed since the last build, preventing stale/mixed object files.
+#  * Parallel build with -j<NCORES>, computed the same way the Makefile does.
+#  * `set -euo pipefail` so any failure stops the script.
+#
+# Run from the repository root.
+#
+# Example usage:
+#   build_tools/rockstest.sh db_test --gtest_filter=*MixedSlowdown*
+#   build_tools/rockstest.sh env_test # Slow to run many tests serially
+#
+# Install mode:
+#   build_tools/rockstest.sh install
+#     Creates ~/bin/rockstest and ~/bin/rocksptest shims that defer to
+#     build_tools/rockstest.sh / rocksptest.sh in whatever directory you run
+#     them from, so you can just type `rockstest db_test` from any rocksdb
+#     source root.
+
+set -euo pipefail
+
+# Install mode: write thin ~/bin shims that defer to the build_tools scripts in
+# the current directory.
+if [ "${1:-}" = "install" ]; then
+  mkdir -p "$HOME/bin"
+  for name in rockstest rocksptest; do
+    dest="$HOME/bin/$name"
+    cat > "$dest" <<EOF
+#!/usr/bin/env bash
+# Auto-generated by 'build_tools/rockstest.sh install'. Defers to
+# build_tools/$name.sh in the current rocksdb source root.
+if [ -x build_tools/$name.sh ]; then
+  exec build_tools/$name.sh "\$@"
+else
+  echo "build_tools/$name.sh not found (Not in a rocksdb source root directory?)" >&2
+  exit 1
+fi
+EOF
+    chmod +x "$dest"
+    echo "Installed $dest"
+  done
+  exit 0
+fi
+
+if [ "$#" -lt 1 ]; then
+  echo "usage: $0 <test_binary> [test args...]" >&2
+  echo "       $0 install   # create ~/bin/rockstest and ~/bin/rocksptest shims" >&2
+  echo "example: $0 db_test --gtest_filter=*MixedSlowdown*" >&2
+  exit 1
+fi
+
+# First argument must be a test binary, not an option (catches a forgotten name).
+if [ "${1#-}" != "$1" ]; then
+  echo "$0: first argument must be a test binary name, not an option ('$1')" >&2
+  exit 1
+fi
+
+BIN=$1
+shift
+
+# Compute parallelism the same way the Makefile computes NCORES.
+NCORES=$(nproc 2>/dev/null || sysctl -n hw.ncpu 2>/dev/null || echo 4)
+
+make AUTO_CLEAN=1 -j"$NCORES" "$BIN"
+./"$BIN" "$@"

build_tools/update_dependencies.sh

@@ -32,7 +32,7 @@ function get_lib_base()
   local lib_platform=$3
 
   local result="$TP2_LATEST/$lib_name/"
-  
+
   # Lib Version
   if [ -z "$lib_version" ] || [ "$lib_version" = "LATEST" ]; then
     # version is not provided, use latest
@@ -40,7 +40,7 @@ function get_lib_base()
   else
     result="$result/$lib_version/"
   fi
-  
+
   # Lib Platform
   if [ -z "$lib_platform" ]; then
     # platform is not provided, use latest gcc
@@ -49,17 +49,17 @@ function get_lib_base()
     echo $lib_platform
     result="$result/$lib_platform/"
   fi
-  
+
   result=`ls -1d $result/*/ | head -n1`
 
   echo Finding link $result
-  
+
   # lib_name => LIB_NAME_BASE
   local __res_var=${lib_name^^}"_BASE"
   __res_var=`echo $__res_var | tr - _`
   # LIB_NAME_BASE=$result
   eval $__res_var=`readlink -f $result`
-  
+
   log_variable $__res_var
 }
 
@@ -75,17 +75,33 @@ touch "$OUTPUT"
 echo "Writing dependencies to $OUTPUT"
 
 # Compilers locations
-GCC_BASE=`readlink -f $TP2_LATEST/gcc/11.x/centos8-native/*/`
-CLANG_BASE=`readlink -f $TP2_LATEST/llvm-fb/15/platform010/*/`
+# GCC is pinned to 11.x because the only newer GCC in third-party2 (13.x) is
+# built for centos9/glibc>=2.35 -- its libgcc_s.so.1 has a hard reference
+# to _dl_find_object@GLIBC_2.35, but platform010 ships glibc 2.34. Bumping
+# GCC requires a platform with glibc >= 2.35.
+GCC_BASE=`readlink -f $TP2_LATEST/gcc/11.x/centos9-native/*/`
+# Clang is pinned to the latest tested major (21). Bump deliberately.
+CLANG_BASE=`readlink -f $TP2_LATEST/llvm-fb/21/platform010/*/`
 
 log_header
 log_variable GCC_BASE
 log_variable CLANG_BASE
 
 # Libraries locations
+# libgcc is pinned to 11.x to match the GCC 11 compiler above (libstdc++/
+# libgcc runtime, ABI, and C++ headers). Bump in lockstep with GCC_BASE.
 get_lib_base libgcc     11.x    platform010
+# glibc 2.34 is the platform010 ABI baseline (ld.so + libc); it is also the
+# only version available in third-party2, and defines the platform -- do
+# not bump independently.
 get_lib_base glibc      2.34    platform010
 get_lib_base snappy     LATEST  platform010
+# zlib is pinned to 1.2.8: it is the latest version with an x86_64
+# platform010 build (1.2.13 / 1.3.1 are centos9 / aarch64 only). At time of
+# writing, LATEST here doesn't work: get_lib_base picks the newest version
+# dir first and only then appends the platform, with no fallback -- so
+# LATEST would resolve to 1.3.1, find no platform010 build, and emit an
+# empty ZLIB_BASE.
 get_lib_base zlib       1.2.8   platform010
 get_lib_base bzip2      LATEST  platform010
 get_lib_base lz4        LATEST  platform010
@@ -94,12 +110,11 @@ get_lib_base gflags     LATEST  platform010
 get_lib_base jemalloc   LATEST  platform010
 get_lib_base numa       LATEST  platform010
 get_lib_base libunwind  LATEST  platform010
-get_lib_base tbb        2018_U5 platform010
 get_lib_base liburing   LATEST  platform010
 get_lib_base benchmark  LATEST  platform010
 
 get_lib_base kernel-headers fb platform010
-get_lib_base binutils   LATEST centos8-native
+get_lib_base binutils   LATEST centos9-native
 get_lib_base valgrind   LATEST platform010
 
 git diff $OUTPUT

c_api_gen/c_generated_block_based_options_subset.cc.inc

@@ -0,0 +1,43 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+// @generated
+// -----------------------------------------------------------------------------
+// Auto-generated by tools/c_api_gen/generate_c_api.py.
+// DO NOT EDIT THIS FILE DIRECTLY.
+// Source: tools/c_api_gen/spec.json
+// -----------------------------------------------------------------------------
+
+// To regenerate this file:
+//   python3 tools/c_api_gen/generate_c_api.py
+//     --spec tools/c_api_gen/spec.json --section 'BlockBasedOptions simple'
+//     --header-out
+//     c_api_gen/c_generated_block_based_options_subset.h.inc
+//     --source-out
+//     c_api_gen/c_generated_block_based_options_subset.cc.inc
+
+/* BlockBasedOptions simple */
+
+void rocksdb_block_based_options_set_data_block_hash_ratio(
+    rocksdb_block_based_table_options_t* options, double v) {
+  options->rep.data_block_hash_table_util_ratio = v;
+}
+
+void rocksdb_block_based_options_set_top_level_index_pinning_tier(
+    rocksdb_block_based_table_options_t* options, int v) {
+  options->rep.metadata_cache_options.top_level_index_pinning =
+      static_cast<ROCKSDB_NAMESPACE::PinningTier>(v);
+}
+
+void rocksdb_block_based_options_set_partition_pinning_tier(
+    rocksdb_block_based_table_options_t* options, int v) {
+  options->rep.metadata_cache_options.partition_pinning =
+      static_cast<ROCKSDB_NAMESPACE::PinningTier>(v);
+}
+
+void rocksdb_block_based_options_set_unpartitioned_pinning_tier(
+    rocksdb_block_based_table_options_t* options, int v) {
+  options->rep.metadata_cache_options.unpartitioned_pinning =
+      static_cast<ROCKSDB_NAMESPACE::PinningTier>(v);
+}

c_api_gen/c_generated_block_based_options_subset.h.inc

@@ -0,0 +1,36 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+// @generated
+// -----------------------------------------------------------------------------
+// Auto-generated by tools/c_api_gen/generate_c_api.py.
+// DO NOT EDIT THIS FILE DIRECTLY.
+// Source: tools/c_api_gen/spec.json
+// -----------------------------------------------------------------------------
+
+// To regenerate this file:
+//   python3 tools/c_api_gen/generate_c_api.py
+//     --spec tools/c_api_gen/spec.json --section 'BlockBasedOptions simple'
+//     --header-out
+//     c_api_gen/c_generated_block_based_options_subset.h.inc
+//     --source-out
+//     c_api_gen/c_generated_block_based_options_subset.cc.inc
+
+/* BlockBasedOptions simple */
+
+extern ROCKSDB_LIBRARY_API void
+rocksdb_block_based_options_set_data_block_hash_ratio(
+    rocksdb_block_based_table_options_t* options, double v);
+
+extern ROCKSDB_LIBRARY_API void
+rocksdb_block_based_options_set_top_level_index_pinning_tier(
+    rocksdb_block_based_table_options_t* options, int v);
+
+extern ROCKSDB_LIBRARY_API void
+rocksdb_block_based_options_set_partition_pinning_tier(
+    rocksdb_block_based_table_options_t* options, int v);
+
+extern ROCKSDB_LIBRARY_API void
+rocksdb_block_based_options_set_unpartitioned_pinning_tier(
+    rocksdb_block_based_table_options_t* options, int v);

c_api_gen/c_generated_cuckoo_options_subset.cc.inc

@@ -0,0 +1,25 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+// @generated
+// -----------------------------------------------------------------------------
+// Auto-generated by tools/c_api_gen/generate_c_api.py.
+// DO NOT EDIT THIS FILE DIRECTLY.
+// Source: tools/c_api_gen/spec.json
+// -----------------------------------------------------------------------------
+
+// To regenerate this file:
+//   python3 tools/c_api_gen/generate_c_api.py
+//     --spec tools/c_api_gen/spec.json --section 'CuckooOptions simple'
+//     --header-out
+//     c_api_gen/c_generated_cuckoo_options_subset.h.inc
+//     --source-out
+//     c_api_gen/c_generated_cuckoo_options_subset.cc.inc
+
+/* CuckooOptions simple */
+
+void rocksdb_cuckoo_options_set_hash_ratio(
+    rocksdb_cuckoo_table_options_t* options, double v) {
+  options->rep.hash_table_ratio = v;
+}

c_api_gen/c_generated_cuckoo_options_subset.h.inc

@@ -0,0 +1,23 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+// @generated
+// -----------------------------------------------------------------------------
+// Auto-generated by tools/c_api_gen/generate_c_api.py.
+// DO NOT EDIT THIS FILE DIRECTLY.
+// Source: tools/c_api_gen/spec.json
+// -----------------------------------------------------------------------------
+
+// To regenerate this file:
+//   python3 tools/c_api_gen/generate_c_api.py
+//     --spec tools/c_api_gen/spec.json --section 'CuckooOptions simple'
+//     --header-out
+//     c_api_gen/c_generated_cuckoo_options_subset.h.inc
+//     --source-out
+//     c_api_gen/c_generated_cuckoo_options_subset.cc.inc
+
+/* CuckooOptions simple */
+
+extern ROCKSDB_LIBRARY_API void rocksdb_cuckoo_options_set_hash_ratio(
+    rocksdb_cuckoo_table_options_t* options, double v);

c_api_gen/c_generated_db_simple_subset.cc.inc

@@ -0,0 +1,193 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+// @generated
+// -----------------------------------------------------------------------------
+// Auto-generated by tools/c_api_gen/generate_c_api.py.
+// DO NOT EDIT THIS FILE DIRECTLY.
+// Source: tools/c_api_gen/spec.json
+// -----------------------------------------------------------------------------
+
+// To regenerate this file:
+//   python3 tools/c_api_gen/generate_c_api.py
+//     --spec tools/c_api_gen/spec.json --section 'DB data operations'
+//     --header-out
+//     c_api_gen/c_generated_db_simple_subset.h.inc
+//     --source-out
+//     c_api_gen/c_generated_db_simple_subset.cc.inc
+
+/* DB data operations */
+
+void rocksdb_put(rocksdb_t* db, const rocksdb_writeoptions_t* options,
+                 const char* key, size_t keylen, const char* val, size_t vallen,
+                 char** errptr) {
+  SaveError(errptr,
+            db->rep->Put(options->rep, Slice(key, keylen), Slice(val, vallen)));
+}
+
+void rocksdb_put_cf(rocksdb_t* db, const rocksdb_writeoptions_t* options,
+                    rocksdb_column_family_handle_t* column_family,
+                    const char* key, size_t keylen, const char* val,
+                    size_t vallen, char** errptr) {
+  SaveError(errptr, db->rep->Put(options->rep, column_family->rep,
+                                 Slice(key, keylen), Slice(val, vallen)));
+}
+
+void rocksdb_delete(rocksdb_t* db, const rocksdb_writeoptions_t* options,
+                    const char* key, size_t keylen, char** errptr) {
+  SaveError(errptr, db->rep->Delete(options->rep, Slice(key, keylen)));
+}
+
+void rocksdb_delete_cf(rocksdb_t* db, const rocksdb_writeoptions_t* options,
+                       rocksdb_column_family_handle_t* column_family,
+                       const char* key, size_t keylen, char** errptr) {
+  SaveError(errptr, db->rep->Delete(options->rep, column_family->rep,
+                                    Slice(key, keylen)));
+}
+
+void rocksdb_merge(rocksdb_t* db, const rocksdb_writeoptions_t* options,
+                   const char* key, size_t keylen, const char* val,
+                   size_t vallen, char** errptr) {
+  SaveError(errptr, db->rep->Merge(options->rep, Slice(key, keylen),
+                                   Slice(val, vallen)));
+}
+
+void rocksdb_merge_cf(rocksdb_t* db, const rocksdb_writeoptions_t* options,
+                      rocksdb_column_family_handle_t* column_family,
+                      const char* key, size_t keylen, const char* val,
+                      size_t vallen, char** errptr) {
+  SaveError(errptr, db->rep->Merge(options->rep, column_family->rep,
+                                   Slice(key, keylen), Slice(val, vallen)));
+}
+
+void rocksdb_write(rocksdb_t* db, const rocksdb_writeoptions_t* options,
+                   rocksdb_writebatch_t* batch, char** errptr) {
+  SaveError(errptr, db->rep->Write(options->rep, &batch->rep));
+}
+
+void rocksdb_put_with_ts(rocksdb_t* db, const rocksdb_writeoptions_t* options,
+                         const char* key, size_t keylen, const char* ts,
+                         size_t tslen, const char* val, size_t vallen,
+                         char** errptr) {
+  SaveError(errptr, db->rep->Put(options->rep, Slice(key, keylen),
+                                 Slice(ts, tslen), Slice(val, vallen)));
+}
+
+void rocksdb_put_cf_with_ts(rocksdb_t* db,
+                            const rocksdb_writeoptions_t* options,
+                            rocksdb_column_family_handle_t* column_family,
+                            const char* key, size_t keylen, const char* ts,
+                            size_t tslen, const char* val, size_t vallen,
+                            char** errptr) {
+  SaveError(errptr,
+            db->rep->Put(options->rep, column_family->rep, Slice(key, keylen),
+                         Slice(ts, tslen), Slice(val, vallen)));
+}
+
+void rocksdb_delete_with_ts(rocksdb_t* db,
+                            const rocksdb_writeoptions_t* options,
+                            const char* key, size_t keylen, const char* ts,
+                            size_t tslen, char** errptr) {
+  SaveError(errptr, db->rep->Delete(options->rep, Slice(key, keylen),
+                                    Slice(ts, tslen)));
+}
+
+void rocksdb_delete_cf_with_ts(rocksdb_t* db,
+                               const rocksdb_writeoptions_t* options,
+                               rocksdb_column_family_handle_t* column_family,
+                               const char* key, size_t keylen, const char* ts,
+                               size_t tslen, char** errptr) {
+  SaveError(errptr, db->rep->Delete(options->rep, column_family->rep,
+                                    Slice(key, keylen), Slice(ts, tslen)));
+}
+
+void rocksdb_singledelete(rocksdb_t* db, const rocksdb_writeoptions_t* options,
+                          const char* key, size_t keylen, char** errptr) {
+  SaveError(errptr, db->rep->SingleDelete(options->rep, Slice(key, keylen)));
+}
+
+void rocksdb_singledelete_cf(rocksdb_t* db,
+                             const rocksdb_writeoptions_t* options,
+                             rocksdb_column_family_handle_t* column_family,
+                             const char* key, size_t keylen, char** errptr) {
+  SaveError(errptr, db->rep->SingleDelete(options->rep, column_family->rep,
+                                          Slice(key, keylen)));
+}
+
+void rocksdb_singledelete_with_ts(rocksdb_t* db,
+                                  const rocksdb_writeoptions_t* options,
+                                  const char* key, size_t keylen,
+                                  const char* ts, size_t tslen, char** errptr) {
+  SaveError(errptr, db->rep->SingleDelete(options->rep, Slice(key, keylen),
+                                          Slice(ts, tslen)));
+}
+
+void rocksdb_singledelete_cf_with_ts(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, const char* ts, size_t tslen, char** errptr) {
+  SaveError(errptr,
+            db->rep->SingleDelete(options->rep, column_family->rep,
+                                  Slice(key, keylen), Slice(ts, tslen)));
+}
+
+void rocksdb_delete_range_cf(rocksdb_t* db,
+                             const rocksdb_writeoptions_t* options,
+                             rocksdb_column_family_handle_t* column_family,
+                             const char* start_key, size_t start_key_len,
+                             const char* end_key, size_t end_key_len,
+                             char** errptr) {
+  SaveError(errptr, db->rep->DeleteRange(options->rep, column_family->rep,
+                                         Slice(start_key, start_key_len),
+                                         Slice(end_key, end_key_len)));
+}
+
+void rocksdb_flush(rocksdb_t* db, const rocksdb_flushoptions_t* options,
+                   char** errptr) {
+  SaveError(errptr, db->rep->Flush(options->rep));
+}
+
+void rocksdb_flush_cf(rocksdb_t* db, const rocksdb_flushoptions_t* options,
+                      rocksdb_column_family_handle_t* column_family,
+                      char** errptr) {
+  SaveError(errptr, db->rep->Flush(options->rep, column_family->rep));
+}
+
+void rocksdb_flush_wal(rocksdb_t* db, unsigned char sync, char** errptr) {
+  SaveError(errptr, db->rep->FlushWAL(sync));
+}
+
+void rocksdb_pause_background_work(rocksdb_t* db, char** errptr) {
+  SaveError(errptr, db->rep->PauseBackgroundWork());
+}
+
+void rocksdb_continue_background_work(rocksdb_t* db, char** errptr) {
+  SaveError(errptr, db->rep->ContinueBackgroundWork());
+}
+
+void rocksdb_disable_file_deletions(rocksdb_t* db, char** errptr) {
+  SaveError(errptr, db->rep->DisableFileDeletions());
+}
+
+void rocksdb_enable_file_deletions(rocksdb_t* db, char** errptr) {
+  SaveError(errptr, db->rep->EnableFileDeletions());
+}
+
+void rocksdb_verify_checksum(rocksdb_t* db, char** errptr) {
+  SaveError(errptr, db->rep->VerifyChecksum());
+}
+
+void rocksdb_verify_file_checksums(rocksdb_t* db, char** errptr) {
+  SaveError(errptr, db->rep->VerifyFileChecksums(ReadOptions()));
+}
+
+void rocksdb_destroy_db(const rocksdb_options_t* options, const char* name,
+                        char** errptr) {
+  SaveError(errptr, DestroyDB(name, options->rep));
+}
+
+void rocksdb_repair_db(const rocksdb_options_t* options, const char* name,
+                       char** errptr) {
+  SaveError(errptr, RepairDB(name, options->rep));
+}

c_api_gen/c_generated_db_simple_subset.h.inc

@@ -0,0 +1,130 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+// @generated
+// -----------------------------------------------------------------------------
+// Auto-generated by tools/c_api_gen/generate_c_api.py.
+// DO NOT EDIT THIS FILE DIRECTLY.
+// Source: tools/c_api_gen/spec.json
+// -----------------------------------------------------------------------------
+
+// To regenerate this file:
+//   python3 tools/c_api_gen/generate_c_api.py
+//     --spec tools/c_api_gen/spec.json --section 'DB data operations'
+//     --header-out
+//     c_api_gen/c_generated_db_simple_subset.h.inc
+//     --source-out
+//     c_api_gen/c_generated_db_simple_subset.cc.inc
+
+/* DB data operations */
+
+extern ROCKSDB_LIBRARY_API void rocksdb_put(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options, const char* key,
+    size_t keylen, const char* val, size_t vallen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_put_cf(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, const char* val, size_t vallen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_delete(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options, const char* key,
+    size_t keylen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_delete_cf(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_merge(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options, const char* key,
+    size_t keylen, const char* val, size_t vallen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_merge_cf(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, const char* val, size_t vallen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_write(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options,
+    rocksdb_writebatch_t* batch, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_put_with_ts(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options, const char* key,
+    size_t keylen, const char* ts, size_t tslen, const char* val, size_t vallen,
+    char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_put_cf_with_ts(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, const char* ts, size_t tslen, const char* val, size_t vallen,
+    char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_delete_with_ts(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options, const char* key,
+    size_t keylen, const char* ts, size_t tslen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_delete_cf_with_ts(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, const char* ts, size_t tslen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_singledelete(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options, const char* key,
+    size_t keylen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_singledelete_cf(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_singledelete_with_ts(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options, const char* key,
+    size_t keylen, const char* ts, size_t tslen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_singledelete_cf_with_ts(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, const char* ts, size_t tslen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_delete_range_cf(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* start_key,
+    size_t start_key_len, const char* end_key, size_t end_key_len,
+    char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_flush(
+    rocksdb_t* db, const rocksdb_flushoptions_t* options, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_flush_cf(
+    rocksdb_t* db, const rocksdb_flushoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_flush_wal(rocksdb_t* db,
+                                                  unsigned char sync,
+                                                  char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_pause_background_work(rocksdb_t* db,
+                                                              char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_continue_background_work(rocksdb_t* db,
+                                                                 char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_disable_file_deletions(rocksdb_t* db,
+                                                               char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_enable_file_deletions(rocksdb_t* db,
+                                                              char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_verify_checksum(rocksdb_t* db,
+                                                        char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_verify_file_checksums(rocksdb_t* db,
+                                                              char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_destroy_db(
+    const rocksdb_options_t* options, const char* name, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_repair_db(
+    const rocksdb_options_t* options, const char* name, char** errptr);

c_api_gen/c_generated_jobinfo_auto.cc.inc

@@ -0,0 +1,259 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+// @generated
+// -----------------------------------------------------------------------------
+// Auto-generated by tools/c_api_gen/auto_simple_bindings.py.
+// DO NOT EDIT THIS FILE DIRECTLY.
+// Sources:
+//   - include/rocksdb/listener.h
+//   - tools/c_api_gen/auto_simple_bindings_blocklist.json
+//   - include/rocksdb/c.h
+//   - tools/c_api_gen/c_base.cc
+// -----------------------------------------------------------------------------
+
+// To regenerate this file:
+//   python3 tools/c_api_gen/auto_simple_bindings.py
+
+// Output group: Auto-discovered JobInfo metadata simple
+
+/* FlushJobInfo */
+
+uint32_t rocksdb_flushjobinfo_cf_id(const rocksdb_flushjobinfo_t* info) {
+  return info->rep.cf_id;
+}
+
+const char* rocksdb_flushjobinfo_cf_name(const rocksdb_flushjobinfo_t* info,
+                                         size_t* size) {
+  *size = info->rep.cf_name.size();
+  return info->rep.cf_name.data();
+}
+
+const char* rocksdb_flushjobinfo_file_path(const rocksdb_flushjobinfo_t* info,
+                                           size_t* size) {
+  *size = info->rep.file_path.size();
+  return info->rep.file_path.data();
+}
+
+uint64_t rocksdb_flushjobinfo_file_number(const rocksdb_flushjobinfo_t* info) {
+  return info->rep.file_number;
+}
+
+uint64_t rocksdb_flushjobinfo_oldest_blob_file_number(
+    const rocksdb_flushjobinfo_t* info) {
+  return info->rep.oldest_blob_file_number;
+}
+
+uint64_t rocksdb_flushjobinfo_thread_id(const rocksdb_flushjobinfo_t* info) {
+  return info->rep.thread_id;
+}
+
+int rocksdb_flushjobinfo_job_id(const rocksdb_flushjobinfo_t* info) {
+  return info->rep.job_id;
+}
+
+unsigned char rocksdb_flushjobinfo_triggered_writes_slowdown(
+    const rocksdb_flushjobinfo_t* info) {
+  return info->rep.triggered_writes_slowdown;
+}
+
+unsigned char rocksdb_flushjobinfo_triggered_writes_stop(
+    const rocksdb_flushjobinfo_t* info) {
+  return info->rep.triggered_writes_stop;
+}
+
+uint64_t rocksdb_flushjobinfo_smallest_seqno(
+    const rocksdb_flushjobinfo_t* info) {
+  return info->rep.smallest_seqno;
+}
+
+uint64_t rocksdb_flushjobinfo_largest_seqno(
+    const rocksdb_flushjobinfo_t* info) {
+  return info->rep.largest_seqno;
+}
+
+uint32_t rocksdb_flushjobinfo_flush_reason(const rocksdb_flushjobinfo_t* info) {
+  return static_cast<uint32_t>(info->rep.flush_reason);
+}
+
+uint32_t rocksdb_flushjobinfo_blob_compression_type(
+    const rocksdb_flushjobinfo_t* info) {
+  return static_cast<uint32_t>(info->rep.blob_compression_type);
+}
+
+/* CompactionJobInfo */
+
+uint32_t rocksdb_compactionjobinfo_cf_id(
+    const rocksdb_compactionjobinfo_t* info) {
+  return info->rep.cf_id;
+}
+
+const char* rocksdb_compactionjobinfo_cf_name(
+    const rocksdb_compactionjobinfo_t* info, size_t* size) {
+  *size = info->rep.cf_name.size();
+  return info->rep.cf_name.data();
+}
+
+void rocksdb_compactionjobinfo_status(const rocksdb_compactionjobinfo_t* info,
+                                      char** errptr) {
+  SaveError(errptr, info->rep.status);
+}
+
+uint64_t rocksdb_compactionjobinfo_thread_id(
+    const rocksdb_compactionjobinfo_t* info) {
+  return info->rep.thread_id;
+}
+
+int rocksdb_compactionjobinfo_job_id(const rocksdb_compactionjobinfo_t* info) {
+  return info->rep.job_id;
+}
+
+int rocksdb_compactionjobinfo_num_l0_files(
+    const rocksdb_compactionjobinfo_t* info) {
+  return info->rep.num_l0_files;
+}
+
+int rocksdb_compactionjobinfo_base_input_level(
+    const rocksdb_compactionjobinfo_t* info) {
+  return info->rep.base_input_level;
+}
+
+int rocksdb_compactionjobinfo_output_level(
+    const rocksdb_compactionjobinfo_t* info) {
+  return info->rep.output_level;
+}
+
+uint32_t rocksdb_compactionjobinfo_compaction_reason(
+    const rocksdb_compactionjobinfo_t* info) {
+  return static_cast<uint32_t>(info->rep.compaction_reason);
+}
+
+uint32_t rocksdb_compactionjobinfo_compression(
+    const rocksdb_compactionjobinfo_t* info) {
+  return static_cast<uint32_t>(info->rep.compression);
+}
+
+uint32_t rocksdb_compactionjobinfo_blob_compression_type(
+    const rocksdb_compactionjobinfo_t* info) {
+  return static_cast<uint32_t>(info->rep.blob_compression_type);
+}
+
+unsigned char rocksdb_compactionjobinfo_aborted(
+    const rocksdb_compactionjobinfo_t* info) {
+  return info->rep.aborted;
+}
+
+/* SubcompactionJobInfo */
+
+uint32_t rocksdb_subcompactionjobinfo_cf_id(
+    const rocksdb_subcompactionjobinfo_t* info) {
+  return info->rep.cf_id;
+}
+
+const char* rocksdb_subcompactionjobinfo_cf_name(
+    const rocksdb_subcompactionjobinfo_t* info, size_t* size) {
+  *size = info->rep.cf_name.size();
+  return info->rep.cf_name.data();
+}
+
+void rocksdb_subcompactionjobinfo_status(
+    const rocksdb_subcompactionjobinfo_t* info, char** errptr) {
+  SaveError(errptr, info->rep.status);
+}
+
+uint64_t rocksdb_subcompactionjobinfo_thread_id(
+    const rocksdb_subcompactionjobinfo_t* info) {
+  return info->rep.thread_id;
+}
+
+int rocksdb_subcompactionjobinfo_job_id(
+    const rocksdb_subcompactionjobinfo_t* info) {
+  return info->rep.job_id;
+}
+
+int rocksdb_subcompactionjobinfo_subcompaction_job_id(
+    const rocksdb_subcompactionjobinfo_t* info) {
+  return info->rep.subcompaction_job_id;
+}
+
+int rocksdb_subcompactionjobinfo_base_input_level(
+    const rocksdb_subcompactionjobinfo_t* info) {
+  return info->rep.base_input_level;
+}
+
+int rocksdb_subcompactionjobinfo_output_level(
+    const rocksdb_subcompactionjobinfo_t* info) {
+  return info->rep.output_level;
+}
+
+uint32_t rocksdb_subcompactionjobinfo_compaction_reason(
+    const rocksdb_subcompactionjobinfo_t* info) {
+  return static_cast<uint32_t>(info->rep.compaction_reason);
+}
+
+uint32_t rocksdb_subcompactionjobinfo_compression(
+    const rocksdb_subcompactionjobinfo_t* info) {
+  return static_cast<uint32_t>(info->rep.compression);
+}
+
+uint32_t rocksdb_subcompactionjobinfo_blob_compression_type(
+    const rocksdb_subcompactionjobinfo_t* info) {
+  return static_cast<uint32_t>(info->rep.blob_compression_type);
+}
+
+/* ExternalFileIngestionInfo */
+
+const char* rocksdb_externalfileingestioninfo_cf_name(
+    const rocksdb_externalfileingestioninfo_t* info, size_t* size) {
+  *size = info->rep.cf_name.size();
+  return info->rep.cf_name.data();
+}
+
+const char* rocksdb_externalfileingestioninfo_external_file_path(
+    const rocksdb_externalfileingestioninfo_t* info, size_t* size) {
+  *size = info->rep.external_file_path.size();
+  return info->rep.external_file_path.data();
+}
+
+const char* rocksdb_externalfileingestioninfo_internal_file_path(
+    const rocksdb_externalfileingestioninfo_t* info, size_t* size) {
+  *size = info->rep.internal_file_path.size();
+  return info->rep.internal_file_path.data();
+}
+
+uint64_t rocksdb_externalfileingestioninfo_global_seqno(
+    const rocksdb_externalfileingestioninfo_t* info) {
+  return info->rep.global_seqno;
+}
+
+/* MemTableInfo */
+
+const char* rocksdb_memtableinfo_cf_name(const rocksdb_memtableinfo_t* info,
+                                         size_t* size) {
+  *size = info->rep.cf_name.size();
+  return info->rep.cf_name.data();
+}
+
+uint64_t rocksdb_memtableinfo_first_seqno(const rocksdb_memtableinfo_t* info) {
+  return info->rep.first_seqno;
+}
+
+uint64_t rocksdb_memtableinfo_earliest_seqno(
+    const rocksdb_memtableinfo_t* info) {
+  return info->rep.earliest_seqno;
+}
+
+uint64_t rocksdb_memtableinfo_num_entries(const rocksdb_memtableinfo_t* info) {
+  return info->rep.num_entries;
+}
+
+uint64_t rocksdb_memtableinfo_num_deletes(const rocksdb_memtableinfo_t* info) {
+  return info->rep.num_deletes;
+}
+
+const char* rocksdb_memtableinfo_newest_udt(const rocksdb_memtableinfo_t* info,
+                                            size_t* size) {
+  *size = info->rep.newest_udt.size();
+  return info->rep.newest_udt.data();
+}

c_api_gen/c_generated_jobinfo_auto.h.inc

@@ -0,0 +1,177 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+// @generated
+// -----------------------------------------------------------------------------
+// Auto-generated by tools/c_api_gen/auto_simple_bindings.py.
+// DO NOT EDIT THIS FILE DIRECTLY.
+// Sources:
+//   - include/rocksdb/listener.h
+//   - tools/c_api_gen/auto_simple_bindings_blocklist.json
+//   - include/rocksdb/c.h
+//   - tools/c_api_gen/c_base.cc
+// -----------------------------------------------------------------------------
+
+// To regenerate this file:
+//   python3 tools/c_api_gen/auto_simple_bindings.py
+
+// Output group: Auto-discovered JobInfo metadata simple
+
+/* FlushJobInfo */
+
+extern ROCKSDB_LIBRARY_API uint32_t
+rocksdb_flushjobinfo_cf_id(const rocksdb_flushjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API const char* rocksdb_flushjobinfo_cf_name(
+    const rocksdb_flushjobinfo_t* info, size_t* size);
+
+extern ROCKSDB_LIBRARY_API const char* rocksdb_flushjobinfo_file_path(
+    const rocksdb_flushjobinfo_t* info, size_t* size);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_flushjobinfo_file_number(const rocksdb_flushjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_flushjobinfo_oldest_blob_file_number(
+    const rocksdb_flushjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_flushjobinfo_thread_id(const rocksdb_flushjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API int rocksdb_flushjobinfo_job_id(
+    const rocksdb_flushjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API unsigned char
+rocksdb_flushjobinfo_triggered_writes_slowdown(
+    const rocksdb_flushjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API unsigned char
+rocksdb_flushjobinfo_triggered_writes_stop(const rocksdb_flushjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_flushjobinfo_smallest_seqno(const rocksdb_flushjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_flushjobinfo_largest_seqno(const rocksdb_flushjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API uint32_t
+rocksdb_flushjobinfo_flush_reason(const rocksdb_flushjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API uint32_t
+rocksdb_flushjobinfo_blob_compression_type(const rocksdb_flushjobinfo_t* info);
+
+/* CompactionJobInfo */
+
+extern ROCKSDB_LIBRARY_API uint32_t
+rocksdb_compactionjobinfo_cf_id(const rocksdb_compactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API const char* rocksdb_compactionjobinfo_cf_name(
+    const rocksdb_compactionjobinfo_t* info, size_t* size);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_compactionjobinfo_status(
+    const rocksdb_compactionjobinfo_t* info, char** errptr);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_compactionjobinfo_thread_id(const rocksdb_compactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API int rocksdb_compactionjobinfo_job_id(
+    const rocksdb_compactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API int rocksdb_compactionjobinfo_num_l0_files(
+    const rocksdb_compactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API int rocksdb_compactionjobinfo_base_input_level(
+    const rocksdb_compactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API int rocksdb_compactionjobinfo_output_level(
+    const rocksdb_compactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API uint32_t rocksdb_compactionjobinfo_compaction_reason(
+    const rocksdb_compactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API uint32_t
+rocksdb_compactionjobinfo_compression(const rocksdb_compactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API uint32_t
+rocksdb_compactionjobinfo_blob_compression_type(
+    const rocksdb_compactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API unsigned char rocksdb_compactionjobinfo_aborted(
+    const rocksdb_compactionjobinfo_t* info);
+
+/* SubcompactionJobInfo */
+
+extern ROCKSDB_LIBRARY_API uint32_t
+rocksdb_subcompactionjobinfo_cf_id(const rocksdb_subcompactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API const char* rocksdb_subcompactionjobinfo_cf_name(
+    const rocksdb_subcompactionjobinfo_t* info, size_t* size);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_subcompactionjobinfo_status(
+    const rocksdb_subcompactionjobinfo_t* info, char** errptr);
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_subcompactionjobinfo_thread_id(
+    const rocksdb_subcompactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API int rocksdb_subcompactionjobinfo_job_id(
+    const rocksdb_subcompactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API int
+rocksdb_subcompactionjobinfo_subcompaction_job_id(
+    const rocksdb_subcompactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API int rocksdb_subcompactionjobinfo_base_input_level(
+    const rocksdb_subcompactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API int rocksdb_subcompactionjobinfo_output_level(
+    const rocksdb_subcompactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API uint32_t
+rocksdb_subcompactionjobinfo_compaction_reason(
+    const rocksdb_subcompactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API uint32_t rocksdb_subcompactionjobinfo_compression(
+    const rocksdb_subcompactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API uint32_t
+rocksdb_subcompactionjobinfo_blob_compression_type(
+    const rocksdb_subcompactionjobinfo_t* info);
+
+/* ExternalFileIngestionInfo */
+
+extern ROCKSDB_LIBRARY_API const char*
+rocksdb_externalfileingestioninfo_cf_name(
+    const rocksdb_externalfileingestioninfo_t* info, size_t* size);
+
+extern ROCKSDB_LIBRARY_API const char*
+rocksdb_externalfileingestioninfo_external_file_path(
+    const rocksdb_externalfileingestioninfo_t* info, size_t* size);
+
+extern ROCKSDB_LIBRARY_API const char*
+rocksdb_externalfileingestioninfo_internal_file_path(
+    const rocksdb_externalfileingestioninfo_t* info, size_t* size);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_externalfileingestioninfo_global_seqno(
+    const rocksdb_externalfileingestioninfo_t* info);
+
+/* MemTableInfo */
+
+extern ROCKSDB_LIBRARY_API const char* rocksdb_memtableinfo_cf_name(
+    const rocksdb_memtableinfo_t* info, size_t* size);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_memtableinfo_first_seqno(const rocksdb_memtableinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_memtableinfo_earliest_seqno(const rocksdb_memtableinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_memtableinfo_num_entries(const rocksdb_memtableinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_memtableinfo_num_deletes(const rocksdb_memtableinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API const char* rocksdb_memtableinfo_newest_udt(
+    const rocksdb_memtableinfo_t* info, size_t* size);

c_api_gen/c_generated_jobinfo_metadata_subset.cc.inc

@@ -0,0 +1,88 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+// @generated
+// -----------------------------------------------------------------------------
+// Auto-generated by tools/c_api_gen/generate_c_api.py.
+// DO NOT EDIT THIS FILE DIRECTLY.
+// Source: tools/c_api_gen/spec.json
+// -----------------------------------------------------------------------------
+
+// To regenerate this file:
+//   python3 tools/c_api_gen/generate_c_api.py
+//     --spec tools/c_api_gen/spec.json --section 'JobInfo metadata simple'
+//     --header-out
+//     c_api_gen/c_generated_jobinfo_metadata_subset.h.inc
+//     --source-out
+//     c_api_gen/c_generated_jobinfo_metadata_subset.cc.inc
+
+/* JobInfo metadata simple */
+
+size_t rocksdb_compactionjobinfo_input_files_count(
+    const rocksdb_compactionjobinfo_t* info) {
+  return info->rep.input_files.size();
+}
+
+size_t rocksdb_compactionjobinfo_output_files_count(
+    const rocksdb_compactionjobinfo_t* info) {
+  return info->rep.output_files.size();
+}
+
+uint64_t rocksdb_compactionjobinfo_elapsed_micros(
+    const rocksdb_compactionjobinfo_t* info) {
+  return info->rep.stats.elapsed_micros;
+}
+
+uint64_t rocksdb_compactionjobinfo_num_corrupt_keys(
+    const rocksdb_compactionjobinfo_t* info) {
+  return info->rep.stats.num_corrupt_keys;
+}
+
+uint64_t rocksdb_compactionjobinfo_input_records(
+    const rocksdb_compactionjobinfo_t* info) {
+  return info->rep.stats.num_input_records;
+}
+
+uint64_t rocksdb_compactionjobinfo_output_records(
+    const rocksdb_compactionjobinfo_t* info) {
+  return info->rep.stats.num_output_records;
+}
+
+uint64_t rocksdb_compactionjobinfo_total_input_bytes(
+    const rocksdb_compactionjobinfo_t* info) {
+  return info->rep.stats.total_input_bytes;
+}
+
+uint64_t rocksdb_compactionjobinfo_total_output_bytes(
+    const rocksdb_compactionjobinfo_t* info) {
+  return info->rep.stats.total_output_bytes;
+}
+
+size_t rocksdb_compactionjobinfo_num_input_files(
+    const rocksdb_compactionjobinfo_t* info) {
+  return info->rep.stats.num_input_files;
+}
+
+size_t rocksdb_compactionjobinfo_num_input_files_at_output_level(
+    const rocksdb_compactionjobinfo_t* info) {
+  return info->rep.stats.num_input_files_at_output_level;
+}
+
+const char* rocksdb_writestallinfo_cf_name(const rocksdb_writestallinfo_t* info,
+                                           size_t* size) {
+  *size = info->rep.cf_name.size();
+  return info->rep.cf_name.data();
+}
+
+const rocksdb_writestallcondition_t* rocksdb_writestallinfo_cur(
+    const rocksdb_writestallinfo_t* info) {
+  return reinterpret_cast<const rocksdb_writestallcondition_t*>(
+      &info->rep.condition.cur);
+}
+
+const rocksdb_writestallcondition_t* rocksdb_writestallinfo_prev(
+    const rocksdb_writestallinfo_t* info) {
+  return reinterpret_cast<const rocksdb_writestallcondition_t*>(
+      &info->rep.condition.prev);
+}

c_api_gen/c_generated_jobinfo_metadata_subset.h.inc

@@ -0,0 +1,61 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+// @generated
+// -----------------------------------------------------------------------------
+// Auto-generated by tools/c_api_gen/generate_c_api.py.
+// DO NOT EDIT THIS FILE DIRECTLY.
+// Source: tools/c_api_gen/spec.json
+// -----------------------------------------------------------------------------
+
+// To regenerate this file:
+//   python3 tools/c_api_gen/generate_c_api.py
+//     --spec tools/c_api_gen/spec.json --section 'JobInfo metadata simple'
+//     --header-out
+//     c_api_gen/c_generated_jobinfo_metadata_subset.h.inc
+//     --source-out
+//     c_api_gen/c_generated_jobinfo_metadata_subset.cc.inc
+
+/* JobInfo metadata simple */
+
+extern ROCKSDB_LIBRARY_API size_t rocksdb_compactionjobinfo_input_files_count(
+    const rocksdb_compactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API size_t rocksdb_compactionjobinfo_output_files_count(
+    const rocksdb_compactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_compactionjobinfo_elapsed_micros(
+    const rocksdb_compactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_compactionjobinfo_num_corrupt_keys(
+    const rocksdb_compactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_compactionjobinfo_input_records(
+    const rocksdb_compactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_compactionjobinfo_output_records(
+    const rocksdb_compactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_compactionjobinfo_total_input_bytes(
+    const rocksdb_compactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_compactionjobinfo_total_output_bytes(
+    const rocksdb_compactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API size_t rocksdb_compactionjobinfo_num_input_files(
+    const rocksdb_compactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API size_t
+rocksdb_compactionjobinfo_num_input_files_at_output_level(
+    const rocksdb_compactionjobinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API const char* rocksdb_writestallinfo_cf_name(
+    const rocksdb_writestallinfo_t* info, size_t* size);
+
+extern ROCKSDB_LIBRARY_API const rocksdb_writestallcondition_t*
+rocksdb_writestallinfo_cur(const rocksdb_writestallinfo_t* info);
+
+extern ROCKSDB_LIBRARY_API const rocksdb_writestallcondition_t*
+rocksdb_writestallinfo_prev(const rocksdb_writestallinfo_t* info);

c_api_gen/c_generated_metadata_structs_auto.cc.inc

@@ -0,0 +1,527 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+// @generated
+// -----------------------------------------------------------------------------
+// Auto-generated by tools/c_api_gen/auto_simple_bindings.py.
+// DO NOT EDIT THIS FILE DIRECTLY.
+// Sources:
+//   - include/rocksdb/compaction_job_stats.h
+//   - include/rocksdb/listener.h
+//   - include/rocksdb/table_properties.h
+//   - tools/c_api_gen/auto_simple_bindings_blocklist.json
+//   - include/rocksdb/c.h
+//   - tools/c_api_gen/c_base.cc
+// -----------------------------------------------------------------------------
+
+// To regenerate this file:
+//   python3 tools/c_api_gen/auto_simple_bindings.py
+
+// Output group: Auto-discovered metadata view structs simple
+
+/* TableProperties */
+
+uint64_t rocksdb_table_properties_orig_file_number(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.orig_file_number;
+}
+
+uint64_t rocksdb_table_properties_data_size(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.data_size;
+}
+
+uint64_t rocksdb_table_properties_uncompressed_data_size(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.uncompressed_data_size;
+}
+
+uint64_t rocksdb_table_properties_index_size(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.index_size;
+}
+
+uint64_t rocksdb_table_properties_index_partitions(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.index_partitions;
+}
+
+uint64_t rocksdb_table_properties_top_level_index_size(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.top_level_index_size;
+}
+
+uint64_t rocksdb_table_properties_index_key_is_user_key(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.index_key_is_user_key;
+}
+
+uint64_t rocksdb_table_properties_index_value_is_delta_encoded(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.index_value_is_delta_encoded;
+}
+
+uint64_t rocksdb_table_properties_udi_is_primary_index(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.udi_is_primary_index;
+}
+
+uint64_t rocksdb_table_properties_filter_size(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.filter_size;
+}
+
+uint64_t rocksdb_table_properties_raw_key_size(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.raw_key_size;
+}
+
+uint64_t rocksdb_table_properties_raw_value_size(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.raw_value_size;
+}
+
+uint64_t rocksdb_table_properties_num_data_blocks(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.num_data_blocks;
+}
+
+uint64_t rocksdb_table_properties_num_data_blocks_compression_rejected(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.num_data_blocks_compression_rejected;
+}
+
+uint64_t rocksdb_table_properties_num_data_blocks_compression_bypassed(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.num_data_blocks_compression_bypassed;
+}
+
+uint64_t rocksdb_table_properties_num_uniform_blocks(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.num_uniform_blocks;
+}
+
+uint64_t rocksdb_table_properties_num_entries(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.num_entries;
+}
+
+uint64_t rocksdb_table_properties_num_filter_entries(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.num_filter_entries;
+}
+
+uint64_t rocksdb_table_properties_num_deletions(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.num_deletions;
+}
+
+uint64_t rocksdb_table_properties_num_merge_operands(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.num_merge_operands;
+}
+
+uint64_t rocksdb_table_properties_num_range_deletions(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.num_range_deletions;
+}
+
+uint64_t rocksdb_table_properties_format_version(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.format_version;
+}
+
+uint64_t rocksdb_table_properties_fixed_key_len(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.fixed_key_len;
+}
+
+uint64_t rocksdb_table_properties_column_family_id(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.column_family_id;
+}
+
+uint64_t rocksdb_table_properties_creation_time(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.creation_time;
+}
+
+uint64_t rocksdb_table_properties_oldest_key_time(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.oldest_key_time;
+}
+
+uint64_t rocksdb_table_properties_newest_key_time(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.newest_key_time;
+}
+
+uint64_t rocksdb_table_properties_file_creation_time(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.file_creation_time;
+}
+
+uint64_t rocksdb_table_properties_slow_compression_estimated_data_size(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.slow_compression_estimated_data_size;
+}
+
+uint64_t rocksdb_table_properties_fast_compression_estimated_data_size(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.fast_compression_estimated_data_size;
+}
+
+uint64_t rocksdb_table_properties_external_sst_file_global_seqno_offset(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.external_sst_file_global_seqno_offset;
+}
+
+uint64_t rocksdb_table_properties_tail_start_offset(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.tail_start_offset;
+}
+
+uint64_t rocksdb_table_properties_user_defined_timestamps_persisted(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.user_defined_timestamps_persisted;
+}
+
+uint64_t rocksdb_table_properties_key_largest_seqno(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.key_largest_seqno;
+}
+
+uint64_t rocksdb_table_properties_key_smallest_seqno(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.key_smallest_seqno;
+}
+
+uint64_t rocksdb_table_properties_data_block_restart_interval(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.data_block_restart_interval;
+}
+
+uint64_t rocksdb_table_properties_index_block_restart_interval(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.index_block_restart_interval;
+}
+
+uint64_t rocksdb_table_properties_separate_key_value_in_data_block(
+    const rocksdb_table_properties_t* props) {
+  return props->rep.separate_key_value_in_data_block;
+}
+
+const char* rocksdb_table_properties_db_id(
+    const rocksdb_table_properties_t* props, size_t* size) {
+  *size = props->rep.db_id.size();
+  return props->rep.db_id.data();
+}
+
+const char* rocksdb_table_properties_db_session_id(
+    const rocksdb_table_properties_t* props, size_t* size) {
+  *size = props->rep.db_session_id.size();
+  return props->rep.db_session_id.data();
+}
+
+const char* rocksdb_table_properties_db_host_id(
+    const rocksdb_table_properties_t* props, size_t* size) {
+  *size = props->rep.db_host_id.size();
+  return props->rep.db_host_id.data();
+}
+
+const char* rocksdb_table_properties_column_family_name(
+    const rocksdb_table_properties_t* props, size_t* size) {
+  *size = props->rep.column_family_name.size();
+  return props->rep.column_family_name.data();
+}
+
+const char* rocksdb_table_properties_filter_policy_name(
+    const rocksdb_table_properties_t* props, size_t* size) {
+  *size = props->rep.filter_policy_name.size();
+  return props->rep.filter_policy_name.data();
+}
+
+const char* rocksdb_table_properties_comparator_name(
+    const rocksdb_table_properties_t* props, size_t* size) {
+  *size = props->rep.comparator_name.size();
+  return props->rep.comparator_name.data();
+}
+
+const char* rocksdb_table_properties_merge_operator_name(
+    const rocksdb_table_properties_t* props, size_t* size) {
+  *size = props->rep.merge_operator_name.size();
+  return props->rep.merge_operator_name.data();
+}
+
+const char* rocksdb_table_properties_prefix_extractor_name(
+    const rocksdb_table_properties_t* props, size_t* size) {
+  *size = props->rep.prefix_extractor_name.size();
+  return props->rep.prefix_extractor_name.data();
+}
+
+const char* rocksdb_table_properties_property_collectors_names(
+    const rocksdb_table_properties_t* props, size_t* size) {
+  *size = props->rep.property_collectors_names.size();
+  return props->rep.property_collectors_names.data();
+}
+
+const char* rocksdb_table_properties_compression_name(
+    const rocksdb_table_properties_t* props, size_t* size) {
+  *size = props->rep.compression_name.size();
+  return props->rep.compression_name.data();
+}
+
+const char* rocksdb_table_properties_compression_options(
+    const rocksdb_table_properties_t* props, size_t* size) {
+  *size = props->rep.compression_options.size();
+  return props->rep.compression_options.data();
+}
+
+const char* rocksdb_table_properties_seqno_to_time_mapping(
+    const rocksdb_table_properties_t* props, size_t* size) {
+  *size = props->rep.seqno_to_time_mapping.size();
+  return props->rep.seqno_to_time_mapping.data();
+}
+
+/* CompactionJobStats */
+
+uint64_t rocksdb_compaction_job_stats_elapsed_micros(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.elapsed_micros;
+}
+
+uint64_t rocksdb_compaction_job_stats_cpu_micros(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.cpu_micros;
+}
+
+unsigned char rocksdb_compaction_job_stats_has_accurate_num_input_records(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.has_accurate_num_input_records;
+}
+
+uint64_t rocksdb_compaction_job_stats_num_input_records(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.num_input_records;
+}
+
+uint64_t rocksdb_compaction_job_stats_num_blobs_read(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.num_blobs_read;
+}
+
+size_t rocksdb_compaction_job_stats_num_input_files(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.num_input_files;
+}
+
+size_t rocksdb_compaction_job_stats_num_input_files_trivially_moved(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.num_input_files_trivially_moved;
+}
+
+size_t rocksdb_compaction_job_stats_num_input_files_at_output_level(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.num_input_files_at_output_level;
+}
+
+size_t rocksdb_compaction_job_stats_num_filtered_input_files(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.num_filtered_input_files;
+}
+
+size_t rocksdb_compaction_job_stats_num_filtered_input_files_at_output_level(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.num_filtered_input_files_at_output_level;
+}
+
+uint64_t rocksdb_compaction_job_stats_num_output_records(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.num_output_records;
+}
+
+size_t rocksdb_compaction_job_stats_num_output_files(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.num_output_files;
+}
+
+size_t rocksdb_compaction_job_stats_num_output_files_blob(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.num_output_files_blob;
+}
+
+unsigned char rocksdb_compaction_job_stats_is_full_compaction(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.is_full_compaction;
+}
+
+unsigned char rocksdb_compaction_job_stats_is_manual_compaction(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.is_manual_compaction;
+}
+
+unsigned char rocksdb_compaction_job_stats_is_remote_compaction(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.is_remote_compaction;
+}
+
+uint64_t rocksdb_compaction_job_stats_total_input_bytes(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.total_input_bytes;
+}
+
+uint64_t rocksdb_compaction_job_stats_total_blob_bytes_read(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.total_blob_bytes_read;
+}
+
+uint64_t rocksdb_compaction_job_stats_total_output_bytes(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.total_output_bytes;
+}
+
+uint64_t rocksdb_compaction_job_stats_total_output_bytes_blob(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.total_output_bytes_blob;
+}
+
+uint64_t rocksdb_compaction_job_stats_total_skipped_input_bytes(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.total_skipped_input_bytes;
+}
+
+uint64_t rocksdb_compaction_job_stats_num_records_replaced(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.num_records_replaced;
+}
+
+uint64_t rocksdb_compaction_job_stats_total_input_raw_key_bytes(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.total_input_raw_key_bytes;
+}
+
+uint64_t rocksdb_compaction_job_stats_total_input_raw_value_bytes(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.total_input_raw_value_bytes;
+}
+
+uint64_t rocksdb_compaction_job_stats_num_input_deletion_records(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.num_input_deletion_records;
+}
+
+uint64_t rocksdb_compaction_job_stats_num_expired_deletion_records(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.num_expired_deletion_records;
+}
+
+uint64_t rocksdb_compaction_job_stats_num_corrupt_keys(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.num_corrupt_keys;
+}
+
+uint64_t rocksdb_compaction_job_stats_file_write_nanos(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.file_write_nanos;
+}
+
+uint64_t rocksdb_compaction_job_stats_file_range_sync_nanos(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.file_range_sync_nanos;
+}
+
+uint64_t rocksdb_compaction_job_stats_file_fsync_nanos(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.file_fsync_nanos;
+}
+
+uint64_t rocksdb_compaction_job_stats_file_prepare_write_nanos(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.file_prepare_write_nanos;
+}
+
+const char* rocksdb_compaction_job_stats_smallest_output_key_prefix(
+    const rocksdb_compaction_job_stats_t* stats, size_t* size) {
+  *size = stats->rep.smallest_output_key_prefix.size();
+  return stats->rep.smallest_output_key_prefix.data();
+}
+
+const char* rocksdb_compaction_job_stats_largest_output_key_prefix(
+    const rocksdb_compaction_job_stats_t* stats, size_t* size) {
+  *size = stats->rep.largest_output_key_prefix.size();
+  return stats->rep.largest_output_key_prefix.data();
+}
+
+uint64_t rocksdb_compaction_job_stats_num_single_del_fallthru(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.num_single_del_fallthru;
+}
+
+uint64_t rocksdb_compaction_job_stats_num_single_del_mismatch(
+    const rocksdb_compaction_job_stats_t* stats) {
+  return stats->rep.num_single_del_mismatch;
+}
+
+/* CompactionFileInfo */
+
+int rocksdb_compaction_file_info_level(
+    const rocksdb_compaction_file_info_t* info) {
+  return info->rep.level;
+}
+
+uint64_t rocksdb_compaction_file_info_file_number(
+    const rocksdb_compaction_file_info_t* info) {
+  return info->rep.file_number;
+}
+
+uint64_t rocksdb_compaction_file_info_oldest_blob_file_number(
+    const rocksdb_compaction_file_info_t* info) {
+  return info->rep.oldest_blob_file_number;
+}
+
+/* BlobFileAdditionInfo */
+
+const char* rocksdb_blob_file_addition_info_blob_file_path(
+    const rocksdb_blob_file_addition_info_t* info, size_t* size) {
+  *size = info->rep.blob_file_path.size();
+  return info->rep.blob_file_path.data();
+}
+
+uint64_t rocksdb_blob_file_addition_info_blob_file_number(
+    const rocksdb_blob_file_addition_info_t* info) {
+  return info->rep.blob_file_number;
+}
+
+uint64_t rocksdb_blob_file_addition_info_total_blob_count(
+    const rocksdb_blob_file_addition_info_t* info) {
+  return info->rep.total_blob_count;
+}
+
+uint64_t rocksdb_blob_file_addition_info_total_blob_bytes(
+    const rocksdb_blob_file_addition_info_t* info) {
+  return info->rep.total_blob_bytes;
+}
+
+/* BlobFileGarbageInfo */
+
+const char* rocksdb_blob_file_garbage_info_blob_file_path(
+    const rocksdb_blob_file_garbage_info_t* info, size_t* size) {
+  *size = info->rep.blob_file_path.size();
+  return info->rep.blob_file_path.data();
+}
+
+uint64_t rocksdb_blob_file_garbage_info_blob_file_number(
+    const rocksdb_blob_file_garbage_info_t* info) {
+  return info->rep.blob_file_number;
+}
+
+uint64_t rocksdb_blob_file_garbage_info_garbage_blob_count(
+    const rocksdb_blob_file_garbage_info_t* info) {
+  return info->rep.garbage_blob_count;
+}
+
+uint64_t rocksdb_blob_file_garbage_info_garbage_blob_bytes(
+    const rocksdb_blob_file_garbage_info_t* info) {
+  return info->rep.garbage_blob_bytes;
+}

c_api_gen/c_generated_metadata_structs_auto.h.inc

@@ -0,0 +1,381 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+// @generated
+// -----------------------------------------------------------------------------
+// Auto-generated by tools/c_api_gen/auto_simple_bindings.py.
+// DO NOT EDIT THIS FILE DIRECTLY.
+// Sources:
+//   - include/rocksdb/compaction_job_stats.h
+//   - include/rocksdb/listener.h
+//   - include/rocksdb/table_properties.h
+//   - tools/c_api_gen/auto_simple_bindings_blocklist.json
+//   - include/rocksdb/c.h
+//   - tools/c_api_gen/c_base.cc
+// -----------------------------------------------------------------------------
+
+// To regenerate this file:
+//   python3 tools/c_api_gen/auto_simple_bindings.py
+
+// Output group: Auto-discovered metadata view structs simple
+
+/* TableProperties */
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_table_properties_orig_file_number(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_table_properties_data_size(const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_table_properties_uncompressed_data_size(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_table_properties_index_size(const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_table_properties_index_partitions(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_table_properties_top_level_index_size(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_table_properties_index_key_is_user_key(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_table_properties_index_value_is_delta_encoded(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_table_properties_udi_is_primary_index(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_table_properties_filter_size(const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_table_properties_raw_key_size(const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_table_properties_raw_value_size(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_table_properties_num_data_blocks(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_table_properties_num_data_blocks_compression_rejected(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_table_properties_num_data_blocks_compression_bypassed(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_table_properties_num_uniform_blocks(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_table_properties_num_entries(const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_table_properties_num_filter_entries(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_table_properties_num_deletions(const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_table_properties_num_merge_operands(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_table_properties_num_range_deletions(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_table_properties_format_version(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_table_properties_fixed_key_len(const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_table_properties_column_family_id(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_table_properties_creation_time(const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_table_properties_oldest_key_time(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_table_properties_newest_key_time(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_table_properties_file_creation_time(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_table_properties_slow_compression_estimated_data_size(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_table_properties_fast_compression_estimated_data_size(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_table_properties_external_sst_file_global_seqno_offset(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_table_properties_tail_start_offset(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_table_properties_user_defined_timestamps_persisted(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_table_properties_key_largest_seqno(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_table_properties_key_smallest_seqno(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_table_properties_data_block_restart_interval(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_table_properties_index_block_restart_interval(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_table_properties_separate_key_value_in_data_block(
+    const rocksdb_table_properties_t* props);
+
+extern ROCKSDB_LIBRARY_API const char* rocksdb_table_properties_db_id(
+    const rocksdb_table_properties_t* props, size_t* size);
+
+extern ROCKSDB_LIBRARY_API const char* rocksdb_table_properties_db_session_id(
+    const rocksdb_table_properties_t* props, size_t* size);
+
+extern ROCKSDB_LIBRARY_API const char* rocksdb_table_properties_db_host_id(
+    const rocksdb_table_properties_t* props, size_t* size);
+
+extern ROCKSDB_LIBRARY_API const char*
+rocksdb_table_properties_column_family_name(
+    const rocksdb_table_properties_t* props, size_t* size);
+
+extern ROCKSDB_LIBRARY_API const char*
+rocksdb_table_properties_filter_policy_name(
+    const rocksdb_table_properties_t* props, size_t* size);
+
+extern ROCKSDB_LIBRARY_API const char* rocksdb_table_properties_comparator_name(
+    const rocksdb_table_properties_t* props, size_t* size);
+
+extern ROCKSDB_LIBRARY_API const char*
+rocksdb_table_properties_merge_operator_name(
+    const rocksdb_table_properties_t* props, size_t* size);
+
+extern ROCKSDB_LIBRARY_API const char*
+rocksdb_table_properties_prefix_extractor_name(
+    const rocksdb_table_properties_t* props, size_t* size);
+
+extern ROCKSDB_LIBRARY_API const char*
+rocksdb_table_properties_property_collectors_names(
+    const rocksdb_table_properties_t* props, size_t* size);
+
+extern ROCKSDB_LIBRARY_API const char*
+rocksdb_table_properties_compression_name(
+    const rocksdb_table_properties_t* props, size_t* size);
+
+extern ROCKSDB_LIBRARY_API const char*
+rocksdb_table_properties_compression_options(
+    const rocksdb_table_properties_t* props, size_t* size);
+
+extern ROCKSDB_LIBRARY_API const char*
+rocksdb_table_properties_seqno_to_time_mapping(
+    const rocksdb_table_properties_t* props, size_t* size);
+
+/* CompactionJobStats */
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_compaction_job_stats_elapsed_micros(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_compaction_job_stats_cpu_micros(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API unsigned char
+rocksdb_compaction_job_stats_has_accurate_num_input_records(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_compaction_job_stats_num_input_records(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_compaction_job_stats_num_blobs_read(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API size_t rocksdb_compaction_job_stats_num_input_files(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API size_t
+rocksdb_compaction_job_stats_num_input_files_trivially_moved(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API size_t
+rocksdb_compaction_job_stats_num_input_files_at_output_level(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API size_t
+rocksdb_compaction_job_stats_num_filtered_input_files(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API size_t
+rocksdb_compaction_job_stats_num_filtered_input_files_at_output_level(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_compaction_job_stats_num_output_records(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API size_t rocksdb_compaction_job_stats_num_output_files(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API size_t
+rocksdb_compaction_job_stats_num_output_files_blob(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API unsigned char
+rocksdb_compaction_job_stats_is_full_compaction(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API unsigned char
+rocksdb_compaction_job_stats_is_manual_compaction(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API unsigned char
+rocksdb_compaction_job_stats_is_remote_compaction(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_compaction_job_stats_total_input_bytes(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_compaction_job_stats_total_blob_bytes_read(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_compaction_job_stats_total_output_bytes(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_compaction_job_stats_total_output_bytes_blob(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_compaction_job_stats_total_skipped_input_bytes(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_compaction_job_stats_num_records_replaced(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_compaction_job_stats_total_input_raw_key_bytes(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_compaction_job_stats_total_input_raw_value_bytes(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_compaction_job_stats_num_input_deletion_records(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_compaction_job_stats_num_expired_deletion_records(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_compaction_job_stats_num_corrupt_keys(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_compaction_job_stats_file_write_nanos(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_compaction_job_stats_file_range_sync_nanos(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_compaction_job_stats_file_fsync_nanos(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_compaction_job_stats_file_prepare_write_nanos(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API const char*
+rocksdb_compaction_job_stats_smallest_output_key_prefix(
+    const rocksdb_compaction_job_stats_t* stats, size_t* size);
+
+extern ROCKSDB_LIBRARY_API const char*
+rocksdb_compaction_job_stats_largest_output_key_prefix(
+    const rocksdb_compaction_job_stats_t* stats, size_t* size);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_compaction_job_stats_num_single_del_fallthru(
+    const rocksdb_compaction_job_stats_t* stats);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_compaction_job_stats_num_single_del_mismatch(
+    const rocksdb_compaction_job_stats_t* stats);
+
+/* CompactionFileInfo */
+
+extern ROCKSDB_LIBRARY_API int rocksdb_compaction_file_info_level(
+    const rocksdb_compaction_file_info_t* info);
+
+extern ROCKSDB_LIBRARY_API uint64_t rocksdb_compaction_file_info_file_number(
+    const rocksdb_compaction_file_info_t* info);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_compaction_file_info_oldest_blob_file_number(
+    const rocksdb_compaction_file_info_t* info);
+
+/* BlobFileAdditionInfo */
+
+extern ROCKSDB_LIBRARY_API const char*
+rocksdb_blob_file_addition_info_blob_file_path(
+    const rocksdb_blob_file_addition_info_t* info, size_t* size);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_blob_file_addition_info_blob_file_number(
+    const rocksdb_blob_file_addition_info_t* info);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_blob_file_addition_info_total_blob_count(
+    const rocksdb_blob_file_addition_info_t* info);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_blob_file_addition_info_total_blob_bytes(
+    const rocksdb_blob_file_addition_info_t* info);
+
+/* BlobFileGarbageInfo */
+
+extern ROCKSDB_LIBRARY_API const char*
+rocksdb_blob_file_garbage_info_blob_file_path(
+    const rocksdb_blob_file_garbage_info_t* info, size_t* size);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_blob_file_garbage_info_blob_file_number(
+    const rocksdb_blob_file_garbage_info_t* info);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_blob_file_garbage_info_garbage_blob_count(
+    const rocksdb_blob_file_garbage_info_t* info);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_blob_file_garbage_info_garbage_blob_bytes(
+    const rocksdb_blob_file_garbage_info_t* info);

c_api_gen/c_generated_options_subset.cc.inc

@@ -0,0 +1,88 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+// @generated
+// -----------------------------------------------------------------------------
+// Auto-generated by tools/c_api_gen/generate_c_api.py.
+// DO NOT EDIT THIS FILE DIRECTLY.
+// Source: tools/c_api_gen/spec.json
+// -----------------------------------------------------------------------------
+
+// To regenerate this file:
+//   python3 tools/c_api_gen/generate_c_api.py
+//     --spec tools/c_api_gen/spec.json --section 'Options simple'
+//     --header-out
+//     c_api_gen/c_generated_options_subset.h.inc
+//     --source-out
+//     c_api_gen/c_generated_options_subset.cc.inc
+
+/* Options simple */
+
+void rocksdb_options_set_compression_options_zstd_max_train_bytes(
+    rocksdb_options_t* opt, int zstd_max_train_bytes) {
+  opt->rep.compression_opts.zstd_max_train_bytes = zstd_max_train_bytes;
+}
+
+int rocksdb_options_get_compression_options_zstd_max_train_bytes(
+    rocksdb_options_t* opt) {
+  return opt->rep.compression_opts.zstd_max_train_bytes;
+}
+
+void rocksdb_options_set_compression_options_use_zstd_dict_trainer(
+    rocksdb_options_t* opt, unsigned char use_zstd_dict_trainer) {
+  opt->rep.compression_opts.use_zstd_dict_trainer = use_zstd_dict_trainer;
+}
+
+unsigned char rocksdb_options_get_compression_options_use_zstd_dict_trainer(
+    rocksdb_options_t* opt) {
+  return opt->rep.compression_opts.use_zstd_dict_trainer;
+}
+
+void rocksdb_options_set_compression_options_parallel_threads(
+    rocksdb_options_t* opt, int value) {
+  opt->rep.compression_opts.parallel_threads = value;
+}
+
+int rocksdb_options_get_compression_options_parallel_threads(
+    rocksdb_options_t* opt) {
+  return opt->rep.compression_opts.parallel_threads;
+}
+
+void rocksdb_options_set_compression_options_max_dict_buffer_bytes(
+    rocksdb_options_t* opt, uint64_t max_dict_buffer_bytes) {
+  opt->rep.compression_opts.max_dict_buffer_bytes = max_dict_buffer_bytes;
+}
+
+uint64_t rocksdb_options_get_compression_options_max_dict_buffer_bytes(
+    rocksdb_options_t* opt) {
+  return opt->rep.compression_opts.max_dict_buffer_bytes;
+}
+
+unsigned char
+rocksdb_options_get_bottommost_compression_options_use_zstd_dict_trainer(
+    rocksdb_options_t* opt) {
+  return opt->rep.bottommost_compression_opts.use_zstd_dict_trainer;
+}
+
+void rocksdb_options_set_use_fsync(rocksdb_options_t* opt, int use_fsync) {
+  opt->rep.use_fsync = use_fsync;
+}
+
+int rocksdb_options_get_use_fsync(rocksdb_options_t* opt) {
+  return opt->rep.use_fsync;
+}
+
+void rocksdb_options_set_disable_auto_compactions(rocksdb_options_t* opt,
+                                                  int disable) {
+  opt->rep.disable_auto_compactions = disable;
+}
+
+void rocksdb_options_set_optimize_filters_for_hits(rocksdb_options_t* opt,
+                                                   int v) {
+  opt->rep.optimize_filters_for_hits = v;
+}
+
+void rocksdb_options_set_report_bg_io_stats(rocksdb_options_t* opt, int v) {
+  opt->rep.report_bg_io_stats = v;
+}

c_api_gen/c_generated_options_subset.h.inc

@@ -0,0 +1,71 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+// @generated
+// -----------------------------------------------------------------------------
+// Auto-generated by tools/c_api_gen/generate_c_api.py.
+// DO NOT EDIT THIS FILE DIRECTLY.
+// Source: tools/c_api_gen/spec.json
+// -----------------------------------------------------------------------------
+
+// To regenerate this file:
+//   python3 tools/c_api_gen/generate_c_api.py
+//     --spec tools/c_api_gen/spec.json --section 'Options simple'
+//     --header-out
+//     c_api_gen/c_generated_options_subset.h.inc
+//     --source-out
+//     c_api_gen/c_generated_options_subset.cc.inc
+
+/* Options simple */
+
+extern ROCKSDB_LIBRARY_API void
+rocksdb_options_set_compression_options_zstd_max_train_bytes(
+    rocksdb_options_t* opt, int zstd_max_train_bytes);
+
+extern ROCKSDB_LIBRARY_API int
+rocksdb_options_get_compression_options_zstd_max_train_bytes(
+    rocksdb_options_t* opt);
+
+extern ROCKSDB_LIBRARY_API void
+rocksdb_options_set_compression_options_use_zstd_dict_trainer(
+    rocksdb_options_t* opt, unsigned char use_zstd_dict_trainer);
+
+extern ROCKSDB_LIBRARY_API unsigned char
+rocksdb_options_get_compression_options_use_zstd_dict_trainer(
+    rocksdb_options_t* opt);
+
+extern ROCKSDB_LIBRARY_API void
+rocksdb_options_set_compression_options_parallel_threads(rocksdb_options_t* opt,
+                                                         int value);
+
+extern ROCKSDB_LIBRARY_API int
+rocksdb_options_get_compression_options_parallel_threads(
+    rocksdb_options_t* opt);
+
+extern ROCKSDB_LIBRARY_API void
+rocksdb_options_set_compression_options_max_dict_buffer_bytes(
+    rocksdb_options_t* opt, uint64_t max_dict_buffer_bytes);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_options_get_compression_options_max_dict_buffer_bytes(
+    rocksdb_options_t* opt);
+
+extern ROCKSDB_LIBRARY_API unsigned char
+rocksdb_options_get_bottommost_compression_options_use_zstd_dict_trainer(
+    rocksdb_options_t* opt);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_options_set_use_fsync(
+    rocksdb_options_t* opt, int use_fsync);
+
+extern ROCKSDB_LIBRARY_API int rocksdb_options_get_use_fsync(
+    rocksdb_options_t* opt);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_options_set_disable_auto_compactions(
+    rocksdb_options_t* opt, int disable);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_options_set_optimize_filters_for_hits(
+    rocksdb_options_t* opt, int v);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_options_set_report_bg_io_stats(
+    rocksdb_options_t* opt, int v);

c_api_gen/c_generated_readoptions_auto.cc.inc

@@ -0,0 +1,240 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+// @generated
+// -----------------------------------------------------------------------------
+// Auto-generated by tools/c_api_gen/auto_simple_bindings.py.
+// DO NOT EDIT THIS FILE DIRECTLY.
+// Sources:
+//   - include/rocksdb/options.h
+//   - tools/c_api_gen/auto_simple_bindings_blocklist.json
+//   - include/rocksdb/c.h
+//   - tools/c_api_gen/c_base.cc
+// -----------------------------------------------------------------------------
+
+// To regenerate this file:
+//   python3 tools/c_api_gen/auto_simple_bindings.py
+
+// Output group: Auto-discovered ReadOptions simple
+
+/* ReadOptions */
+
+void rocksdb_readoptions_set_deadline(rocksdb_readoptions_t* opt,
+                                      uint64_t microseconds) {
+  opt->rep.deadline = std::chrono::microseconds(microseconds);
+}
+
+uint64_t rocksdb_readoptions_get_deadline(rocksdb_readoptions_t* opt) {
+  return opt->rep.deadline.count();
+}
+
+void rocksdb_readoptions_set_io_timeout(rocksdb_readoptions_t* opt,
+                                        uint64_t microseconds) {
+  opt->rep.io_timeout = std::chrono::microseconds(microseconds);
+}
+
+uint64_t rocksdb_readoptions_get_io_timeout(rocksdb_readoptions_t* opt) {
+  return opt->rep.io_timeout.count();
+}
+
+void rocksdb_readoptions_set_read_tier(rocksdb_readoptions_t* opt, int v) {
+  opt->rep.read_tier = static_cast<decltype(opt->rep.read_tier)>(v);
+}
+
+int rocksdb_readoptions_get_read_tier(rocksdb_readoptions_t* opt) {
+  return static_cast<int>(opt->rep.read_tier);
+}
+
+void rocksdb_readoptions_set_rate_limiter_priority(rocksdb_readoptions_t* opt,
+                                                   int v) {
+  opt->rep.rate_limiter_priority =
+      static_cast<decltype(opt->rep.rate_limiter_priority)>(v);
+}
+
+int rocksdb_readoptions_get_rate_limiter_priority(rocksdb_readoptions_t* opt) {
+  return static_cast<int>(opt->rep.rate_limiter_priority);
+}
+
+void rocksdb_readoptions_set_value_size_soft_limit(rocksdb_readoptions_t* opt,
+                                                   uint64_t v) {
+  opt->rep.value_size_soft_limit = v;
+}
+
+uint64_t rocksdb_readoptions_get_value_size_soft_limit(
+    rocksdb_readoptions_t* opt) {
+  return opt->rep.value_size_soft_limit;
+}
+
+void rocksdb_readoptions_set_verify_checksums(rocksdb_readoptions_t* opt,
+                                              unsigned char v) {
+  opt->rep.verify_checksums = v;
+}
+
+unsigned char rocksdb_readoptions_get_verify_checksums(
+    rocksdb_readoptions_t* opt) {
+  return opt->rep.verify_checksums;
+}
+
+void rocksdb_readoptions_set_fill_cache(rocksdb_readoptions_t* opt,
+                                        unsigned char v) {
+  opt->rep.fill_cache = v;
+}
+
+unsigned char rocksdb_readoptions_get_fill_cache(rocksdb_readoptions_t* opt) {
+  return opt->rep.fill_cache;
+}
+
+void rocksdb_readoptions_set_ignore_range_deletions(rocksdb_readoptions_t* opt,
+                                                    unsigned char v) {
+  opt->rep.ignore_range_deletions = v;
+}
+
+unsigned char rocksdb_readoptions_get_ignore_range_deletions(
+    rocksdb_readoptions_t* opt) {
+  return opt->rep.ignore_range_deletions;
+}
+
+void rocksdb_readoptions_set_async_io(rocksdb_readoptions_t* opt,
+                                      unsigned char v) {
+  opt->rep.async_io = v;
+}
+
+unsigned char rocksdb_readoptions_get_async_io(rocksdb_readoptions_t* opt) {
+  return opt->rep.async_io;
+}
+
+void rocksdb_readoptions_set_optimize_multiget_for_io(
+    rocksdb_readoptions_t* opt, unsigned char v) {
+  opt->rep.optimize_multiget_for_io = v;
+}
+
+unsigned char rocksdb_readoptions_get_optimize_multiget_for_io(
+    rocksdb_readoptions_t* opt) {
+  return opt->rep.optimize_multiget_for_io;
+}
+
+void rocksdb_readoptions_set_readahead_size(rocksdb_readoptions_t* opt,
+                                            size_t v) {
+  opt->rep.readahead_size = v;
+}
+
+size_t rocksdb_readoptions_get_readahead_size(rocksdb_readoptions_t* opt) {
+  return opt->rep.readahead_size;
+}
+
+void rocksdb_readoptions_set_max_skippable_internal_keys(
+    rocksdb_readoptions_t* opt, uint64_t v) {
+  opt->rep.max_skippable_internal_keys = v;
+}
+
+uint64_t rocksdb_readoptions_get_max_skippable_internal_keys(
+    rocksdb_readoptions_t* opt) {
+  return opt->rep.max_skippable_internal_keys;
+}
+
+void rocksdb_readoptions_set_tailing(rocksdb_readoptions_t* opt,
+                                     unsigned char v) {
+  opt->rep.tailing = v;
+}
+
+unsigned char rocksdb_readoptions_get_tailing(rocksdb_readoptions_t* opt) {
+  return opt->rep.tailing;
+}
+
+void rocksdb_readoptions_set_total_order_seek(rocksdb_readoptions_t* opt,
+                                              unsigned char v) {
+  opt->rep.total_order_seek = v;
+}
+
+unsigned char rocksdb_readoptions_get_total_order_seek(
+    rocksdb_readoptions_t* opt) {
+  return opt->rep.total_order_seek;
+}
+
+void rocksdb_readoptions_set_auto_prefix_mode(rocksdb_readoptions_t* opt,
+                                              unsigned char v) {
+  opt->rep.auto_prefix_mode = v;
+}
+
+unsigned char rocksdb_readoptions_get_auto_prefix_mode(
+    rocksdb_readoptions_t* opt) {
+  return opt->rep.auto_prefix_mode;
+}
+
+void rocksdb_readoptions_set_prefix_same_as_start(rocksdb_readoptions_t* opt,
+                                                  unsigned char v) {
+  opt->rep.prefix_same_as_start = v;
+}
+
+unsigned char rocksdb_readoptions_get_prefix_same_as_start(
+    rocksdb_readoptions_t* opt) {
+  return opt->rep.prefix_same_as_start;
+}
+
+void rocksdb_readoptions_set_pin_data(rocksdb_readoptions_t* opt,
+                                      unsigned char v) {
+  opt->rep.pin_data = v;
+}
+
+unsigned char rocksdb_readoptions_get_pin_data(rocksdb_readoptions_t* opt) {
+  return opt->rep.pin_data;
+}
+
+void rocksdb_readoptions_set_adaptive_readahead(rocksdb_readoptions_t* opt,
+                                                unsigned char v) {
+  opt->rep.adaptive_readahead = v;
+}
+
+unsigned char rocksdb_readoptions_get_adaptive_readahead(
+    rocksdb_readoptions_t* opt) {
+  return opt->rep.adaptive_readahead;
+}
+
+void rocksdb_readoptions_set_background_purge_on_iterator_cleanup(
+    rocksdb_readoptions_t* opt, unsigned char v) {
+  opt->rep.background_purge_on_iterator_cleanup = v;
+}
+
+unsigned char rocksdb_readoptions_get_background_purge_on_iterator_cleanup(
+    rocksdb_readoptions_t* opt) {
+  return opt->rep.background_purge_on_iterator_cleanup;
+}
+
+void rocksdb_readoptions_set_auto_readahead_size(rocksdb_readoptions_t* opt,
+                                                 unsigned char v) {
+  opt->rep.auto_readahead_size = v;
+}
+
+unsigned char rocksdb_readoptions_get_auto_readahead_size(
+    rocksdb_readoptions_t* opt) {
+  return opt->rep.auto_readahead_size;
+}
+
+void rocksdb_readoptions_set_allow_unprepared_value(rocksdb_readoptions_t* opt,
+                                                    unsigned char v) {
+  opt->rep.allow_unprepared_value = v;
+}
+
+unsigned char rocksdb_readoptions_get_allow_unprepared_value(
+    rocksdb_readoptions_t* opt) {
+  return opt->rep.allow_unprepared_value;
+}
+
+void rocksdb_readoptions_set_auto_refresh_iterator_with_snapshot(
+    rocksdb_readoptions_t* opt, unsigned char v) {
+  opt->rep.auto_refresh_iterator_with_snapshot = v;
+}
+
+unsigned char rocksdb_readoptions_get_auto_refresh_iterator_with_snapshot(
+    rocksdb_readoptions_t* opt) {
+  return opt->rep.auto_refresh_iterator_with_snapshot;
+}
+
+void rocksdb_readoptions_set_io_activity(rocksdb_readoptions_t* opt, int v) {
+  opt->rep.io_activity = static_cast<decltype(opt->rep.io_activity)>(v);
+}
+
+int rocksdb_readoptions_get_io_activity(rocksdb_readoptions_t* opt) {
+  return static_cast<int>(opt->rep.io_activity);
+}

c_api_gen/c_generated_readoptions_auto.h.inc

@@ -0,0 +1,165 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+// @generated
+// -----------------------------------------------------------------------------
+// Auto-generated by tools/c_api_gen/auto_simple_bindings.py.
+// DO NOT EDIT THIS FILE DIRECTLY.
+// Sources:
+//   - include/rocksdb/options.h
+//   - tools/c_api_gen/auto_simple_bindings_blocklist.json
+//   - include/rocksdb/c.h
+//   - tools/c_api_gen/c_base.cc
+// -----------------------------------------------------------------------------
+
+// To regenerate this file:
+//   python3 tools/c_api_gen/auto_simple_bindings.py
+
+// Output group: Auto-discovered ReadOptions simple
+
+/* ReadOptions */
+
+extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_deadline(
+    rocksdb_readoptions_t* opt, uint64_t microseconds);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_readoptions_get_deadline(rocksdb_readoptions_t* opt);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_io_timeout(
+    rocksdb_readoptions_t* opt, uint64_t microseconds);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_readoptions_get_io_timeout(rocksdb_readoptions_t* opt);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_read_tier(
+    rocksdb_readoptions_t* opt, int v);
+
+extern ROCKSDB_LIBRARY_API int rocksdb_readoptions_get_read_tier(
+    rocksdb_readoptions_t* opt);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_rate_limiter_priority(
+    rocksdb_readoptions_t* opt, int v);
+
+extern ROCKSDB_LIBRARY_API int rocksdb_readoptions_get_rate_limiter_priority(
+    rocksdb_readoptions_t* opt);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_value_size_soft_limit(
+    rocksdb_readoptions_t* opt, uint64_t v);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_readoptions_get_value_size_soft_limit(rocksdb_readoptions_t* opt);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_verify_checksums(
+    rocksdb_readoptions_t* opt, unsigned char v);
+
+extern ROCKSDB_LIBRARY_API unsigned char
+rocksdb_readoptions_get_verify_checksums(rocksdb_readoptions_t* opt);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_fill_cache(
+    rocksdb_readoptions_t* opt, unsigned char v);
+
+extern ROCKSDB_LIBRARY_API unsigned char rocksdb_readoptions_get_fill_cache(
+    rocksdb_readoptions_t* opt);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_ignore_range_deletions(
+    rocksdb_readoptions_t* opt, unsigned char v);
+
+extern ROCKSDB_LIBRARY_API unsigned char
+rocksdb_readoptions_get_ignore_range_deletions(rocksdb_readoptions_t* opt);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_async_io(
+    rocksdb_readoptions_t* opt, unsigned char v);
+
+extern ROCKSDB_LIBRARY_API unsigned char rocksdb_readoptions_get_async_io(
+    rocksdb_readoptions_t* opt);
+
+extern ROCKSDB_LIBRARY_API void
+rocksdb_readoptions_set_optimize_multiget_for_io(rocksdb_readoptions_t* opt,
+                                                 unsigned char v);
+
+extern ROCKSDB_LIBRARY_API unsigned char
+rocksdb_readoptions_get_optimize_multiget_for_io(rocksdb_readoptions_t* opt);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_readahead_size(
+    rocksdb_readoptions_t* opt, size_t v);
+
+extern ROCKSDB_LIBRARY_API size_t
+rocksdb_readoptions_get_readahead_size(rocksdb_readoptions_t* opt);
+
+extern ROCKSDB_LIBRARY_API void
+rocksdb_readoptions_set_max_skippable_internal_keys(rocksdb_readoptions_t* opt,
+                                                    uint64_t v);
+
+extern ROCKSDB_LIBRARY_API uint64_t
+rocksdb_readoptions_get_max_skippable_internal_keys(rocksdb_readoptions_t* opt);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_tailing(
+    rocksdb_readoptions_t* opt, unsigned char v);
+
+extern ROCKSDB_LIBRARY_API unsigned char rocksdb_readoptions_get_tailing(
+    rocksdb_readoptions_t* opt);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_total_order_seek(
+    rocksdb_readoptions_t* opt, unsigned char v);
+
+extern ROCKSDB_LIBRARY_API unsigned char
+rocksdb_readoptions_get_total_order_seek(rocksdb_readoptions_t* opt);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_auto_prefix_mode(
+    rocksdb_readoptions_t* opt, unsigned char v);
+
+extern ROCKSDB_LIBRARY_API unsigned char
+rocksdb_readoptions_get_auto_prefix_mode(rocksdb_readoptions_t* opt);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_prefix_same_as_start(
+    rocksdb_readoptions_t* opt, unsigned char v);
+
+extern ROCKSDB_LIBRARY_API unsigned char
+rocksdb_readoptions_get_prefix_same_as_start(rocksdb_readoptions_t* opt);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_pin_data(
+    rocksdb_readoptions_t* opt, unsigned char v);
+
+extern ROCKSDB_LIBRARY_API unsigned char rocksdb_readoptions_get_pin_data(
+    rocksdb_readoptions_t* opt);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_adaptive_readahead(
+    rocksdb_readoptions_t* opt, unsigned char v);
+
+extern ROCKSDB_LIBRARY_API unsigned char
+rocksdb_readoptions_get_adaptive_readahead(rocksdb_readoptions_t* opt);
+
+extern ROCKSDB_LIBRARY_API void
+rocksdb_readoptions_set_background_purge_on_iterator_cleanup(
+    rocksdb_readoptions_t* opt, unsigned char v);
+
+extern ROCKSDB_LIBRARY_API unsigned char
+rocksdb_readoptions_get_background_purge_on_iterator_cleanup(
+    rocksdb_readoptions_t* opt);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_auto_readahead_size(
+    rocksdb_readoptions_t* opt, unsigned char v);
+
+extern ROCKSDB_LIBRARY_API unsigned char
+rocksdb_readoptions_get_auto_readahead_size(rocksdb_readoptions_t* opt);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_allow_unprepared_value(
+    rocksdb_readoptions_t* opt, unsigned char v);
+
+extern ROCKSDB_LIBRARY_API unsigned char
+rocksdb_readoptions_get_allow_unprepared_value(rocksdb_readoptions_t* opt);
+
+extern ROCKSDB_LIBRARY_API void
+rocksdb_readoptions_set_auto_refresh_iterator_with_snapshot(
+    rocksdb_readoptions_t* opt, unsigned char v);
+
+extern ROCKSDB_LIBRARY_API unsigned char
+rocksdb_readoptions_get_auto_refresh_iterator_with_snapshot(
+    rocksdb_readoptions_t* opt);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_io_activity(
+    rocksdb_readoptions_t* opt, int v);
+
+extern ROCKSDB_LIBRARY_API int rocksdb_readoptions_get_io_activity(
+    rocksdb_readoptions_t* opt);

c_api_gen/c_generated_subset.cc.inc

@@ -0,0 +1,375 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+// @generated
+// -----------------------------------------------------------------------------
+// Auto-generated by tools/c_api_gen/generate_c_api.py.
+// DO NOT EDIT THIS FILE DIRECTLY.
+// Source: tools/c_api_gen/spec.json
+// -----------------------------------------------------------------------------
+
+// To regenerate this file:
+//   python3 tools/c_api_gen/generate_c_api.py
+//     --spec tools/c_api_gen/spec.json --section 'DB data operations'
+//     --section 'WriteBatch' --section 'Transaction'
+//     --section 'TransactionDB simple'
+//     --header-out
+//     c_api_gen/c_generated_subset.h.inc
+//     --source-out
+//     c_api_gen/c_generated_subset.cc.inc
+
+/* DB data operations */
+
+void rocksdb_put(rocksdb_t* db, const rocksdb_writeoptions_t* options,
+                 const char* key, size_t keylen, const char* val, size_t vallen,
+                 char** errptr) {
+  SaveError(errptr,
+            db->rep->Put(options->rep, Slice(key, keylen), Slice(val, vallen)));
+}
+
+void rocksdb_put_cf(rocksdb_t* db, const rocksdb_writeoptions_t* options,
+                    rocksdb_column_family_handle_t* column_family,
+                    const char* key, size_t keylen, const char* val,
+                    size_t vallen, char** errptr) {
+  SaveError(errptr, db->rep->Put(options->rep, column_family->rep,
+                                 Slice(key, keylen), Slice(val, vallen)));
+}
+
+void rocksdb_delete(rocksdb_t* db, const rocksdb_writeoptions_t* options,
+                    const char* key, size_t keylen, char** errptr) {
+  SaveError(errptr, db->rep->Delete(options->rep, Slice(key, keylen)));
+}
+
+void rocksdb_delete_cf(rocksdb_t* db, const rocksdb_writeoptions_t* options,
+                       rocksdb_column_family_handle_t* column_family,
+                       const char* key, size_t keylen, char** errptr) {
+  SaveError(errptr, db->rep->Delete(options->rep, column_family->rep,
+                                    Slice(key, keylen)));
+}
+
+void rocksdb_merge(rocksdb_t* db, const rocksdb_writeoptions_t* options,
+                   const char* key, size_t keylen, const char* val,
+                   size_t vallen, char** errptr) {
+  SaveError(errptr, db->rep->Merge(options->rep, Slice(key, keylen),
+                                   Slice(val, vallen)));
+}
+
+void rocksdb_merge_cf(rocksdb_t* db, const rocksdb_writeoptions_t* options,
+                      rocksdb_column_family_handle_t* column_family,
+                      const char* key, size_t keylen, const char* val,
+                      size_t vallen, char** errptr) {
+  SaveError(errptr, db->rep->Merge(options->rep, column_family->rep,
+                                   Slice(key, keylen), Slice(val, vallen)));
+}
+
+void rocksdb_write(rocksdb_t* db, const rocksdb_writeoptions_t* options,
+                   rocksdb_writebatch_t* batch, char** errptr) {
+  SaveError(errptr, db->rep->Write(options->rep, &batch->rep));
+}
+
+void rocksdb_put_with_ts(rocksdb_t* db, const rocksdb_writeoptions_t* options,
+                         const char* key, size_t keylen, const char* ts,
+                         size_t tslen, const char* val, size_t vallen,
+                         char** errptr) {
+  SaveError(errptr, db->rep->Put(options->rep, Slice(key, keylen),
+                                 Slice(ts, tslen), Slice(val, vallen)));
+}
+
+void rocksdb_put_cf_with_ts(rocksdb_t* db,
+                            const rocksdb_writeoptions_t* options,
+                            rocksdb_column_family_handle_t* column_family,
+                            const char* key, size_t keylen, const char* ts,
+                            size_t tslen, const char* val, size_t vallen,
+                            char** errptr) {
+  SaveError(errptr,
+            db->rep->Put(options->rep, column_family->rep, Slice(key, keylen),
+                         Slice(ts, tslen), Slice(val, vallen)));
+}
+
+void rocksdb_delete_with_ts(rocksdb_t* db,
+                            const rocksdb_writeoptions_t* options,
+                            const char* key, size_t keylen, const char* ts,
+                            size_t tslen, char** errptr) {
+  SaveError(errptr, db->rep->Delete(options->rep, Slice(key, keylen),
+                                    Slice(ts, tslen)));
+}
+
+void rocksdb_delete_cf_with_ts(rocksdb_t* db,
+                               const rocksdb_writeoptions_t* options,
+                               rocksdb_column_family_handle_t* column_family,
+                               const char* key, size_t keylen, const char* ts,
+                               size_t tslen, char** errptr) {
+  SaveError(errptr, db->rep->Delete(options->rep, column_family->rep,
+                                    Slice(key, keylen), Slice(ts, tslen)));
+}
+
+void rocksdb_singledelete(rocksdb_t* db, const rocksdb_writeoptions_t* options,
+                          const char* key, size_t keylen, char** errptr) {
+  SaveError(errptr, db->rep->SingleDelete(options->rep, Slice(key, keylen)));
+}
+
+void rocksdb_singledelete_cf(rocksdb_t* db,
+                             const rocksdb_writeoptions_t* options,
+                             rocksdb_column_family_handle_t* column_family,
+                             const char* key, size_t keylen, char** errptr) {
+  SaveError(errptr, db->rep->SingleDelete(options->rep, column_family->rep,
+                                          Slice(key, keylen)));
+}
+
+void rocksdb_singledelete_with_ts(rocksdb_t* db,
+                                  const rocksdb_writeoptions_t* options,
+                                  const char* key, size_t keylen,
+                                  const char* ts, size_t tslen, char** errptr) {
+  SaveError(errptr, db->rep->SingleDelete(options->rep, Slice(key, keylen),
+                                          Slice(ts, tslen)));
+}
+
+void rocksdb_singledelete_cf_with_ts(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, const char* ts, size_t tslen, char** errptr) {
+  SaveError(errptr,
+            db->rep->SingleDelete(options->rep, column_family->rep,
+                                  Slice(key, keylen), Slice(ts, tslen)));
+}
+
+void rocksdb_delete_range_cf(rocksdb_t* db,
+                             const rocksdb_writeoptions_t* options,
+                             rocksdb_column_family_handle_t* column_family,
+                             const char* start_key, size_t start_key_len,
+                             const char* end_key, size_t end_key_len,
+                             char** errptr) {
+  SaveError(errptr, db->rep->DeleteRange(options->rep, column_family->rep,
+                                         Slice(start_key, start_key_len),
+                                         Slice(end_key, end_key_len)));
+}
+
+void rocksdb_flush(rocksdb_t* db, const rocksdb_flushoptions_t* options,
+                   char** errptr) {
+  SaveError(errptr, db->rep->Flush(options->rep));
+}
+
+void rocksdb_flush_cf(rocksdb_t* db, const rocksdb_flushoptions_t* options,
+                      rocksdb_column_family_handle_t* column_family,
+                      char** errptr) {
+  SaveError(errptr, db->rep->Flush(options->rep, column_family->rep));
+}
+
+void rocksdb_flush_wal(rocksdb_t* db, unsigned char sync, char** errptr) {
+  SaveError(errptr, db->rep->FlushWAL(sync));
+}
+
+void rocksdb_pause_background_work(rocksdb_t* db, char** errptr) {
+  SaveError(errptr, db->rep->PauseBackgroundWork());
+}
+
+void rocksdb_continue_background_work(rocksdb_t* db, char** errptr) {
+  SaveError(errptr, db->rep->ContinueBackgroundWork());
+}
+
+void rocksdb_disable_file_deletions(rocksdb_t* db, char** errptr) {
+  SaveError(errptr, db->rep->DisableFileDeletions());
+}
+
+void rocksdb_enable_file_deletions(rocksdb_t* db, char** errptr) {
+  SaveError(errptr, db->rep->EnableFileDeletions());
+}
+
+void rocksdb_verify_checksum(rocksdb_t* db, char** errptr) {
+  SaveError(errptr, db->rep->VerifyChecksum());
+}
+
+void rocksdb_verify_file_checksums(rocksdb_t* db, char** errptr) {
+  SaveError(errptr, db->rep->VerifyFileChecksums(ReadOptions()));
+}
+
+void rocksdb_destroy_db(const rocksdb_options_t* options, const char* name,
+                        char** errptr) {
+  SaveError(errptr, DestroyDB(name, options->rep));
+}
+
+void rocksdb_repair_db(const rocksdb_options_t* options, const char* name,
+                       char** errptr) {
+  SaveError(errptr, RepairDB(name, options->rep));
+}
+
+/* WriteBatch */
+
+void rocksdb_writebatch_clear(rocksdb_writebatch_t* b) { b->rep.Clear(); }
+
+void rocksdb_writebatch_put(rocksdb_writebatch_t* b, const char* key,
+                            size_t klen, const char* val, size_t vlen) {
+  b->rep.Put(Slice(key, klen), Slice(val, vlen));
+}
+
+void rocksdb_writebatch_put_cf(rocksdb_writebatch_t* b,
+                               rocksdb_column_family_handle_t* column_family,
+                               const char* key, size_t klen, const char* val,
+                               size_t vlen) {
+  b->rep.Put(column_family->rep, Slice(key, klen), Slice(val, vlen));
+}
+
+void rocksdb_writebatch_delete(rocksdb_writebatch_t* b, const char* key,
+                               size_t klen) {
+  b->rep.Delete(Slice(key, klen));
+}
+
+void rocksdb_writebatch_put_log_data(rocksdb_writebatch_t* b, const char* blob,
+                                     size_t len) {
+  b->rep.PutLogData(Slice(blob, len));
+}
+
+void rocksdb_writebatch_set_save_point(rocksdb_writebatch_t* b) {
+  b->rep.SetSavePoint();
+}
+
+void rocksdb_writebatch_rollback_to_save_point(rocksdb_writebatch_t* b,
+                                               char** errptr) {
+  SaveError(errptr, b->rep.RollbackToSavePoint());
+}
+
+void rocksdb_writebatch_pop_save_point(rocksdb_writebatch_t* b, char** errptr) {
+  SaveError(errptr, b->rep.PopSavePoint());
+}
+
+void rocksdb_writebatch_verify_checksum(rocksdb_writebatch_t* b,
+                                        char** errptr) {
+  SaveError(errptr, b->rep.VerifyChecksum());
+}
+
+/* Transaction */
+
+void rocksdb_transaction_put(rocksdb_transaction_t* txn, const char* key,
+                             size_t klen, const char* val, size_t vlen,
+                             char** errptr) {
+  SaveError(errptr, txn->rep->Put(Slice(key, klen), Slice(val, vlen)));
+}
+
+void rocksdb_transaction_put_cf(rocksdb_transaction_t* txn,
+                                rocksdb_column_family_handle_t* column_family,
+                                const char* key, size_t klen, const char* val,
+                                size_t vlen, char** errptr) {
+  SaveError(errptr, txn->rep->Put(column_family->rep, Slice(key, klen),
+                                  Slice(val, vlen)));
+}
+
+void rocksdb_transaction_merge(rocksdb_transaction_t* txn, const char* key,
+                               size_t klen, const char* val, size_t vlen,
+                               char** errptr) {
+  SaveError(errptr, txn->rep->Merge(Slice(key, klen), Slice(val, vlen)));
+}
+
+void rocksdb_transaction_merge_cf(rocksdb_transaction_t* txn,
+                                  rocksdb_column_family_handle_t* column_family,
+                                  const char* key, size_t klen, const char* val,
+                                  size_t vlen, char** errptr) {
+  SaveError(errptr, txn->rep->Merge(column_family->rep, Slice(key, klen),
+                                    Slice(val, vlen)));
+}
+
+void rocksdb_transaction_delete(rocksdb_transaction_t* txn, const char* key,
+                                size_t klen, char** errptr) {
+  SaveError(errptr, txn->rep->Delete(Slice(key, klen)));
+}
+
+void rocksdb_transaction_delete_cf(
+    rocksdb_transaction_t* txn, rocksdb_column_family_handle_t* column_family,
+    const char* key, size_t klen, char** errptr) {
+  SaveError(errptr, txn->rep->Delete(column_family->rep, Slice(key, klen)));
+}
+
+void rocksdb_transaction_commit(rocksdb_transaction_t* txn, char** errptr) {
+  SaveError(errptr, txn->rep->Commit());
+}
+
+void rocksdb_transaction_rollback(rocksdb_transaction_t* txn, char** errptr) {
+  SaveError(errptr, txn->rep->Rollback());
+}
+
+void rocksdb_transaction_put_log_data(rocksdb_transaction_t* txn,
+                                      const char* blob, size_t len) {
+  txn->rep->PutLogData(Slice(blob, len));
+}
+
+void rocksdb_transaction_set_savepoint(rocksdb_transaction_t* txn) {
+  txn->rep->SetSavePoint();
+}
+
+void rocksdb_transaction_rollback_to_savepoint(rocksdb_transaction_t* txn,
+                                               char** errptr) {
+  SaveError(errptr, txn->rep->RollbackToSavePoint());
+}
+
+/* TransactionDB simple */
+
+void rocksdb_transactiondb_put(rocksdb_transactiondb_t* txn_db,
+                               const rocksdb_writeoptions_t* options,
+                               const char* key, size_t klen, const char* val,
+                               size_t vlen, char** errptr) {
+  SaveError(errptr,
+            txn_db->rep->Put(options->rep, Slice(key, klen), Slice(val, vlen)));
+}
+
+void rocksdb_transactiondb_put_cf(rocksdb_transactiondb_t* txn_db,
+                                  const rocksdb_writeoptions_t* options,
+                                  rocksdb_column_family_handle_t* column_family,
+                                  const char* key, size_t keylen,
+                                  const char* val, size_t vallen,
+                                  char** errptr) {
+  SaveError(errptr, txn_db->rep->Put(options->rep, column_family->rep,
+                                     Slice(key, keylen), Slice(val, vallen)));
+}
+
+void rocksdb_transactiondb_write(rocksdb_transactiondb_t* db,
+                                 const rocksdb_writeoptions_t* options,
+                                 rocksdb_writebatch_t* batch, char** errptr) {
+  SaveError(errptr, db->rep->Write(options->rep, &batch->rep));
+}
+
+void rocksdb_transactiondb_merge(rocksdb_transactiondb_t* txn_db,
+                                 const rocksdb_writeoptions_t* options,
+                                 const char* key, size_t klen, const char* val,
+                                 size_t vlen, char** errptr) {
+  SaveError(errptr, txn_db->rep->Merge(options->rep, Slice(key, klen),
+                                       Slice(val, vlen)));
+}
+
+void rocksdb_transactiondb_merge_cf(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key, size_t klen,
+    const char* val, size_t vlen, char** errptr) {
+  SaveError(errptr, txn_db->rep->Merge(options->rep, column_family->rep,
+                                       Slice(key, klen), Slice(val, vlen)));
+}
+
+void rocksdb_transactiondb_delete(rocksdb_transactiondb_t* txn_db,
+                                  const rocksdb_writeoptions_t* options,
+                                  const char* key, size_t klen, char** errptr) {
+  SaveError(errptr, txn_db->rep->Delete(options->rep, Slice(key, klen)));
+}
+
+void rocksdb_transactiondb_delete_cf(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, char** errptr) {
+  SaveError(errptr, txn_db->rep->Delete(options->rep, column_family->rep,
+                                        Slice(key, keylen)));
+}
+
+void rocksdb_transactiondb_flush_wal(rocksdb_transactiondb_t* txn_db,
+                                     unsigned char sync, char** errptr) {
+  SaveError(errptr, txn_db->rep->FlushWAL(sync));
+}
+
+void rocksdb_transactiondb_flush(rocksdb_transactiondb_t* txn_db,
+                                 const rocksdb_flushoptions_t* options,
+                                 char** errptr) {
+  SaveError(errptr, txn_db->rep->Flush(options->rep));
+}
+
+void rocksdb_transactiondb_flush_cf(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_flushoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, char** errptr) {
+  SaveError(errptr, txn_db->rep->Flush(options->rep, column_family->rep));
+}

c_api_gen/c_generated_subset.h.inc

@@ -0,0 +1,249 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+// @generated
+// -----------------------------------------------------------------------------
+// Auto-generated by tools/c_api_gen/generate_c_api.py.
+// DO NOT EDIT THIS FILE DIRECTLY.
+// Source: tools/c_api_gen/spec.json
+// -----------------------------------------------------------------------------
+
+// To regenerate this file:
+//   python3 tools/c_api_gen/generate_c_api.py
+//     --spec tools/c_api_gen/spec.json --section 'DB data operations'
+//     --section 'WriteBatch' --section 'Transaction'
+//     --section 'TransactionDB simple'
+//     --header-out
+//     c_api_gen/c_generated_subset.h.inc
+//     --source-out
+//     c_api_gen/c_generated_subset.cc.inc
+
+/* DB data operations */
+
+extern ROCKSDB_LIBRARY_API void rocksdb_put(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options, const char* key,
+    size_t keylen, const char* val, size_t vallen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_put_cf(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, const char* val, size_t vallen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_delete(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options, const char* key,
+    size_t keylen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_delete_cf(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_merge(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options, const char* key,
+    size_t keylen, const char* val, size_t vallen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_merge_cf(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, const char* val, size_t vallen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_write(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options,
+    rocksdb_writebatch_t* batch, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_put_with_ts(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options, const char* key,
+    size_t keylen, const char* ts, size_t tslen, const char* val, size_t vallen,
+    char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_put_cf_with_ts(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, const char* ts, size_t tslen, const char* val, size_t vallen,
+    char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_delete_with_ts(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options, const char* key,
+    size_t keylen, const char* ts, size_t tslen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_delete_cf_with_ts(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, const char* ts, size_t tslen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_singledelete(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options, const char* key,
+    size_t keylen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_singledelete_cf(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_singledelete_with_ts(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options, const char* key,
+    size_t keylen, const char* ts, size_t tslen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_singledelete_cf_with_ts(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, const char* ts, size_t tslen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_delete_range_cf(
+    rocksdb_t* db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* start_key,
+    size_t start_key_len, const char* end_key, size_t end_key_len,
+    char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_flush(
+    rocksdb_t* db, const rocksdb_flushoptions_t* options, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_flush_cf(
+    rocksdb_t* db, const rocksdb_flushoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_flush_wal(rocksdb_t* db,
+                                                  unsigned char sync,
+                                                  char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_pause_background_work(rocksdb_t* db,
+                                                              char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_continue_background_work(rocksdb_t* db,
+                                                                 char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_disable_file_deletions(rocksdb_t* db,
+                                                               char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_enable_file_deletions(rocksdb_t* db,
+                                                              char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_verify_checksum(rocksdb_t* db,
+                                                        char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_verify_file_checksums(rocksdb_t* db,
+                                                              char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_destroy_db(
+    const rocksdb_options_t* options, const char* name, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_repair_db(
+    const rocksdb_options_t* options, const char* name, char** errptr);
+
+/* WriteBatch */
+
+extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_clear(
+    rocksdb_writebatch_t* b);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_put(rocksdb_writebatch_t* b,
+                                                       const char* key,
+                                                       size_t klen,
+                                                       const char* val,
+                                                       size_t vlen);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_put_cf(
+    rocksdb_writebatch_t* b, rocksdb_column_family_handle_t* column_family,
+    const char* key, size_t klen, const char* val, size_t vlen);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_delete(
+    rocksdb_writebatch_t* b, const char* key, size_t klen);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_put_log_data(
+    rocksdb_writebatch_t* b, const char* blob, size_t len);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_set_save_point(
+    rocksdb_writebatch_t* b);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_rollback_to_save_point(
+    rocksdb_writebatch_t* b, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_pop_save_point(
+    rocksdb_writebatch_t* b, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_verify_checksum(
+    rocksdb_writebatch_t* b, char** errptr);
+
+/* Transaction */
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transaction_put(
+    rocksdb_transaction_t* txn, const char* key, size_t klen, const char* val,
+    size_t vlen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transaction_put_cf(
+    rocksdb_transaction_t* txn, rocksdb_column_family_handle_t* column_family,
+    const char* key, size_t klen, const char* val, size_t vlen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transaction_merge(
+    rocksdb_transaction_t* txn, const char* key, size_t klen, const char* val,
+    size_t vlen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transaction_merge_cf(
+    rocksdb_transaction_t* txn, rocksdb_column_family_handle_t* column_family,
+    const char* key, size_t klen, const char* val, size_t vlen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transaction_delete(
+    rocksdb_transaction_t* txn, const char* key, size_t klen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transaction_delete_cf(
+    rocksdb_transaction_t* txn, rocksdb_column_family_handle_t* column_family,
+    const char* key, size_t klen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transaction_commit(
+    rocksdb_transaction_t* txn, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transaction_rollback(
+    rocksdb_transaction_t* txn, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transaction_put_log_data(
+    rocksdb_transaction_t* txn, const char* blob, size_t len);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transaction_set_savepoint(
+    rocksdb_transaction_t* txn);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transaction_rollback_to_savepoint(
+    rocksdb_transaction_t* txn, char** errptr);
+
+/* TransactionDB simple */
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transactiondb_put(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_writeoptions_t* options,
+    const char* key, size_t klen, const char* val, size_t vlen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transactiondb_put_cf(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, const char* val, size_t vallen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transactiondb_write(
+    rocksdb_transactiondb_t* db, const rocksdb_writeoptions_t* options,
+    rocksdb_writebatch_t* batch, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transactiondb_merge(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_writeoptions_t* options,
+    const char* key, size_t klen, const char* val, size_t vlen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transactiondb_merge_cf(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key, size_t klen,
+    const char* val, size_t vlen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transactiondb_delete(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_writeoptions_t* options,
+    const char* key, size_t klen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transactiondb_delete_cf(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transactiondb_flush_wal(
+    rocksdb_transactiondb_t* txn_db, unsigned char sync, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transactiondb_flush(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_flushoptions_t* options,
+    char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transactiondb_flush_cf(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_flushoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, char** errptr);

c_api_gen/c_generated_transaction_subset.cc.inc

@@ -0,0 +1,81 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+// @generated
+// -----------------------------------------------------------------------------
+// Auto-generated by tools/c_api_gen/generate_c_api.py.
+// DO NOT EDIT THIS FILE DIRECTLY.
+// Source: tools/c_api_gen/spec.json
+// -----------------------------------------------------------------------------
+
+// To regenerate this file:
+//   python3 tools/c_api_gen/generate_c_api.py
+//     --spec tools/c_api_gen/spec.json --section 'Transaction'
+//     --header-out
+//     c_api_gen/c_generated_transaction_subset.h.inc
+//     --source-out
+//     c_api_gen/c_generated_transaction_subset.cc.inc
+
+/* Transaction */
+
+void rocksdb_transaction_put(rocksdb_transaction_t* txn, const char* key,
+                             size_t klen, const char* val, size_t vlen,
+                             char** errptr) {
+  SaveError(errptr, txn->rep->Put(Slice(key, klen), Slice(val, vlen)));
+}
+
+void rocksdb_transaction_put_cf(rocksdb_transaction_t* txn,
+                                rocksdb_column_family_handle_t* column_family,
+                                const char* key, size_t klen, const char* val,
+                                size_t vlen, char** errptr) {
+  SaveError(errptr, txn->rep->Put(column_family->rep, Slice(key, klen),
+                                  Slice(val, vlen)));
+}
+
+void rocksdb_transaction_merge(rocksdb_transaction_t* txn, const char* key,
+                               size_t klen, const char* val, size_t vlen,
+                               char** errptr) {
+  SaveError(errptr, txn->rep->Merge(Slice(key, klen), Slice(val, vlen)));
+}
+
+void rocksdb_transaction_merge_cf(rocksdb_transaction_t* txn,
+                                  rocksdb_column_family_handle_t* column_family,
+                                  const char* key, size_t klen, const char* val,
+                                  size_t vlen, char** errptr) {
+  SaveError(errptr, txn->rep->Merge(column_family->rep, Slice(key, klen),
+                                    Slice(val, vlen)));
+}
+
+void rocksdb_transaction_delete(rocksdb_transaction_t* txn, const char* key,
+                                size_t klen, char** errptr) {
+  SaveError(errptr, txn->rep->Delete(Slice(key, klen)));
+}
+
+void rocksdb_transaction_delete_cf(
+    rocksdb_transaction_t* txn, rocksdb_column_family_handle_t* column_family,
+    const char* key, size_t klen, char** errptr) {
+  SaveError(errptr, txn->rep->Delete(column_family->rep, Slice(key, klen)));
+}
+
+void rocksdb_transaction_commit(rocksdb_transaction_t* txn, char** errptr) {
+  SaveError(errptr, txn->rep->Commit());
+}
+
+void rocksdb_transaction_rollback(rocksdb_transaction_t* txn, char** errptr) {
+  SaveError(errptr, txn->rep->Rollback());
+}
+
+void rocksdb_transaction_put_log_data(rocksdb_transaction_t* txn,
+                                      const char* blob, size_t len) {
+  txn->rep->PutLogData(Slice(blob, len));
+}
+
+void rocksdb_transaction_set_savepoint(rocksdb_transaction_t* txn) {
+  txn->rep->SetSavePoint();
+}
+
+void rocksdb_transaction_rollback_to_savepoint(rocksdb_transaction_t* txn,
+                                               char** errptr) {
+  SaveError(errptr, txn->rep->RollbackToSavePoint());
+}

c_api_gen/c_generated_transaction_subset.h.inc

@@ -0,0 +1,58 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+// @generated
+// -----------------------------------------------------------------------------
+// Auto-generated by tools/c_api_gen/generate_c_api.py.
+// DO NOT EDIT THIS FILE DIRECTLY.
+// Source: tools/c_api_gen/spec.json
+// -----------------------------------------------------------------------------
+
+// To regenerate this file:
+//   python3 tools/c_api_gen/generate_c_api.py
+//     --spec tools/c_api_gen/spec.json --section 'Transaction'
+//     --header-out
+//     c_api_gen/c_generated_transaction_subset.h.inc
+//     --source-out
+//     c_api_gen/c_generated_transaction_subset.cc.inc
+
+/* Transaction */
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transaction_put(
+    rocksdb_transaction_t* txn, const char* key, size_t klen, const char* val,
+    size_t vlen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transaction_put_cf(
+    rocksdb_transaction_t* txn, rocksdb_column_family_handle_t* column_family,
+    const char* key, size_t klen, const char* val, size_t vlen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transaction_merge(
+    rocksdb_transaction_t* txn, const char* key, size_t klen, const char* val,
+    size_t vlen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transaction_merge_cf(
+    rocksdb_transaction_t* txn, rocksdb_column_family_handle_t* column_family,
+    const char* key, size_t klen, const char* val, size_t vlen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transaction_delete(
+    rocksdb_transaction_t* txn, const char* key, size_t klen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transaction_delete_cf(
+    rocksdb_transaction_t* txn, rocksdb_column_family_handle_t* column_family,
+    const char* key, size_t klen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transaction_commit(
+    rocksdb_transaction_t* txn, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transaction_rollback(
+    rocksdb_transaction_t* txn, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transaction_put_log_data(
+    rocksdb_transaction_t* txn, const char* blob, size_t len);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transaction_set_savepoint(
+    rocksdb_transaction_t* txn);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transaction_rollback_to_savepoint(
+    rocksdb_transaction_t* txn, char** errptr);

c_api_gen/c_generated_transactiondb_subset.cc.inc

@@ -0,0 +1,91 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+// @generated
+// -----------------------------------------------------------------------------
+// Auto-generated by tools/c_api_gen/generate_c_api.py.
+// DO NOT EDIT THIS FILE DIRECTLY.
+// Source: tools/c_api_gen/spec.json
+// -----------------------------------------------------------------------------
+
+// To regenerate this file:
+//   python3 tools/c_api_gen/generate_c_api.py
+//     --spec tools/c_api_gen/spec.json --section 'TransactionDB simple'
+//     --header-out
+//     c_api_gen/c_generated_transactiondb_subset.h.inc
+//     --source-out
+//     c_api_gen/c_generated_transactiondb_subset.cc.inc
+
+/* TransactionDB simple */
+
+void rocksdb_transactiondb_put(rocksdb_transactiondb_t* txn_db,
+                               const rocksdb_writeoptions_t* options,
+                               const char* key, size_t klen, const char* val,
+                               size_t vlen, char** errptr) {
+  SaveError(errptr,
+            txn_db->rep->Put(options->rep, Slice(key, klen), Slice(val, vlen)));
+}
+
+void rocksdb_transactiondb_put_cf(rocksdb_transactiondb_t* txn_db,
+                                  const rocksdb_writeoptions_t* options,
+                                  rocksdb_column_family_handle_t* column_family,
+                                  const char* key, size_t keylen,
+                                  const char* val, size_t vallen,
+                                  char** errptr) {
+  SaveError(errptr, txn_db->rep->Put(options->rep, column_family->rep,
+                                     Slice(key, keylen), Slice(val, vallen)));
+}
+
+void rocksdb_transactiondb_write(rocksdb_transactiondb_t* db,
+                                 const rocksdb_writeoptions_t* options,
+                                 rocksdb_writebatch_t* batch, char** errptr) {
+  SaveError(errptr, db->rep->Write(options->rep, &batch->rep));
+}
+
+void rocksdb_transactiondb_merge(rocksdb_transactiondb_t* txn_db,
+                                 const rocksdb_writeoptions_t* options,
+                                 const char* key, size_t klen, const char* val,
+                                 size_t vlen, char** errptr) {
+  SaveError(errptr, txn_db->rep->Merge(options->rep, Slice(key, klen),
+                                       Slice(val, vlen)));
+}
+
+void rocksdb_transactiondb_merge_cf(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key, size_t klen,
+    const char* val, size_t vlen, char** errptr) {
+  SaveError(errptr, txn_db->rep->Merge(options->rep, column_family->rep,
+                                       Slice(key, klen), Slice(val, vlen)));
+}
+
+void rocksdb_transactiondb_delete(rocksdb_transactiondb_t* txn_db,
+                                  const rocksdb_writeoptions_t* options,
+                                  const char* key, size_t klen, char** errptr) {
+  SaveError(errptr, txn_db->rep->Delete(options->rep, Slice(key, klen)));
+}
+
+void rocksdb_transactiondb_delete_cf(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, char** errptr) {
+  SaveError(errptr, txn_db->rep->Delete(options->rep, column_family->rep,
+                                        Slice(key, keylen)));
+}
+
+void rocksdb_transactiondb_flush_wal(rocksdb_transactiondb_t* txn_db,
+                                     unsigned char sync, char** errptr) {
+  SaveError(errptr, txn_db->rep->FlushWAL(sync));
+}
+
+void rocksdb_transactiondb_flush(rocksdb_transactiondb_t* txn_db,
+                                 const rocksdb_flushoptions_t* options,
+                                 char** errptr) {
+  SaveError(errptr, txn_db->rep->Flush(options->rep));
+}
+
+void rocksdb_transactiondb_flush_cf(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_flushoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, char** errptr) {
+  SaveError(errptr, txn_db->rep->Flush(options->rep, column_family->rep));
+}

c_api_gen/c_generated_transactiondb_subset.h.inc

@@ -0,0 +1,62 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+// @generated
+// -----------------------------------------------------------------------------
+// Auto-generated by tools/c_api_gen/generate_c_api.py.
+// DO NOT EDIT THIS FILE DIRECTLY.
+// Source: tools/c_api_gen/spec.json
+// -----------------------------------------------------------------------------
+
+// To regenerate this file:
+//   python3 tools/c_api_gen/generate_c_api.py
+//     --spec tools/c_api_gen/spec.json --section 'TransactionDB simple'
+//     --header-out
+//     c_api_gen/c_generated_transactiondb_subset.h.inc
+//     --source-out
+//     c_api_gen/c_generated_transactiondb_subset.cc.inc
+
+/* TransactionDB simple */
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transactiondb_put(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_writeoptions_t* options,
+    const char* key, size_t klen, const char* val, size_t vlen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transactiondb_put_cf(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, const char* val, size_t vallen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transactiondb_write(
+    rocksdb_transactiondb_t* db, const rocksdb_writeoptions_t* options,
+    rocksdb_writebatch_t* batch, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transactiondb_merge(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_writeoptions_t* options,
+    const char* key, size_t klen, const char* val, size_t vlen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transactiondb_merge_cf(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key, size_t klen,
+    const char* val, size_t vlen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transactiondb_delete(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_writeoptions_t* options,
+    const char* key, size_t klen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transactiondb_delete_cf(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transactiondb_flush_wal(
+    rocksdb_transactiondb_t* txn_db, unsigned char sync, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transactiondb_flush(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_flushoptions_t* options,
+    char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_transactiondb_flush_cf(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_flushoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, char** errptr);

c_api_gen/c_generated_writebatch_subset.cc.inc

@@ -0,0 +1,62 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+// @generated
+// -----------------------------------------------------------------------------
+// Auto-generated by tools/c_api_gen/generate_c_api.py.
+// DO NOT EDIT THIS FILE DIRECTLY.
+// Source: tools/c_api_gen/spec.json
+// -----------------------------------------------------------------------------
+
+// To regenerate this file:
+//   python3 tools/c_api_gen/generate_c_api.py
+//     --spec tools/c_api_gen/spec.json --section 'WriteBatch'
+//     --header-out
+//     c_api_gen/c_generated_writebatch_subset.h.inc
+//     --source-out
+//     c_api_gen/c_generated_writebatch_subset.cc.inc
+
+/* WriteBatch */
+
+void rocksdb_writebatch_clear(rocksdb_writebatch_t* b) { b->rep.Clear(); }
+
+void rocksdb_writebatch_put(rocksdb_writebatch_t* b, const char* key,
+                            size_t klen, const char* val, size_t vlen) {
+  b->rep.Put(Slice(key, klen), Slice(val, vlen));
+}
+
+void rocksdb_writebatch_put_cf(rocksdb_writebatch_t* b,
+                               rocksdb_column_family_handle_t* column_family,
+                               const char* key, size_t klen, const char* val,
+                               size_t vlen) {
+  b->rep.Put(column_family->rep, Slice(key, klen), Slice(val, vlen));
+}
+
+void rocksdb_writebatch_delete(rocksdb_writebatch_t* b, const char* key,
+                               size_t klen) {
+  b->rep.Delete(Slice(key, klen));
+}
+
+void rocksdb_writebatch_put_log_data(rocksdb_writebatch_t* b, const char* blob,
+                                     size_t len) {
+  b->rep.PutLogData(Slice(blob, len));
+}
+
+void rocksdb_writebatch_set_save_point(rocksdb_writebatch_t* b) {
+  b->rep.SetSavePoint();
+}
+
+void rocksdb_writebatch_rollback_to_save_point(rocksdb_writebatch_t* b,
+                                               char** errptr) {
+  SaveError(errptr, b->rep.RollbackToSavePoint());
+}
+
+void rocksdb_writebatch_pop_save_point(rocksdb_writebatch_t* b, char** errptr) {
+  SaveError(errptr, b->rep.PopSavePoint());
+}
+
+void rocksdb_writebatch_verify_checksum(rocksdb_writebatch_t* b,
+                                        char** errptr) {
+  SaveError(errptr, b->rep.VerifyChecksum());
+}

c_api_gen/c_generated_writebatch_subset.h.inc

@@ -0,0 +1,51 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+// @generated
+// -----------------------------------------------------------------------------
+// Auto-generated by tools/c_api_gen/generate_c_api.py.
+// DO NOT EDIT THIS FILE DIRECTLY.
+// Source: tools/c_api_gen/spec.json
+// -----------------------------------------------------------------------------
+
+// To regenerate this file:
+//   python3 tools/c_api_gen/generate_c_api.py
+//     --spec tools/c_api_gen/spec.json --section 'WriteBatch'
+//     --header-out
+//     c_api_gen/c_generated_writebatch_subset.h.inc
+//     --source-out
+//     c_api_gen/c_generated_writebatch_subset.cc.inc
+
+/* WriteBatch */
+
+extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_clear(
+    rocksdb_writebatch_t* b);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_put(rocksdb_writebatch_t* b,
+                                                       const char* key,
+                                                       size_t klen,
+                                                       const char* val,
+                                                       size_t vlen);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_put_cf(
+    rocksdb_writebatch_t* b, rocksdb_column_family_handle_t* column_family,
+    const char* key, size_t klen, const char* val, size_t vlen);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_delete(
+    rocksdb_writebatch_t* b, const char* key, size_t klen);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_put_log_data(
+    rocksdb_writebatch_t* b, const char* blob, size_t len);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_set_save_point(
+    rocksdb_writebatch_t* b);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_rollback_to_save_point(
+    rocksdb_writebatch_t* b, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_pop_save_point(
+    rocksdb_writebatch_t* b, char** errptr);
+
+extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_verify_checksum(
+    rocksdb_writebatch_t* b, char** errptr);

cache/cache_key.h

@@ -127,6 +127,24 @@ class OffsetableCacheKey : private CacheKey {
     return CacheKey(file_num_etc64_, offset_etc64_ ^ offset);
   }
 
+  // Construct a CacheKey for a record located at byte `offset` within a file in
+  // which every distinct cached region is at least 5 bytes long. The low two
+  // offset bits are dropped, which is collision-free under that >= 5-byte (>= 4
+  // would suffice) guarantee and lets different kinds of records in the same
+  // file share a single keyspace without colliding.
+  //
+  // This is the canonical cache-key scheme for byte-offset records that carry
+  // the block-based "min 5 byte" guarantee: block-based SST blocks (see
+  // BlockBasedTable::GetCacheKey) and SimpleGen2Blob records (which always
+  // include a >= 5-byte trailer; see db/blob/blob_gen2_format.h). Because both
+  // use this same function, an SST's data blocks and its embedded blob records
+  // never collide even when the block cache and blob cache are the same cache.
+  // Keeping a single implementation here avoids a divergence bug that would
+  // silently alias the two keyspaces.
+  inline CacheKey WithOffsetForMinSizeRecord(uint64_t offset) const {
+    return WithOffset(offset >> 2);
+  }
+
   // The "common prefix" is a shared prefix for all the returned CacheKeys.
   // It is specific to the file but the same for all offsets within the file.
   static constexpr size_t kCommonPrefixSize = 8;

cache/clock_cache.cc

@@ -3108,7 +3108,7 @@ AutoHyperClockTable::HandleImpl* AutoHyperClockTable::Lookup(
   HandleImpl* const arr = array_.Get();
   NextWithShift next_with_shift = arr[home].head_next_with_shift.LoadRelaxed();
   for (size_t i = 0; !next_with_shift.IsEnd() && i < 10; ++i) {
-    HandleImpl* h = &arr[next_with_shift.IsEnd()];
+    HandleImpl* h = &arr[next_with_shift.GetNext()];
     // Attempt cheap key match without acquiring a read ref. This could give a
     // false positive, which is re-checked after acquiring read ref, or false
     // negative, which is re-checked in the full Lookup. Also, this is a

claude_md/add_option.md

@@ -46,7 +46,7 @@ This document provides guidance on how to add new options to RocksDB's public AP
 
 ## Pattern 1: Adding a Standard Column Family Option
 
-Example reference: commit `94e65a2e0b4f817aa4bfa4c96cdf867e7980d7bc` (memtable_veirfy_per_key_checksum_on_seek)
+Example reference: commit `94e65a2e0b4f817aa4bfa4c96cdf867e7980d7bc` (memtable_verify_per_key_checksum_on_seek)
 
 ### Step 1: Define the Option in Public Header
 
@@ -63,7 +63,7 @@ Add the option with documentation in `AdvancedColumnFamilyOptions` struct:
 // operation.
 // This option depends on memtable_protection_bytes_per_key to be non zero.
 // If memtable_protection_bytes_per_key is zero, no validation is performed.
-bool memtable_veirfy_per_key_checksum_on_seek = false;
+bool memtable_verify_per_key_checksum_on_seek = false;
 ```
 
 ### Step 2: Add to Internal Options Structs
@@ -74,14 +74,14 @@ Add to `MutableCFOptions` struct (or `ImmutableCFOptions` for immutable options)
 
 ```cpp
 // In MutableCFOptions constructor from Options:
-memtable_veirfy_per_key_checksum_on_seek(
-    options.memtable_veirfy_per_key_checksum_on_seek),
+memtable_verify_per_key_checksum_on_seek(
+    options.memtable_verify_per_key_checksum_on_seek),
 
 // In MutableCFOptions default constructor:
-memtable_veirfy_per_key_checksum_on_seek(false),
+memtable_verify_per_key_checksum_on_seek(false),
 
 // In MutableCFOptions struct member declarations:
-bool memtable_veirfy_per_key_checksum_on_seek;
+bool memtable_verify_per_key_checksum_on_seek;
 ```
 
 ### Step 3: Register for Serialization/Deserialization
@@ -91,9 +91,9 @@ bool memtable_veirfy_per_key_checksum_on_seek;
 Add to the options type info map for serialization:
 
 ```cpp
-{"memtable_veirfy_per_key_checksum_on_seek",
+{"memtable_verify_per_key_checksum_on_seek",
  {offsetof(struct MutableCFOptions,
-           memtable_veirfy_per_key_checksum_on_seek),
+           memtable_verify_per_key_checksum_on_seek),
   OptionType::kBoolean, OptionVerificationType::kNormal,
   OptionTypeFlags::kMutable}},
 ```
@@ -101,8 +101,8 @@ Add to the options type info map for serialization:
 Add logging in `MutableCFOptions::Dump()`:
 
 ```cpp
-ROCKS_LOG_INFO(log, "memtable_veirfy_per_key_checksum_on_seek: %d",
-               memtable_veirfy_per_key_checksum_on_seek);
+ROCKS_LOG_INFO(log, "memtable_verify_per_key_checksum_on_seek: %d",
+               memtable_verify_per_key_checksum_on_seek);
 ```
 
 ### Step 4: Update Options Helper
@@ -112,8 +112,8 @@ ROCKS_LOG_INFO(log, "memtable_veirfy_per_key_checksum_on_seek: %d",
 Add to `UpdateColumnFamilyOptions()`:
 
 ```cpp
-cf_opts->memtable_veirfy_per_key_checksum_on_seek =
-    moptions.memtable_veirfy_per_key_checksum_on_seek;
+cf_opts->memtable_verify_per_key_checksum_on_seek =
+    moptions.memtable_verify_per_key_checksum_on_seek;
 ```
 
 ### Step 5: Add to Options Settable Test
@@ -123,7 +123,7 @@ cf_opts->memtable_veirfy_per_key_checksum_on_seek =
 Add to the test string in `ColumnFamilyOptionsAllFieldsSettable`:
 
 ```cpp
-"memtable_veirfy_per_key_checksum_on_seek=1;"
+"memtable_verify_per_key_checksum_on_seek=1;"
 ```
 
 ### Step 6: Add db_stress Support
@@ -131,23 +131,23 @@ Add to the test string in `ColumnFamilyOptionsAllFieldsSettable`:
 **File: `db_stress_tool/db_stress_common.h`**
 
 ```cpp
-DECLARE_bool(memtable_veirfy_per_key_checksum_on_seek);
+DECLARE_bool(memtable_verify_per_key_checksum_on_seek);
 ```
 
 **File: `db_stress_tool/db_stress_gflags.cc`**
 
 ```cpp
 DEFINE_bool(
-    memtable_veirfy_per_key_checksum_on_seek,
-    ROCKSDB_NAMESPACE::Options().memtable_veirfy_per_key_checksum_on_seek,
-    "Sets CF option memtable_veirfy_per_key_checksum_on_seek.");
+    memtable_verify_per_key_checksum_on_seek,
+    ROCKSDB_NAMESPACE::Options().memtable_verify_per_key_checksum_on_seek,
+    "Sets CF option memtable_verify_per_key_checksum_on_seek.");
 ```
 
 **File: `db_stress_tool/db_stress_test_base.cc`**
 
 ```cpp
-options.memtable_veirfy_per_key_checksum_on_seek =
-    FLAGS_memtable_veirfy_per_key_checksum_on_seek;
+options.memtable_verify_per_key_checksum_on_seek =
+    FLAGS_memtable_verify_per_key_checksum_on_seek;
 ```
 
 ### Step 7: Add db_bench Support
@@ -156,12 +156,12 @@ options.memtable_veirfy_per_key_checksum_on_seek =
 
 ```cpp
 // Flag definition (near related flags):
-DEFINE_bool(memtable_veirfy_per_key_checksum_on_seek, false,
-            "Sets CF option memtable_veirfy_per_key_checksum_on_seek");
+DEFINE_bool(memtable_verify_per_key_checksum_on_seek, false,
+            "Sets CF option memtable_verify_per_key_checksum_on_seek");
 
 // Apply flag to options (in InitializeOptionsFromFlags or similar):
-options.memtable_veirfy_per_key_checksum_on_seek =
-    FLAGS_memtable_veirfy_per_key_checksum_on_seek;
+options.memtable_verify_per_key_checksum_on_seek =
+    FLAGS_memtable_verify_per_key_checksum_on_seek;
 ```
 
 ### Step 8: Add Crash Test Support
@@ -169,7 +169,7 @@ options.memtable_veirfy_per_key_checksum_on_seek =
 **File: `tools/db_crashtest.py`**
 
 ```python
-"memtable_veirfy_per_key_checksum_on_seek": lambda: random.choice([0] * 7 + [1]),
+"memtable_verify_per_key_checksum_on_seek": lambda: random.choice([0] * 7 + [1]),
 ```
 
 Also add constraint handling in `finalize_and_sanitize()` if needed:
@@ -177,7 +177,7 @@ Also add constraint handling in `finalize_and_sanitize()` if needed:
 ```python
 # only skip list memtable representation supports paranoid memory checks
 if dest_params.get("memtablerep") != "skip_list":
-    dest_params["memtable_veirfy_per_key_checksum_on_seek"] = 0
+    dest_params["memtable_verify_per_key_checksum_on_seek"] = 0
 ```
 
 ### Step 9: Add Release Note
@@ -185,7 +185,7 @@ if dest_params.get("memtablerep") != "skip_list":
 **File: `unreleased_history/new_features/<descriptive_name>.md`**
 
 ```markdown
-A new flag memtable_veirfy_per_key_checksum_on_seek is added to AdvancedColumnFamilyOptions. When it is enabled, it will validate key checksum along the binary search path on skiplist based memtable during seek operation.
+A new flag memtable_verify_per_key_checksum_on_seek is added to AdvancedColumnFamilyOptions. When it is enabled, it will validate key checksum along the binary search path on skiplist based memtable during seek operation.
 ```
 
 ---
@@ -421,6 +421,30 @@ Example reference: commit `429b36c22d76403d275dd0e6877b08d4cea2bc90` (block_alig
 
 If an option already exists but needs C API support:
 
+**First decide whether the option is auto-managed.**
+
+Many simple option fields are no longer maintained by hand in
+`include/rocksdb/c.h` and `db/c.cc`. Before adding a manual wrapper:
+
+- Check whether the option belongs to a family handled by
+  `tools/c_api_gen/auto_simple_bindings.py`.
+- If it is a simple scalar/enum/string field in an auto-managed family, run
+  `python3 tools/c_api_gen/regen_all.py` and let the generator add the C API.
+- After regenerating, run
+  `python3 tools/c_api_gen/verify_generated_up_to_date.py` to confirm the
+  checked-in generated fragments stay stable.
+- Do not edit generated `.inc` files under `c_api_gen/` by hand.
+- If the new field is not ready for C API support yet, add an entry to
+  `tools/c_api_gen/auto_simple_bindings_blocklist.json` with
+  `"policy": "deferred"` and a concrete reason so the build stays intentional
+  rather than silently drifting.
+- If the field needs custom ownership, callback, vector, or nested-struct
+  marshalling, keep it manual or move it into `tools/c_api_gen/spec.json`
+  depending on the API shape.
+
+If the option is not auto-managed, or the C API really is a custom/manual case,
+the traditional hand-written pattern still applies:
+
 **File: `db/c.cc`**
 
 ```cpp
@@ -509,4 +533,3 @@ Common option types used in serialization:
 - [ ] unreleased_history markdown file
 - [ ] Java API (for BlockBasedTableOptions)
 - [ ] C API (if needed)
-

claude_md/add_public_api.md

@@ -168,9 +168,18 @@ Status YourNewAPI(const YourAPIOptions& /*options*/,
 
 ### Step 5: Add C API Bindings
 
-**Header:** `include/rocksdb/c.h`
+> **Important:** `include/rocksdb/c.h` and `db/c.cc` are `@generated` and must
+> NOT be edited by hand — your changes would be overwritten on the next
+> regeneration. Choose the right path first (see "Before writing C API code by
+> hand" below). For hand-written (manual) wrappers, edit the source templates
+> `tools/c_api_gen/c_base.h` (declarations) and `tools/c_api_gen/c_base.cc`
+> (implementations), then run `python3 tools/c_api_gen/regen_all.py` to produce
+> `c.h` / `c.cc`. The snippets below show the *declaration* and *implementation*
+> you would add to those templates.
 
-\`\`\`c
+**Header (declaration → `tools/c_api_gen/c_base.h`):**
+
+```c
 // Basic version
 extern ROCKSDB_LIBRARY_API void rocksdb_your_new_api(
     rocksdb_t* db,
@@ -191,7 +200,7 @@ extern ROCKSDB_LIBRARY_API void rocksdb_your_new_api_opt(
     char** errptr);
 \`\`\`
 
-**Implementation:** `db/c.cc`
+**Implementation (→ `tools/c_api_gen/c_base.cc`):**
 
 \`\`\`cpp
 void rocksdb_your_new_api(rocksdb_t* db, const char* start_key,
@@ -217,6 +226,38 @@ void rocksdb_your_new_api_cf(rocksdb_t* db,
 }
 \`\`\`
 
+**Before writing C API code by hand, choose the right path:**
+
+- **Auto-managed simple struct fields:** If you added a new field to a managed
+  public struct such as `ReadOptions`, selected option structs, or managed
+  metadata structs, first check `tools/c_api_gen/auto_simple_bindings.py`.
+  Simple scalar/enum/string fields should be regenerated with
+  `python3 tools/c_api_gen/regen_all.py` rather than hand-editing
+  `include/rocksdb/c.h` / `db/c.cc`.
+  After regenerating, run
+  `python3 tools/c_api_gen/verify_generated_up_to_date.py` to confirm the
+  checked-in generated fragments are stable.
+- **Spec-driven wrappers:** If the API is still mechanically generated but needs
+  an explicit C shape, naming, or `Status`/`char** errptr` policy, add it to
+  `tools/c_api_gen/spec.json`.
+- **Fully manual wrappers:** If the API uses callbacks, ownership transfer,
+  vectors/maps, open flows, or otherwise irregular marshalling, keep the C API
+  hand-written.
+
+**Temporary deferral for auto-managed families:**
+
+If a new field lands in an auto-managed family but the C API shape is not ready
+yet, add a checked-in entry to
+`tools/c_api_gen/auto_simple_bindings_blocklist.json` with:
+
+- `policy: "manual"` when the field is intentionally outside simple auto-gen
+- `policy: "deferred"` when the field should be revisited later
+- a concrete `reason`
+- `tracking_issue` when available
+
+If a field in an auto-managed family is not supported by the generator and is
+not covered by the blocklist, regeneration should fail. That is intentional.
+
 **If you have options, also add:**
 
 \`\`\`cpp
@@ -482,8 +523,8 @@ public class YourAPIOptionsTest {
 | StackableDB | `include/rocksdb/utilities/stackable_db.h` | ✓ |
 | Secondary DB | `db/db_impl/db_impl_secondary.h` | If not supported |
 | Compacted DB | `db/db_impl/compacted_db_impl.h` | If not supported |
-| C API Header | `include/rocksdb/c.h` | ✓ |
-| C API Implementation | `db/c.cc` | ✓ |
+| C API Header (manual wrappers) | `tools/c_api_gen/c_base.h` → regen → `include/rocksdb/c.h` (`@generated`) | ✓ |
+| C API Implementation (manual wrappers) | `tools/c_api_gen/c_base.cc` → regen → `db/c.cc` (`@generated`) | ✓ |
 | Java API | `java/src/main/java/org/rocksdb/RocksDB.java` | ✓ |
 | Java Options | `java/src/main/java/org/rocksdb/YourAPIOptions.java` | If needed |
 | JNI Implementation | `java/rocksjni/rocksjni.cc` | ✓ |

cmake/RocksDBConfig.cmake.in

@@ -44,10 +44,6 @@ if(@WITH_NUMA@)
   find_dependency(NUMA)
 endif()
 
-if(@WITH_TBB@)
-  find_dependency(TBB)
-endif()
-
 find_dependency(Threads)
 
 include("${CMAKE_CURRENT_LIST_DIR}/RocksDBTargets.cmake")

cmake/modules/FindTBB.cmake

@@ -1,33 +0,0 @@
-# - Find TBB
-# Find the Thread Building Blocks library and includes
-#
-# TBB_INCLUDE_DIRS - where to find tbb.h, etc.
-# TBB_LIBRARIES - List of libraries when using TBB.
-# TBB_FOUND - True if TBB found.
-
-if(NOT DEFINED TBB_ROOT_DIR)
-  set(TBB_ROOT_DIR "$ENV{TBBROOT}")
-endif()
-
-find_path(TBB_INCLUDE_DIRS
-  NAMES tbb/tbb.h
-  HINTS ${TBB_ROOT_DIR}/include)
-
-find_library(TBB_LIBRARIES
-  NAMES tbb
-  HINTS ${TBB_ROOT_DIR}/lib ENV LIBRARY_PATH)
-
-include(FindPackageHandleStandardArgs)
-find_package_handle_standard_args(TBB DEFAULT_MSG TBB_LIBRARIES TBB_INCLUDE_DIRS)
-
-mark_as_advanced(
-  TBB_LIBRARIES
-  TBB_INCLUDE_DIRS)
-
-if(TBB_FOUND AND NOT (TARGET TBB::TBB))
-  add_library (TBB::TBB UNKNOWN IMPORTED)
-  set_target_properties(TBB::TBB
-    PROPERTIES
-      IMPORTED_LOCATION ${TBB_LIBRARIES}
-      INTERFACE_INCLUDE_DIRECTORIES ${TBB_INCLUDE_DIRS})
-endif()

db/arena_wrapped_db_iter.cc

@@ -30,13 +30,78 @@ inline static SequenceNumber GetSeqNum(const DBImpl* db, const Snapshot* s) {
 Status ArenaWrappedDBIter::GetProperty(std::string prop_name,
                                        std::string* prop) {
   if (prop_name == "rocksdb.iterator.super-version-number") {
+    if (!internal_iter_initialized_) {
+      *prop = std::to_string(sv_number_);
+      return Status::OK();
+    }
     // First try to pass the value returned from inner iterator.
     if (!db_iter_->GetProperty(prop_name, prop).ok()) {
       *prop = std::to_string(sv_number_);
     }
     return Status::OK();
   }
-  return db_iter_->GetProperty(prop_name, prop);
+  if (!EnsureInternalIteratorInitialized(nullptr).ok()) {
+    return db_iter_->status();
+  }
+  return db_iter_->GetProperty(std::move(prop_name), prop);
+}
+
+void ArenaWrappedDBIter::CleanupDeferredSuperVersion() {
+  if (deferred_sv_ != nullptr) {
+    assert(db_impl_ != nullptr);
+    db_impl_->CleanupIteratorSuperVersion(
+        deferred_sv_, read_options_.background_purge_on_iterator_cleanup);
+    deferred_sv_ = nullptr;
+  }
+}
+
+void ArenaWrappedDBIter::DestroyDBIter() {
+  db_iter_->~DBIter();
+  CleanupDeferredSuperVersion();
+}
+
+void ArenaWrappedDBIter::ColumnFamilyDataUnrefDeleter::operator()(
+    ColumnFamilyData* cfd) const {
+  if (cfd == nullptr) {
+    return;
+  }
+  assert(db_impl != nullptr);
+
+  InstrumentedMutexLock lock(db_impl->mutex());
+  cfd->UnrefAndTryDelete();
+}
+
+void ArenaWrappedDBIter::DestroyDBIterAndArena() {
+  DestroyDBIter();
+  arena_.~Arena();
+}
+
+Status ArenaWrappedDBIter::EnsureInternalIteratorInitialized(
+    const MultiScanArgs* scan_opts) {
+  if (internal_iter_initialized_) {
+    return Status::OK();
+  }
+  if (db_impl_ == nullptr || deferred_cfd_ == nullptr ||
+      deferred_sv_ == nullptr) {
+    Status s = Status::InvalidArgument(
+        "Internal iterator cannot be initialized without deferred DB state");
+    db_iter_->set_status(s);
+    db_iter_->set_valid(false);
+    return s;
+  }
+
+  const MultiScanArgs* pruning_scan_opts =
+      scan_opts != nullptr && scan_opts->HasBoundedScanRanges() ? scan_opts
+                                                                : nullptr;
+  child_read_options_ = read_options_;
+  child_read_options_.snapshot = nullptr;
+  InternalIterator* internal_iter = db_impl_->NewInternalIterator(
+      child_read_options_, deferred_cfd_, deferred_sv_, &arena_, sequence_,
+      /*allow_unprepared_value=*/true, this, pruning_scan_opts);
+  deferred_cfd_ = nullptr;
+  deferred_sv_ = nullptr;
+  SetIterUnderDBIterImpl(internal_iter);
+  return Status::OK();
 }
 
 void ArenaWrappedDBIter::Init(
@@ -44,18 +109,26 @@ void ArenaWrappedDBIter::Init(
     const MutableCFOptions& mutable_cf_options, const Version* version,
     const SequenceNumber& sequence, uint64_t version_number,
     ReadCallback* read_callback, ColumnFamilyHandleImpl* cfh,
-    bool expose_blob_index, bool allow_refresh, ReadOnlyMemTable* active_mem) {
+    bool expose_blob_index, bool allow_refresh, ReadOnlyMemTable* active_mem,
+    DBImpl* db_impl, ColumnFamilyData* cfd) {
   read_options_ = read_options;
+  child_read_options_ = read_options;
   if (!CheckFSFeatureSupport(env->GetFileSystem().get(),
                              FSSupportedOps::kAsyncIO)) {
     read_options_.async_io = false;
   }
   read_options_.total_order_seek |= ioptions.prefix_seek_opt_in_only;
 
-  db_iter_ = DBIter::NewIter(
-      env, read_options_, ioptions, mutable_cf_options,
-      ioptions.user_comparator, /*internal_iter=*/nullptr, version, sequence,
-      read_callback, active_mem, cfh, expose_blob_index, &arena_);
+  if (cfh != nullptr) {
+    db_impl = cfh->db();
+    cfd = cfh->cfd();
+  }
+
+  db_iter_ = DBIter::NewIter(env, read_options_, ioptions, mutable_cf_options,
+                             ioptions.user_comparator,
+                             /*internal_iter=*/nullptr, version, sequence,
+                             read_callback, active_mem, /*cfh=*/nullptr,
+                             expose_blob_index, &arena_, db_impl, cfd);
 
   sv_number_ = version_number;
   allow_refresh_ = allow_refresh;
@@ -65,13 +138,13 @@ void ArenaWrappedDBIter::Init(
 
 void ArenaWrappedDBIter::MaybeAutoRefresh(bool is_seek,
                                           DBIter::Direction direction) {
-  if (cfh_ != nullptr && read_options_.snapshot != nullptr && allow_refresh_ &&
-      read_options_.auto_refresh_iterator_with_snapshot) {
+  if (cfd_ref_ != nullptr && read_options_.snapshot != nullptr &&
+      allow_refresh_ && read_options_.auto_refresh_iterator_with_snapshot) {
     // The intent here is to capture the superversion number change
     // reasonably soon from the time it actually happened. As such,
     // we're fine with weaker synchronization / ordering guarantees
     // provided by relaxed atomic (in favor of less CPU / mem overhead).
-    uint64_t cur_sv_number = cfh_->cfd()->GetSuperVersionNumberRelaxed();
+    uint64_t cur_sv_number = cfd_ref_->GetSuperVersionNumberRelaxed();
     if ((sv_number_ != cur_sv_number) && status().ok()) {
       // Changing iterators' direction is pretty heavy-weight operation and
       // could have unintended consequences when it comes to prefix seek.
@@ -150,13 +223,13 @@ void ArenaWrappedDBIter::DoRefresh(const Snapshot* snapshot,
   // present in the error log, but won't be reflected in the iterator status.
   // This is by design as we expect compaction to clean up those obsolete files
   // eventually.
-  db_iter_->~DBIter();
-
-  arena_.~Arena();
+  DestroyDBIterAndArena();
   new (&arena_) Arena();
 
-  auto cfd = cfh_->cfd();
-  auto db_impl = cfh_->db();
+  auto cfd = cfd_ref_.get();
+  auto db_impl = db_impl_;
+  assert(cfd != nullptr);
+  assert(db_impl != nullptr);
 
   SuperVersion* sv = cfd->GetReferencedSuperVersion(db_impl);
   assert(sv->version_number >= sv_number);
@@ -164,23 +237,27 @@ void ArenaWrappedDBIter::DoRefresh(const Snapshot* snapshot,
   if (read_callback_) {
     read_callback_->Refresh(read_seq);
   }
+  // TODO: Preserve Prepare() scan options across Refresh() so a refreshed
+  // MultiScan iterator can rebuild the same pruned tree.
   Init(env, read_options_, cfd->ioptions(), sv->mutable_cf_options, sv->current,
-       read_seq, sv->version_number, read_callback_, cfh_, expose_blob_index_,
-       allow_refresh_, allow_mark_memtable_for_flush_ ? sv->mem : nullptr);
+       read_seq, sv->version_number, read_callback_, nullptr,
+       expose_blob_index_, allow_refresh_,
+       allow_mark_memtable_for_flush_ ? sv->mem : nullptr, db_impl, cfd);
 
   InternalIterator* internal_iter = db_impl->NewInternalIterator(
       read_options_, cfd, sv, &arena_, read_seq,
       /* allow_unprepared_value */ true, /* db_iter */ this);
   SetIterUnderDBIter(internal_iter);
+  internal_iter_initialized_ = true;
 }
 
 Status ArenaWrappedDBIter::Refresh(const Snapshot* snapshot) {
-  if (cfh_ == nullptr || !allow_refresh_) {
+  if (cfd_ref_ == nullptr || db_impl_ == nullptr || !allow_refresh_) {
     return Status::NotSupported("Creating renew iterator is not allowed.");
   }
   assert(db_iter_ != nullptr);
-  auto cfd = cfh_->cfd();
-  auto db_impl = cfh_->db();
+  auto cfd = cfd_ref_.get();
+  auto db_impl = db_impl_;
 
   // TODO(yiwu): For last_seq_same_as_publish_seq_==false, this is not the
   // correct behavior. Will be corrected automatically when we take a snapshot
@@ -193,6 +270,13 @@ Status ArenaWrappedDBIter::Refresh(const Snapshot* snapshot) {
   TEST_SYNC_POINT("ArenaWrappedDBIter::Refresh:1");
   TEST_SYNC_POINT("ArenaWrappedDBIter::Refresh:2");
 
+  if (!internal_iter_initialized_) {
+    Status s = EnsureInternalIteratorInitialized(nullptr);
+    if (!s.ok()) {
+      return s;
+    }
+  }
+
   while (true) {
     if (sv_number_ != cur_sv_number) {
       DoRefresh(snapshot, cur_sv_number);
@@ -250,25 +334,47 @@ Status ArenaWrappedDBIter::Refresh(const Snapshot* snapshot) {
   return Status::OK();
 }
 
+void ArenaWrappedDBIter::Prepare(const MultiScanArgs& scan_opts) {
+  if (prepare_called_) {
+    db_iter_->set_status(Status::InvalidArgument(
+        "Prepare called more than once on the same iterator"));
+    db_iter_->set_valid(false);
+    return;
+  }
+  prepare_called_ = true;
+
+  Status s = db_iter_->SetScanOptionsForPrepare(scan_opts);
+  if (!s.ok()) {
+    return;
+  }
+
+  const MultiScanArgs* pruning_scan_opts =
+      scan_opts.HasBoundedScanRanges() ? &scan_opts : nullptr;
+  s = EnsureInternalIteratorInitialized(pruning_scan_opts);
+  if (!s.ok()) {
+    return;
+  }
+
+  db_iter_->PrepareInternalChildren();
+}
+
 ArenaWrappedDBIter* NewArenaWrappedDbIterator(
     Env* env, const ReadOptions& read_options, ColumnFamilyHandleImpl* cfh,
     SuperVersion* sv, const SequenceNumber& sequence,
     ReadCallback* read_callback, DBImpl* db_impl, bool expose_blob_index,
     bool allow_refresh, bool allow_mark_memtable_for_flush) {
   ArenaWrappedDBIter* db_iter = new ArenaWrappedDBIter();
+  ColumnFamilyData* cfd = cfh->cfd();
   db_iter->Init(env, read_options, cfh->cfd()->ioptions(),
                 sv->mutable_cf_options, sv->current, sequence,
                 sv->version_number, read_callback, cfh, expose_blob_index,
                 allow_refresh,
                 allow_mark_memtable_for_flush ? sv->mem : nullptr);
-  if (cfh != nullptr && allow_refresh) {
-    db_iter->StoreRefreshInfo(cfh, read_callback, expose_blob_index);
+  if (allow_refresh && cfd != nullptr) {
+    db_iter->StoreRefreshInfo(read_callback, expose_blob_index);
   }
-
-  InternalIterator* internal_iter = db_impl->NewInternalIterator(
-      db_iter->GetReadOptions(), cfh->cfd(), sv, db_iter->GetArena(), sequence,
-      /*allow_unprepared_value=*/true, db_iter);
-  db_iter->SetIterUnderDBIter(internal_iter);
+  db_iter->StoreDeferredInitInfo(db_impl, cfd, sv, sequence,
+                                 allow_mark_memtable_for_flush);
 
   return db_iter;
 }

db/arena_wrapped_db_iter.h

@@ -10,6 +10,7 @@
 #pragma once
 #include <stdint.h>
 
+#include <memory>
 #include <string>
 
 #include "db/db_impl/db_impl.h"
@@ -33,10 +34,17 @@ class Version;
 // When using the class's Iterator interface, the behavior is exactly
 // the same as the inner DBIter.
 class ArenaWrappedDBIter : public Iterator {
+  struct ColumnFamilyDataUnrefDeleter {
+    DBImpl* db_impl = nullptr;
+    void operator()(ColumnFamilyData* cfd) const;
+  };
+  using ColumnFamilyDataRef =
+      std::unique_ptr<ColumnFamilyData, ColumnFamilyDataUnrefDeleter>;
+
  public:
   ~ArenaWrappedDBIter() override {
     if (db_iter_ != nullptr) {
-      db_iter_->~DBIter();
+      DestroyDBIter();
     } else {
       assert(false);
     }
@@ -51,7 +59,7 @@ class ArenaWrappedDBIter : public Iterator {
   // Set the internal iterator wrapped inside the DB Iterator. Usually it is
   // a merging iterator.
   virtual void SetIterUnderDBIter(InternalIterator* iter) {
-    db_iter_->SetIter(iter);
+    SetIterUnderDBIterImpl(iter);
   }
 
   void SetMemtableRangetombstoneIter(
@@ -60,26 +68,48 @@ class ArenaWrappedDBIter : public Iterator {
   }
 
   bool Valid() const override { return db_iter_->Valid(); }
-  void SeekToFirst() override { db_iter_->SeekToFirst(); }
-  void SeekToLast() override { db_iter_->SeekToLast(); }
+  void SeekToFirst() override {
+    if (!EnsureInternalIteratorInitialized(nullptr).ok()) {
+      return;
+    }
+    db_iter_->SeekToFirst();
+  }
+  void SeekToLast() override {
+    if (!EnsureInternalIteratorInitialized(nullptr).ok()) {
+      return;
+    }
+    db_iter_->SeekToLast();
+  }
   // 'target' does not contain timestamp, even if user timestamp feature is
   // enabled.
   void Seek(const Slice& target) override {
+    if (!EnsureInternalIteratorInitialized(nullptr).ok()) {
+      return;
+    }
     MaybeAutoRefresh(true /* is_seek */, DBIter::kForward);
     db_iter_->Seek(target);
   }
 
   void SeekForPrev(const Slice& target) override {
+    if (!EnsureInternalIteratorInitialized(nullptr).ok()) {
+      return;
+    }
     MaybeAutoRefresh(true /* is_seek */, DBIter::kReverse);
     db_iter_->SeekForPrev(target);
   }
 
   void Next() override {
+    if (!EnsureInternalIteratorInitialized(nullptr).ok()) {
+      return;
+    }
     db_iter_->Next();
     MaybeAutoRefresh(false /* is_seek */, DBIter::kForward);
   }
 
   void Prev() override {
+    if (!EnsureInternalIteratorInitialized(nullptr).ok()) {
+      return;
+    }
     db_iter_->Prev();
     MaybeAutoRefresh(false /* is_seek */, DBIter::kReverse);
   }
@@ -96,12 +126,13 @@ class ArenaWrappedDBIter : public Iterator {
   Status Refresh() override;
   Status Refresh(const Snapshot*) override;
 
-  bool PrepareValue() override { return db_iter_->PrepareValue(); }
-
-  void Prepare(const MultiScanArgs& scan_opts) override {
-    db_iter_->Prepare(scan_opts);
+  bool PrepareValue() override {
+    return EnsureInternalIteratorInitialized(nullptr).ok() &&
+           db_iter_->PrepareValue();
   }
 
+  void Prepare(const MultiScanArgs& scan_opts) override;
+
   // FIXME: we could just pass SV in for mutable cf option, version and version
   // number, but this is used by SstFileReader which does not have a SV.
   void Init(Env* env, const ReadOptions& read_options,
@@ -110,27 +141,53 @@ class ArenaWrappedDBIter : public Iterator {
             const SequenceNumber& sequence, uint64_t version_number,
             ReadCallback* read_callback, ColumnFamilyHandleImpl* cfh,
             bool expose_blob_index, bool allow_refresh,
-            ReadOnlyMemTable* active_mem);
+            ReadOnlyMemTable* active_mem, DBImpl* db_impl = nullptr,
+            ColumnFamilyData* cfd = nullptr);
 
-  // Store some parameters so we can refresh the iterator at a later point
-  // with these same params
-  void StoreRefreshInfo(ColumnFamilyHandleImpl* cfh,
-                        ReadCallback* read_callback, bool expose_blob_index) {
-    cfh_ = cfh;
+  // Store parameters used only by explicit/auto-refresh.
+  void StoreRefreshInfo(ReadCallback* read_callback, bool expose_blob_index) {
     read_callback_ = read_callback;
     expose_blob_index_ = expose_blob_index;
   }
 
+  void StoreDeferredInitInfo(DBImpl* db_impl, ColumnFamilyData* cfd,
+                             SuperVersion* sv, const SequenceNumber& sequence,
+                             bool allow_mark_memtable_for_flush) {
+    assert(cfd != nullptr);
+    db_impl_ = db_impl;
+    cfd->Ref();
+    cfd_ref_ = ColumnFamilyDataRef(cfd, ColumnFamilyDataUnrefDeleter{db_impl});
+    deferred_cfd_ = cfd;
+    deferred_sv_ = sv;
+    sequence_ = sequence;
+    allow_mark_memtable_for_flush_ = allow_mark_memtable_for_flush;
+  }
+
  private:
+  Status EnsureInternalIteratorInitialized(const MultiScanArgs* scan_opts);
+  void SetIterUnderDBIterImpl(InternalIterator* iter) {
+    db_iter_->SetIter(iter);
+    internal_iter_initialized_ = true;
+  }
+  void CleanupDeferredSuperVersion();
+  void DestroyDBIter();
+  void DestroyDBIterAndArena();
   void DoRefresh(const Snapshot* snapshot, uint64_t sv_number);
   void MaybeAutoRefresh(bool is_seek, DBIter::Direction direction);
 
   DBIter* db_iter_ = nullptr;
   Arena arena_;
-  uint64_t sv_number_;
-  ColumnFamilyHandleImpl* cfh_ = nullptr;
+  uint64_t sv_number_ = 0;
+  DBImpl* db_impl_ = nullptr;
+  ColumnFamilyDataRef cfd_ref_{nullptr, ColumnFamilyDataUnrefDeleter{}};
+  ColumnFamilyData* deferred_cfd_ = nullptr;
+  SuperVersion* deferred_sv_ = nullptr;
+  SequenceNumber sequence_ = kMaxSequenceNumber;
+  bool internal_iter_initialized_ = false;
+  bool prepare_called_ = false;
   ReadOptions read_options_;
-  ReadCallback* read_callback_;
+  ReadOptions child_read_options_;
+  ReadCallback* read_callback_ = nullptr;
   bool expose_blob_index_ = false;
   bool allow_refresh_ = true;
   bool allow_mark_memtable_for_flush_ = true;

db/blob/blob_constants.h

@@ -11,6 +11,24 @@
 
 namespace ROCKSDB_NAMESPACE {
 
+// WARNING: This value is == kCurrentFileBlobIndexFileNumber.
+// Use this name only where file number zero means "no valid blob file" or
+// "current file" is not understood/supported.
 constexpr uint64_t kInvalidBlobFileNumber = 0;
 
+// WARNING: This value is == kInvalidBlobFileNumber.
+// Use this name only for BlobIndex references to the same physical file as what
+// is currently being read; generic blob-file metadata must treat zero as
+// invalid. (Using a distinct value like 1 was found to be more problematic,
+// e.g. because of legacy "stackable" blob implementation.)
+//
+// This "zero is invalid unless you are the embedded reader/writer" contract is
+// enforced by integrity checks that reject file number zero on generic paths;
+// see FileMetaData::UpdateBoundaries (write/output path) and Version::GetBlob /
+// Version::MultiGetBlob (read path). Same-file references must be resolved (by
+// EmbeddedBlobResolvingIterator) before they reach those paths. Do not weaken
+// those checks -- they catch leaks/corruption closer to the root cause.
+constexpr uint64_t kCurrentFileBlobIndexFileNumber = kInvalidBlobFileNumber;
+static_assert(kCurrentFileBlobIndexFileNumber == kInvalidBlobFileNumber);
+
 }  // namespace ROCKSDB_NAMESPACE

db/blob/blob_fetcher.cc

@@ -7,6 +7,7 @@
 
 #include "db/blob/blob_file_partition_manager.h"
 #include "db/blob/blob_index.h"
+#include "db/blob/same_file_blob_reader.h"
 #include "db/version_set.h"
 
 namespace ROCKSDB_NAMESPACE {
@@ -25,21 +26,47 @@ Status BlobFetcher::FetchBlob(const Slice& user_key,
   return status;
 }
 
-Status BlobFetcher::FetchBlob(const Slice& user_key,
-                              const BlobIndex& blob_index,
-                              FilePrefetchBuffer* prefetch_buffer,
-                              PinnableSlice* blob_value,
-                              uint64_t* bytes_read) const {
+Status VersionBlobFetcherBase::FetchBlob(const Slice& user_key,
+                                         const BlobIndex& blob_index,
+                                         FilePrefetchBuffer* prefetch_buffer,
+                                         PinnableSlice* blob_value,
+                                         uint64_t* bytes_read) const {
+  const ReadOptions& read_options = this->read_options();
   if (!allow_write_path_fallback_) {
     assert(version_);
 
-    return version_->GetBlob(read_options_, user_key, blob_index,
+    return version_->GetBlob(read_options, user_key, blob_index,
                              prefetch_buffer, blob_value, bytes_read);
   }
 
   return BlobFilePartitionManager::ResolveBlobDirectWriteIndex(
-      read_options_, user_key, blob_index, version_, blob_file_cache_,
+      read_options, user_key, blob_index, version_, blob_file_cache_,
       prefetch_buffer, blob_value, bytes_read);
 }
 
+Status EmbeddedAwareBlobFetcher::FetchBlob(const Slice& user_key,
+                                           const BlobIndex& blob_index,
+                                           FilePrefetchBuffer* prefetch_buffer,
+                                           PinnableSlice* blob_value,
+                                           uint64_t* bytes_read) const {
+  // Only an enabled decorator is ever routed to (see EffectiveFetcher()).
+  assert(same_file_reader_ != nullptr);
+  if (blob_index.IsSameFile()) {
+    // Same-file ("embedded") blob records live in the current SST and are read
+    // by the SameFileBlobReader (the BlockBasedTable). A separate blob-file
+    // prefetch buffer and bytes_read counter do not apply here (embedded reads
+    // account their own BLOB_DB_* stats via BlobSource).
+    return same_file_reader_->GetSameFileBlob(read_options(), blob_index,
+                                              blob_value);
+  }
+
+  if (base_ == nullptr) {
+    return Status::Corruption(
+        "Cannot resolve non-same-file blob reference without a base blob "
+        "fetcher");
+  }
+  return base_->FetchBlob(user_key, blob_index, prefetch_buffer, blob_value,
+                          bytes_read);
+}
+
 }  // namespace ROCKSDB_NAMESPACE

db/blob/blob_fetcher.h

@@ -5,6 +5,8 @@
 
 #pragma once
 
+#include <utility>
+
 #include "rocksdb/options.h"
 #include "rocksdb/status.h"
 
@@ -16,32 +18,174 @@ class Slice;
 class FilePrefetchBuffer;
 class PinnableSlice;
 class BlobIndex;
+class SameFileBlobReader;
 
-// A thin wrapper around blob retrieval. By default it reads through Version,
-// and it can optionally fall back to direct-write blob files that are not yet
-// manifest-visible.
+// Abstract interface for blob retrieval on the read path. An implementation
+// resolves a BlobIndex to its value, pinning it into a PinnableSlice. Callers
+// hold a plain const BlobFetcher* so resolution can be intercepted by a
+// decorator (e.g. EmbeddedAwareBlobFetcher) without their knowledge.
 class BlobFetcher {
  public:
-  BlobFetcher(const Version* version, const ReadOptions& read_options,
-              BlobFileCache* blob_file_cache = nullptr,
-              bool allow_write_path_fallback = false)
-      : version_(version),
-        read_options_(read_options),
-        blob_file_cache_(blob_file_cache),
-        allow_write_path_fallback_(allow_write_path_fallback) {}
+  virtual ~BlobFetcher() = default;
 
+  // Convenience overload: decodes `blob_index_slice` into a BlobIndex and
+  // dispatches to the virtual overload below. Non-virtual; shared by all
+  // implementations.
   Status FetchBlob(const Slice& user_key, const Slice& blob_index_slice,
                    FilePrefetchBuffer* prefetch_buffer,
                    PinnableSlice* blob_value, uint64_t* bytes_read) const;
 
+  virtual Status FetchBlob(const Slice& user_key, const BlobIndex& blob_index,
+                           FilePrefetchBuffer* prefetch_buffer,
+                           PinnableSlice* blob_value,
+                           uint64_t* bytes_read) const = 0;
+
+  // The read options this fetcher operates under. Used (via an
+  // EmbeddedAwareBlobFetcher decorator) to resolve same-file blob references,
+  // which need the current read_tier.
+  virtual const ReadOptions& read_options() const = 0;
+};
+
+// Shared implementation of the two Version-backed BlobFetchers below. Reads
+// through a Version, optionally falling back to direct-write blob files that
+// are not yet manifest-visible. Everything except the ReadOptions is stored
+// here; the concrete subclass decides whether it references a caller-owned
+// ReadOptions (VersionBlobFetcher) or owns its own copy
+// (OwningVersionBlobFetcher). FetchBlob reads the options via the virtual
+// read_options() so both storage strategies share one implementation.
+class VersionBlobFetcherBase : public BlobFetcher {
+ public:
+  // Un-hide the base's Slice-based convenience overload.
+  using BlobFetcher::FetchBlob;
+
   Status FetchBlob(const Slice& user_key, const BlobIndex& blob_index,
                    FilePrefetchBuffer* prefetch_buffer,
-                   PinnableSlice* blob_value, uint64_t* bytes_read) const;
+                   PinnableSlice* blob_value,
+                   uint64_t* bytes_read) const override;
+
+  // True if this fetcher has enough context to resolve a (non-inlined,
+  // non-same-file) blob reference: either a Version to read through, or the
+  // direct-write fallback (a blob file cache) enabled. Callers that may hold a
+  // null Version (e.g. DBIter) can check this to surface a clean Corruption on
+  // an unexpected blob index instead of dereferencing null in FetchBlob.
+  bool CanResolve() const {
+    return version_ != nullptr ||
+           (allow_write_path_fallback_ && blob_file_cache_ != nullptr);
+  }
+
+ protected:
+  VersionBlobFetcherBase(const Version* version, BlobFileCache* blob_file_cache,
+                         bool allow_write_path_fallback)
+      : version_(version),
+        blob_file_cache_(blob_file_cache),
+        allow_write_path_fallback_(allow_write_path_fallback) {}
 
- private:
   const Version* version_;
-  ReadOptions read_options_;
   BlobFileCache* blob_file_cache_;
   bool allow_write_path_fallback_;
 };
+
+// The default BlobFetcher for transient/stack use: holds a reference to a
+// caller-owned ReadOptions rather than a copy, avoiding a per-fetcher copy of
+// the (growing) ReadOptions on the hot path. The referenced ReadOptions must
+// outlive the fetcher, so this is only appropriate for fetchers whose lifetime
+// is bounded by the caller's (e.g. a stack local per Get / per resolution).
+// Use OwningVersionBlobFetcher for lifetime-independent fetchers.
+class VersionBlobFetcher : public VersionBlobFetcherBase {
+ public:
+  VersionBlobFetcher(const Version* version, const ReadOptions& read_options,
+                     BlobFileCache* blob_file_cache = nullptr,
+                     bool allow_write_path_fallback = false)
+      : VersionBlobFetcherBase(version, blob_file_cache,
+                               allow_write_path_fallback),
+        read_options_(read_options) {}
+
+  // Reject binding to a temporary ReadOptions, which would leave a dangling
+  // reference. Callers that cannot guarantee the ReadOptions outlives the
+  // fetcher must use OwningVersionBlobFetcher instead.
+  VersionBlobFetcher(const Version* version, ReadOptions&& read_options,
+                     BlobFileCache* blob_file_cache = nullptr,
+                     bool allow_write_path_fallback = false) = delete;
+
+  const ReadOptions& read_options() const override { return read_options_; }
+
+ private:
+  const ReadOptions& read_options_;
+};
+
+// A BlobFetcher that owns its ReadOptions, for fetchers whose lifetime is
+// independent of the caller that created them (the lazy read-path resolver and
+// compaction). Takes ownership of the ReadOptions by rvalue so ownership
+// transfer is explicit and no hidden copy is made on the way in; later reads
+// never touch caller state. Callers holding a borrowed ReadOptions they cannot
+// move must make the copy explicit (e.g. ReadOptions(read_options)).
+class OwningVersionBlobFetcher : public VersionBlobFetcherBase {
+ public:
+  OwningVersionBlobFetcher(const Version* version, ReadOptions&& read_options,
+                           BlobFileCache* blob_file_cache = nullptr,
+                           bool allow_write_path_fallback = false)
+      : VersionBlobFetcherBase(version, blob_file_cache,
+                               allow_write_path_fallback),
+        read_options_(std::move(read_options)) {}
+
+  const ReadOptions& read_options() const override { return read_options_; }
+
+ private:
+  ReadOptions read_options_;
+};
+
+// A BlobFetcher decorator that resolves same-file ("embedded") blob references
+// against a SameFileBlobReader (the current SST) and delegates all other
+// references to a base BlobFetcher. It is cheap to construct as a stack local
+// per resolution: the same-file target is per-SST while the base fetcher is one
+// per-Get object shared across SSTs, so an immutable per-SST decorator avoids
+// shared-mutable-state hazards under async MultiGet.
+//
+// A decorator is "enabled" when it has a same-file reader and "disabled"
+// (unusable) when constructed with a null `same_file_reader` (e.g. via the
+// default constructor). A disabled decorator must not be routed through
+// directly; callers construct one unconditionally and use EffectiveFetcher(),
+// which returns this decorator when enabled or the undecorated base fetcher
+// otherwise. This avoids std::optional / branching at call sites while adding
+// no indirection on the common (non-embedded) path.
+//
+// The base fetcher may be null for an entity whose blob columns are all
+// same-file (no separate-file blob support wired, e.g. SstFileReader); a
+// non-same-file reference then surfaces as a Corruption rather than silently.
+class EmbeddedAwareBlobFetcher : public BlobFetcher {
+ public:
+  explicit EmbeddedAwareBlobFetcher(
+      const BlobFetcher* base = nullptr,
+      const SameFileBlobReader* same_file_reader = nullptr)
+      : base_(base), same_file_reader_(same_file_reader) {}
+
+  // The fetcher to route resolution through: this decorator when it has a
+  // same-file reader, otherwise the (undecorated) base fetcher. Only an enabled
+  // decorator's FetchBlob is ever invoked.
+  const BlobFetcher* EffectiveFetcher() const {
+    return same_file_reader_ != nullptr ? this : base_;
+  }
+
+  // Un-hide the base's Slice-based convenience overload.
+  using BlobFetcher::FetchBlob;
+
+  Status FetchBlob(const Slice& user_key, const BlobIndex& blob_index,
+                   FilePrefetchBuffer* prefetch_buffer,
+                   PinnableSlice* blob_value,
+                   uint64_t* bytes_read) const override;
+
+  const ReadOptions& read_options() const override {
+    // Same-file resolution uses the base fetcher's read options; when there is
+    // no base (e.g. SstFileReader) a default (kReadAllTier) suffices.
+    static const ReadOptions kDefaultReadOptions;
+    return base_ != nullptr ? base_->read_options() : kDefaultReadOptions;
+  }
+
+ private:
+  const BlobFetcher* base_;
+  // Null iff this decorator is disabled (see class comment); otherwise the
+  // same-file blob reader for the current SST.
+  const SameFileBlobReader* same_file_reader_;
+};
+
 }  // namespace ROCKSDB_NAMESPACE

db/blob/blob_file_builder.cc

@@ -22,6 +22,7 @@
 #include "logging/logging.h"
 #include "options/cf_options.h"
 #include "options/options_helper.h"
+#include "rocksdb/file_system.h"
 #include "rocksdb/slice.h"
 #include "rocksdb/status.h"
 #include "test_util/sync_point.h"
@@ -200,6 +201,8 @@ Status BlobFileBuilder::OpenBlobFileIfNeeded() {
   {
     assert(file_options_);
     fo_copy = *file_options_;
+    fo_copy.open_contract = FileOpenContract::kNoReopenForWrite |
+                            FileOpenContract::kNoReadersWhileOpenForWrite;
     fo_copy.write_hint = write_hint_;
     Status s = NewWritableFile(fs_, blob_file_path, &file, fo_copy);
 

db/blob/blob_file_cache.cc

@@ -123,6 +123,7 @@ Status BlobFileCache::OpenBlobFileReaderUncached(
   Statistics* const statistics = immutable_options_->stats;
   RecordTick(statistics, NO_FILE_OPENS);
 
+  assert(file_options_);
   Status s = BlobFileReader::Create(
       *immutable_options_, read_options, *file_options_, column_family_id_,
       blob_file_read_hist_, blob_file_number, io_tracer_,

db/blob/blob_file_partition_manager.cc

@@ -24,6 +24,7 @@
 #include "file/read_write_util.h"
 #include "file/writable_file_writer.h"
 #include "logging/logging.h"
+#include "rocksdb/file_system.h"
 #include "util/aligned_buffer.h"
 #include "util/compression.h"
 #include "util/mutexlock.h"
@@ -199,7 +200,9 @@ Status BlobFilePartitionManager::OpenNewBlobFile(Partition* partition,
   }
 
   std::unique_ptr<FSWritableFile> file;
-  Status s = NewWritableFile(fs_, blob_file_path, &file, file_options_);
+  FileOptions writer_file_options = file_options_;
+  writer_file_options.open_contract = FileOpenContract::kNoReopenForWrite;
+  Status s = NewWritableFile(fs_, blob_file_path, &file, writer_file_options);
   if (!s.ok()) {
     RemoveFilePartitionMapping(blob_file_number);
     return s;
@@ -208,7 +211,7 @@ Status BlobFilePartitionManager::OpenNewBlobFile(Partition* partition,
   const bool perform_data_verification =
       checksum_handoff_file_types_.Contains(FileType::kBlobFile);
   auto file_writer = std::make_unique<WritableFileWriter>(
-      std::move(file), blob_file_path, file_options_, clock_, io_tracer_,
+      std::move(file), blob_file_path, writer_file_options, clock_, io_tracer_,
       statistics_, Histograms::BLOB_DB_BLOB_FILE_WRITE_MICROS, listeners_,
       file_checksum_gen_factory_, perform_data_verification);
 

db/blob/blob_file_reader.cc

@@ -12,6 +12,7 @@
 #include "db/blob/blob_log_format.h"
 #include "file/file_prefetch_buffer.h"
 #include "file/filename.h"
+#include "file/read_write_util.h"
 #include "monitoring/statistics_impl.h"
 #include "options/cf_options.h"
 #include "rocksdb/file_system.h"
@@ -20,6 +21,7 @@
 #include "table/format.h"
 #include "table/multiget_context.h"
 #include "test_util/sync_point.h"
+#include "util/aligned_buffer.h"
 #include "util/compression.h"
 #include "util/stop_watch.h"
 
@@ -105,24 +107,6 @@ Status BlobFileReader::OpenFile(
 
   constexpr IODebugContext* dbg = nullptr;
 
-  {
-    TEST_SYNC_POINT("BlobFileReader::OpenFile:GetFileSize");
-
-    const Status s =
-        fs->GetFileSize(blob_file_path, IOOptions(), file_size, dbg);
-    if (!s.ok()) {
-      return s;
-    }
-  }
-
-  if (!skip_footer_size_check &&
-      *file_size < BlobLogHeader::kSize + BlobLogFooter::kSize) {
-    return Status::Corruption("Malformed blob file");
-  }
-  if (skip_footer_size_check && *file_size < BlobLogHeader::kSize) {
-    return Status::Corruption("Malformed blob file");
-  }
-
   std::unique_ptr<FSRandomAccessFile> file;
   FileOptions reader_file_opts = file_opts;
 
@@ -142,6 +126,24 @@ Status BlobFileReader::OpenFile(
 
   assert(file);
 
+  {
+    Status s = GetFileSizeFromOpenFileOrPath(
+        file.get(), fs, blob_file_path, file_size, dbg,
+        FileSizeFallback::kNotSupportedOnly,
+        []() { TEST_SYNC_POINT("BlobFileReader::OpenFile:GetFileSize"); });
+    if (!s.ok()) {
+      return s;
+    }
+  }
+
+  if (!skip_footer_size_check &&
+      *file_size < BlobLogHeader::kSize + BlobLogFooter::kSize) {
+    return Status::Corruption("Malformed blob file");
+  }
+  if (skip_footer_size_check && *file_size < BlobLogHeader::kSize) {
+    return Status::Corruption("Malformed blob file");
+  }
+
   if (immutable_options.advise_random_on_open) {
     file->Hint(FSRandomAccessFile::kRandom);
   }
@@ -165,7 +167,7 @@ Status BlobFileReader::ReadHeader(const RandomAccessFileReader* file_reader,
 
   Slice header_slice;
   Buffer buf;
-  AlignedBuf aligned_buf;
+  AlignedBuffer direct_io_buffer;
 
   {
     TEST_SYNC_POINT("BlobFileReader::ReadHeader:ReadFromFile");
@@ -175,7 +177,7 @@ Status BlobFileReader::ReadHeader(const RandomAccessFileReader* file_reader,
 
     const Status s =
         ReadFromFile(file_reader, read_options, read_offset, read_size,
-                     statistics, &header_slice, &buf, &aligned_buf);
+                     statistics, &header_slice, &buf, &direct_io_buffer);
     if (!s.ok()) {
       return s;
     }
@@ -216,7 +218,7 @@ Status BlobFileReader::ReadFooter(const RandomAccessFileReader* file_reader,
 
   Slice footer_slice;
   Buffer buf;
-  AlignedBuf aligned_buf;
+  AlignedBuffer direct_io_buffer;
 
   {
     TEST_SYNC_POINT("BlobFileReader::ReadFooter:ReadFromFile");
@@ -226,7 +228,7 @@ Status BlobFileReader::ReadFooter(const RandomAccessFileReader* file_reader,
 
     const Status s =
         ReadFromFile(file_reader, read_options, read_offset, read_size,
-                     statistics, &footer_slice, &buf, &aligned_buf);
+                     statistics, &footer_slice, &buf, &direct_io_buffer);
     if (!s.ok()) {
       return s;
     }
@@ -257,10 +259,11 @@ Status BlobFileReader::ReadFromFile(const RandomAccessFileReader* file_reader,
                                     const ReadOptions& read_options,
                                     uint64_t read_offset, size_t read_size,
                                     Statistics* statistics, Slice* slice,
-                                    Buffer* buf, AlignedBuf* aligned_buf) {
+                                    Buffer* buf,
+                                    AlignedBuffer* direct_io_buffer) {
   assert(slice);
   assert(buf);
-  assert(aligned_buf);
+  assert(direct_io_buffer);
 
   assert(file_reader);
 
@@ -277,15 +280,15 @@ Status BlobFileReader::ReadFromFile(const RandomAccessFileReader* file_reader,
 
   if (file_reader->use_direct_io()) {
     constexpr char* scratch = nullptr;
+    AlignedBufferAllocationContext direct_io_context{direct_io_buffer};
 
     s = file_reader->Read(io_options, read_offset, read_size, slice, scratch,
-                          aligned_buf, &dbg);
+                          &direct_io_context, &dbg);
   } else {
     buf->reset(new char[read_size]);
-    constexpr AlignedBuf* aligned_scratch = nullptr;
 
     s = file_reader->Read(io_options, read_offset, read_size, slice, buf->get(),
-                          aligned_scratch, &dbg);
+                          nullptr, &dbg);
   }
 
   if (!s.ok()) {
@@ -349,7 +352,7 @@ Status BlobFileReader::GetBlob(
 
   Slice record_slice;
   Buffer buf;
-  AlignedBuf aligned_buf;
+  AlignedBuffer direct_io_buffer;
 
   bool prefetched = false;
 
@@ -379,7 +382,7 @@ Status BlobFileReader::GetBlob(
     const Status s =
         ReadFromFile(file_reader_.get(), read_options, record_offset,
                      static_cast<size_t>(record_size), statistics_,
-                     &record_slice, &buf, &aligned_buf);
+                     &record_slice, &buf, &direct_io_buffer);
     if (!s.ok()) {
       return s;
     }
@@ -477,7 +480,7 @@ void BlobFileReader::MultiGetBlob(
   }
 
   Buffer buf;
-  AlignedBuf aligned_buf;
+  AlignedBuffer direct_io_buffer;
 
   Status s;
   bool direct_io = file_reader_->use_direct_io();
@@ -500,8 +503,9 @@ void BlobFileReader::MultiGetBlob(
   IODebugContext dbg;
   s = file_reader_->PrepareIOOptions(read_options, opts, &dbg);
   if (s.ok()) {
+    AlignedBufferAllocationContext direct_io_context{&direct_io_buffer};
     s = file_reader_->MultiRead(opts, read_reqs.data(), read_reqs.size(),
-                                direct_io ? &aligned_buf : nullptr, &dbg);
+                                &direct_io_context, &dbg);
   }
   if (!s.ok()) {
     for (auto& req : read_reqs) {

db/blob/blob_file_reader.h

@@ -108,7 +108,7 @@ class BlobFileReader {
                              const ReadOptions& read_options,
                              uint64_t read_offset, size_t read_size,
                              Statistics* statistics, Slice* slice, Buffer* buf,
-                             AlignedBuf* aligned_buf);
+                             AlignedBuffer* direct_io_buffer);
 
   static Status VerifyBlob(const Slice& record_slice, const Slice& user_key,
                            uint64_t value_size);

db/blob/blob_file_reader_test.cc

@@ -7,10 +7,12 @@
 
 #include <cassert>
 #include <string>
+#include <utility>
 
 #include "db/blob/blob_contents.h"
 #include "db/blob/blob_log_format.h"
 #include "db/blob/blob_log_writer.h"
+#include "env/composite_env_wrapper.h"
 #include "env/mock_env.h"
 #include "file/filename.h"
 #include "file/read_write_util.h"
@@ -136,6 +138,167 @@ class BlobFileReaderTest : public testing::Test {
   std::unique_ptr<Env> mock_env_;
 };
 
+namespace {
+
+// Returns a stale path-level size for one target blob file while leaving the
+// opened file handle untouched. This lets the test verify
+// BlobFileReader::Create prefers the opened handle's size when the path stat
+// lags behind.
+class StalePathSizeFileSystem : public FileSystemWrapper {
+ public:
+  static const char* kClassName() { return "StalePathSizeFileSystem"; }
+
+  StalePathSizeFileSystem(const std::shared_ptr<FileSystem>& target,
+                          std::string stale_path)
+      : FileSystemWrapper(target), stale_path_(std::move(stale_path)) {}
+
+  const char* Name() const override { return kClassName(); }
+
+  IOStatus GetFileSize(const std::string& fname, const IOOptions& options,
+                       uint64_t* file_size, IODebugContext* dbg) override {
+    if (fname == stale_path_) {
+      *file_size = 0;
+      return IOStatus::OK();
+    }
+    return FileSystemWrapper::GetFileSize(fname, options, file_size, dbg);
+  }
+
+ private:
+  const std::string stale_path_;
+};
+
+// Makes opened blob file handles fail GetFileSize() with a hard error so the
+// reader path can verify that such errors are propagated directly.
+class FailingOpenFileSizeRandomAccessFile
+    : public FSRandomAccessFileOwnerWrapper {
+ public:
+  explicit FailingOpenFileSizeRandomAccessFile(
+      std::unique_ptr<FSRandomAccessFile>&& target)
+      : FSRandomAccessFileOwnerWrapper(std::move(target)) {}
+
+  IOStatus GetFileSize(uint64_t* /*result*/) override {
+    return IOStatus::IOError("open file size failed");
+  }
+};
+
+// Makes opened blob file handles report GetFileSize() as unsupported so the
+// reader path exercises its path-level fallback logic.
+class NotSupportedOpenFileSizeRandomAccessFile
+    : public FSRandomAccessFileOwnerWrapper {
+ public:
+  explicit NotSupportedOpenFileSizeRandomAccessFile(
+      std::unique_ptr<FSRandomAccessFile>&& target)
+      : FSRandomAccessFileOwnerWrapper(std::move(target)) {}
+
+  IOStatus GetFileSize(uint64_t* /*result*/) override {
+    return IOStatus::NotSupported("open file size not supported");
+  }
+};
+
+class OpenFileSizeFileSystem : public FileSystemWrapper {
+ public:
+  OpenFileSizeFileSystem(const std::shared_ptr<FileSystem>& target,
+                         std::string target_path)
+      : FileSystemWrapper(target), target_path_(std::move(target_path)) {}
+
+  IOStatus NewRandomAccessFile(const std::string& fname,
+                               const FileOptions& options,
+                               std::unique_ptr<FSRandomAccessFile>* result,
+                               IODebugContext* dbg) override {
+    std::unique_ptr<FSRandomAccessFile> file;
+    IOStatus s =
+        FileSystemWrapper::NewRandomAccessFile(fname, options, &file, dbg);
+    if (!s.ok()) {
+      return IOStatus(std::move(s));
+    }
+    if (fname == target_path_) {
+      *result = WrapTargetFile(std::move(file));
+    } else {
+      *result = std::move(file);
+    }
+    return IOStatus::OK();
+  }
+
+ protected:
+  virtual std::unique_ptr<FSRandomAccessFile> WrapTargetFile(
+      std::unique_ptr<FSRandomAccessFile>&& file) = 0;
+
+ private:
+  const std::string target_path_;
+};
+
+// Wraps the target blob file in FailingOpenFileSizeRandomAccessFile while
+// leaving all other files alone.
+class FailingOpenFileSizeFileSystem : public OpenFileSizeFileSystem {
+ public:
+  static const char* kClassName() { return "FailingOpenFileSizeFileSystem"; }
+
+  FailingOpenFileSizeFileSystem(const std::shared_ptr<FileSystem>& target,
+                                std::string target_path)
+      : OpenFileSizeFileSystem(target, std::move(target_path)) {}
+
+  const char* Name() const override { return kClassName(); }
+
+ private:
+  std::unique_ptr<FSRandomAccessFile> WrapTargetFile(
+      std::unique_ptr<FSRandomAccessFile>&& file) override {
+    return std::unique_ptr<FSRandomAccessFile>(
+        new FailingOpenFileSizeRandomAccessFile(std::move(file)));
+  }
+};
+
+// Wraps the target blob file in NotSupportedOpenFileSizeRandomAccessFile while
+// leaving all other files alone.
+class NotSupportedOpenFileSizeFileSystem : public OpenFileSizeFileSystem {
+ public:
+  static const char* kClassName() {
+    return "NotSupportedOpenFileSizeFileSystem";
+  }
+
+  NotSupportedOpenFileSizeFileSystem(const std::shared_ptr<FileSystem>& target,
+                                     std::string target_path)
+      : OpenFileSizeFileSystem(target, std::move(target_path)) {}
+
+  const char* Name() const override { return kClassName(); }
+
+ private:
+  std::unique_ptr<FSRandomAccessFile> WrapTargetFile(
+      std::unique_ptr<FSRandomAccessFile>&& file) override {
+    return std::unique_ptr<FSRandomAccessFile>(
+        new NotSupportedOpenFileSizeRandomAccessFile(std::move(file)));
+  }
+};
+
+// Forces the target blob file onto the NotSupported open-handle-size branch and
+// then fails the fallback path-level GetFileSize() call.
+class FailingFallbackPathSizeFileSystem
+    : public NotSupportedOpenFileSizeFileSystem {
+ public:
+  static const char* kClassName() {
+    return "FailingFallbackPathSizeFileSystem";
+  }
+
+  FailingFallbackPathSizeFileSystem(const std::shared_ptr<FileSystem>& target,
+                                    std::string target_path)
+      : NotSupportedOpenFileSizeFileSystem(target, target_path),
+        target_path_(std::move(target_path)) {}
+
+  const char* Name() const override { return kClassName(); }
+
+  IOStatus GetFileSize(const std::string& fname, const IOOptions& options,
+                       uint64_t* file_size, IODebugContext* dbg) override {
+    if (fname == target_path_) {
+      return IOStatus::IOError("fallback path size failed");
+    }
+    return FileSystemWrapper::GetFileSize(fname, options, file_size, dbg);
+  }
+
+ private:
+  const std::string target_path_;
+};
+
+}  // namespace
+
 TEST_F(BlobFileReaderTest, CreateReaderAndGetBlob) {
   Options options;
   options.env = mock_env_.get();
@@ -428,6 +591,211 @@ TEST_F(BlobFileReaderTest, CreateReaderAndGetBlob) {
   }
 }
 
+// Verifies Create() prefers the file size reported by the opened handle over a
+// stale path-level size query by returning a readable reader even when the path
+// stat intentionally reports 0 for the blob file.
+TEST_F(BlobFileReaderTest, CreateReaderUsesOpenedFileSizeWhenPathSizeIsStale) {
+  const std::string db_path = test::PerThreadDBPath(
+      mock_env_.get(),
+      "BlobFileReaderTest_CreateReaderUsesOpenedFileSizeWhenPathSizeIsStale");
+
+  constexpr uint64_t blob_file_number = 1;
+  const std::string blob_file_path = BlobFileName(db_path, blob_file_number);
+
+  auto stale_size_fs = std::make_shared<StalePathSizeFileSystem>(
+      mock_env_->GetFileSystem(), blob_file_path);
+  CompositeEnvWrapper stale_size_env(mock_env_.get(), stale_size_fs);
+
+  Options options;
+  options.env = &stale_size_env;
+  options.cf_paths.emplace_back(db_path, 0);
+  options.enable_blob_files = true;
+
+  ImmutableOptions immutable_options(options);
+
+  constexpr uint32_t column_family_id = 1;
+  constexpr bool has_ttl = false;
+  constexpr ExpirationRange expiration_range;
+  constexpr char key[] = "key";
+  constexpr char blob[] = "blob";
+
+  uint64_t blob_offset = 0;
+  uint64_t blob_size = 0;
+  WriteBlobFile(immutable_options, column_family_id, has_ttl, expiration_range,
+                expiration_range, blob_file_number, key, blob, kNoCompression,
+                &blob_offset, &blob_size);
+
+  std::unique_ptr<BlobFileReader> reader;
+  ReadOptions read_options;
+  read_options.verify_checksums = false;
+  ASSERT_OK(BlobFileReader::Create(
+      immutable_options, read_options, FileOptions(), column_family_id,
+      /*blob_file_read_hist=*/nullptr, blob_file_number, nullptr /*IOTracer*/,
+      &reader));
+  ASSERT_NE(reader, nullptr);
+
+  std::unique_ptr<BlobContents> value;
+  uint64_t bytes_read = 0;
+  ASSERT_OK(reader->GetBlob(read_options, key, blob_offset, blob_size,
+                            kNoCompression,
+                            /*prefetch_buffer=*/nullptr,
+                            /*allocator=*/nullptr, &value, &bytes_read));
+  ASSERT_NE(value, nullptr);
+  ASSERT_EQ(value->data(), Slice(blob));
+  ASSERT_EQ(bytes_read, blob_size);
+}
+
+// Verifies Create() falls back to the path-level GetFileSize() call when the
+// opened file handle reports that size queries are unsupported.
+TEST_F(BlobFileReaderTest,
+       CreateReaderFallsBackToPathSizeWhenOpenedFileSizeNotSupported) {
+  const std::string db_path = test::PerThreadDBPath(
+      mock_env_.get(),
+      "BlobFileReaderTest_"
+      "CreateReaderFallsBackToPathSizeWhenOpenedFileSizeNotSupported");
+
+  constexpr uint64_t blob_file_number = 1;
+  const std::string blob_file_path = BlobFileName(db_path, blob_file_number);
+
+  auto not_supported_open_file_size_fs =
+      std::make_shared<NotSupportedOpenFileSizeFileSystem>(
+          mock_env_->GetFileSystem(), blob_file_path);
+  CompositeEnvWrapper not_supported_open_file_size_env(
+      mock_env_.get(), not_supported_open_file_size_fs);
+
+  Options options;
+  options.env = &not_supported_open_file_size_env;
+  options.cf_paths.emplace_back(db_path, 0);
+  options.enable_blob_files = true;
+
+  ImmutableOptions immutable_options(options);
+
+  constexpr uint32_t column_family_id = 1;
+  constexpr bool has_ttl = false;
+  constexpr ExpirationRange expiration_range;
+  constexpr char key[] = "key";
+  constexpr char blob[] = "blob";
+
+  uint64_t blob_offset = 0;
+  uint64_t blob_size = 0;
+  WriteBlobFile(immutable_options, column_family_id, has_ttl, expiration_range,
+                expiration_range, blob_file_number, key, blob, kNoCompression,
+                &blob_offset, &blob_size);
+
+  std::unique_ptr<BlobFileReader> reader;
+  ReadOptions read_options;
+  read_options.verify_checksums = false;
+  ASSERT_OK(BlobFileReader::Create(
+      immutable_options, read_options, FileOptions(), column_family_id,
+      /*blob_file_read_hist=*/nullptr, blob_file_number, nullptr /*IOTracer*/,
+      &reader));
+  ASSERT_NE(reader, nullptr);
+
+  std::unique_ptr<BlobContents> value;
+  uint64_t bytes_read = 0;
+  ASSERT_OK(reader->GetBlob(read_options, key, blob_offset, blob_size,
+                            kNoCompression,
+                            /*prefetch_buffer=*/nullptr,
+                            /*allocator=*/nullptr, &value, &bytes_read));
+  ASSERT_NE(value, nullptr);
+  ASSERT_EQ(value->data(), Slice(blob));
+  ASSERT_EQ(bytes_read, blob_size);
+}
+
+// Verifies Create() propagates a real error from the opened file handle's
+// GetFileSize() rather than masking it with any path-level fallback.
+TEST_F(BlobFileReaderTest, CreateReaderPropagatesOpenedFileSizeError) {
+  const std::string db_path = test::PerThreadDBPath(
+      mock_env_.get(),
+      "BlobFileReaderTest_CreateReaderPropagatesOpenedFileSizeError");
+
+  constexpr uint64_t blob_file_number = 1;
+  const std::string blob_file_path = BlobFileName(db_path, blob_file_number);
+
+  auto failing_open_file_size_fs =
+      std::make_shared<FailingOpenFileSizeFileSystem>(
+          mock_env_->GetFileSystem(), blob_file_path);
+  CompositeEnvWrapper failing_open_file_size_env(mock_env_.get(),
+                                                 failing_open_file_size_fs);
+
+  Options options;
+  options.env = &failing_open_file_size_env;
+  options.cf_paths.emplace_back(db_path, 0);
+  options.enable_blob_files = true;
+
+  ImmutableOptions immutable_options(options);
+
+  constexpr uint32_t column_family_id = 1;
+  constexpr bool has_ttl = false;
+  constexpr ExpirationRange expiration_range;
+  constexpr char key[] = "key";
+  constexpr char blob[] = "blob";
+
+  uint64_t blob_offset = 0;
+  uint64_t blob_size = 0;
+  WriteBlobFile(immutable_options, column_family_id, has_ttl, expiration_range,
+                expiration_range, blob_file_number, key, blob, kNoCompression,
+                &blob_offset, &blob_size);
+
+  std::unique_ptr<BlobFileReader> reader;
+  ReadOptions read_options;
+  Status s = BlobFileReader::Create(
+      immutable_options, read_options, FileOptions(), column_family_id,
+      /*blob_file_read_hist=*/nullptr, blob_file_number, nullptr /*IOTracer*/,
+      &reader);
+  ASSERT_TRUE(s.IsIOError());
+  ASSERT_EQ(reader, nullptr);
+}
+
+// Verifies Create() also propagates errors from the path-level fallback size
+// query when the opened file handle only reports NotSupported for GetFileSize.
+TEST_F(
+    BlobFileReaderTest,
+    CreateReaderPropagatesFallbackPathSizeErrorWhenOpenedFileSizeNotSupported) {
+  const std::string db_path =
+      test::PerThreadDBPath(mock_env_.get(),
+                            "BlobFileReaderTest_"
+                            "CreateReaderPropagatesFallbackPathSizeErrorWhenOpe"
+                            "nedFileSizeNotSupported");
+
+  constexpr uint64_t blob_file_number = 1;
+  const std::string blob_file_path = BlobFileName(db_path, blob_file_number);
+
+  auto failing_fallback_path_size_fs =
+      std::make_shared<FailingFallbackPathSizeFileSystem>(
+          mock_env_->GetFileSystem(), blob_file_path);
+  CompositeEnvWrapper failing_fallback_path_size_env(
+      mock_env_.get(), failing_fallback_path_size_fs);
+
+  Options options;
+  options.env = &failing_fallback_path_size_env;
+  options.cf_paths.emplace_back(db_path, 0);
+  options.enable_blob_files = true;
+
+  ImmutableOptions immutable_options(options);
+
+  constexpr uint32_t column_family_id = 1;
+  constexpr bool has_ttl = false;
+  constexpr ExpirationRange expiration_range;
+  constexpr char key[] = "key";
+  constexpr char blob[] = "blob";
+
+  uint64_t blob_offset = 0;
+  uint64_t blob_size = 0;
+  WriteBlobFile(immutable_options, column_family_id, has_ttl, expiration_range,
+                expiration_range, blob_file_number, key, blob, kNoCompression,
+                &blob_offset, &blob_size);
+
+  std::unique_ptr<BlobFileReader> reader;
+  ReadOptions read_options;
+  Status s = BlobFileReader::Create(
+      immutable_options, read_options, FileOptions(), column_family_id,
+      /*blob_file_read_hist=*/nullptr, blob_file_number, nullptr /*IOTracer*/,
+      &reader);
+  ASSERT_TRUE(s.IsIOError());
+  ASSERT_EQ(reader, nullptr);
+}
+
 TEST_F(BlobFileReaderTest, Malformed) {
   // Write a blob file consisting of nothing but a header, and make sure we
   // detect the error when we open it for reading
@@ -849,7 +1217,6 @@ class BlobFileReaderIOErrorTest
 
 INSTANTIATE_TEST_CASE_P(BlobFileReaderTest, BlobFileReaderIOErrorTest,
                         ::testing::ValuesIn(std::vector<std::string>{
-                            "BlobFileReader::OpenFile:GetFileSize",
                             "BlobFileReader::OpenFile:NewRandomAccessFile",
                             "BlobFileReader::ReadHeader:ReadFromFile",
                             "BlobFileReader::ReadFooter:ReadFromFile",

db/blob/blob_garbage_meter.cc

@@ -32,6 +32,9 @@ Status BlobGarbageMeter::GetBlobReferenceDetails(const ParsedInternalKey& ikey,
   if (blob_index.IsInlined() || blob_index.HasTTL()) {
     return Status::Corruption("Unexpected TTL/inlined blob index");
   }
+  if (blob_index.IsSameFile()) {
+    return Status::OK();
+  }
 
   *blob_file_number = blob_index.file_number();
   *bytes =
@@ -125,7 +128,9 @@ Status BlobGarbageMeter::ProcessEntityBlobReferences(
             !s.ok()) {
           return s;
         }
-        AddFlow(file_number, blob_bytes, is_inflow);
+        if (file_number != kInvalidBlobFileNumber) {
+          AddFlow(file_number, blob_bytes, is_inflow);
+        }
         return Status::OK();
       });
 }

db/blob/blob_garbage_meter_test.cc

@@ -151,6 +151,24 @@ TEST(BlobGarbageMeterTest, PlainValue) {
   ASSERT_TRUE(blob_garbage_meter.flows().empty());
 }
 
+TEST(BlobGarbageMeterTest, SameFileBlobIndex) {
+  constexpr char user_key[] = "user_key";
+  constexpr SequenceNumber seq = 123;
+
+  const InternalKey key(user_key, seq, kTypeBlobIndex);
+  const Slice key_slice = key.Encode();
+
+  std::string value;
+  BlobIndex::EncodeBlob(&value, kCurrentFileBlobIndexFileNumber,
+                        /*offset=*/123, /*size=*/456, kNoCompression);
+
+  BlobGarbageMeter blob_garbage_meter;
+
+  ASSERT_OK(blob_garbage_meter.ProcessInFlow(key_slice, value));
+  ASSERT_OK(blob_garbage_meter.ProcessOutFlow(key_slice, value));
+  ASSERT_TRUE(blob_garbage_meter.flows().empty());
+}
+
 TEST(BlobGarbageMeterTest, CorruptInternalKey) {
   constexpr char corrupt_key[] = "i_am_corrupt";
   const Slice key_slice(corrupt_key);

db/blob/blob_gen2_format.cc

@@ -0,0 +1,111 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+#include "db/blob/blob_gen2_format.h"
+
+#include <array>
+#include <cstring>
+#include <string>
+
+#include "file/random_access_file_reader.h"
+#include "file/writable_file_writer.h"
+#include "rocksdb/options.h"
+#include "rocksdb/slice.h"
+#include "table/format.h"
+#include "util/cast_util.h"
+#include "util/coding.h"
+
+namespace ROCKSDB_NAMESPACE {
+
+Status ReadAndVerifySimpleGen2BlobRecord(
+    const ReadOptions& read_options, RandomAccessFileReader* file,
+    uint64_t record_offset, size_t payload_size, size_t record_size,
+    ChecksumType checksum_type, uint32_t base_context_checksum,
+    CompressionType expected_compression, char* buf) {
+  assert(file != nullptr);
+  assert(buf != nullptr);
+  assert(record_size == payload_size + kSimpleGen2BlobTrailerSize);
+
+  Slice result;
+  IOOptions opts;
+  IODebugContext dbg;
+  Status s = file->PrepareIOOptions(read_options, opts, &dbg);
+  if (s.ok()) {
+    s = file->Read(opts, record_offset, record_size, &result, buf, nullptr,
+                   &dbg);
+  }
+  if (!s.ok()) {
+    return s;
+  }
+  if (result.size() != record_size) {
+    return Status::Corruption("Could not read complete blob record");
+  }
+  // With mmap reads the data lands outside `buf`; copy it in so the caller can
+  // rely on `buf` owning the bytes (this is the only copy on the mmap path).
+  // TODO: fix this extra memcpy in the mmap case
+  if (result.data() != buf) {
+    memcpy(buf, result.data(), record_size);
+  }
+
+  const char* record = buf;
+  const CompressionType compression =
+      static_cast<CompressionType>(record[payload_size]);
+  if (compression != expected_compression) {
+    return Status::Corruption(
+        "Blob record compression does not match blob index");
+  }
+
+  if (read_options.verify_checksums) {
+    uint32_t stored = DecodeFixed32(record + payload_size + 1);
+    stored -= ChecksumModifierForContext(base_context_checksum, record_offset);
+    const uint32_t computed = ComputeBuiltinChecksumWithLastByte(
+        checksum_type, record, payload_size, record[payload_size]);
+    if (stored != computed) {
+      return Status::Corruption("Blob record checksum mismatch in " +
+                                file->file_name() + " offset " +
+                                std::to_string(record_offset) + " size " +
+                                std::to_string(payload_size));
+    }
+  }
+
+  if (compression != kNoCompression) {
+    return Status::Corruption("Blob record compression is not supported");
+  }
+  return Status::OK();
+}
+
+IOStatus WriteSimpleGen2BlobRecord(WritableFileWriter* file,
+                                   const WriteOptions& write_options,
+                                   ChecksumType checksum_type,
+                                   uint32_t base_context_checksum,
+                                   uint64_t record_offset, const Slice& payload,
+                                   CompressionType compression) {
+  assert(file != nullptr);
+  // Placeholder for future embedded blob compression support; only
+  // uncompressed payloads are currently written.
+  assert(compression == kNoCompression);
+
+  std::array<char, kSimpleGen2BlobTrailerSize> trailer;
+  trailer[0] = lossless_cast<char>(compression);
+  uint32_t checksum = ComputeBuiltinChecksumWithLastByte(
+      checksum_type, payload.data(), payload.size(), /*last_byte=*/trailer[0]);
+  checksum += ChecksumModifierForContext(base_context_checksum, record_offset);
+  EncodeFixed32(trailer.data() + 1, checksum);
+
+  IOOptions opts;
+  IOStatus io_s = WritableFileWriter::PrepareIOOptions(write_options, opts);
+  if (!io_s.ok()) {
+    return io_s;
+  }
+  if (!payload.empty()) {
+    io_s = file->Append(opts, payload);
+    if (!io_s.ok()) {
+      return io_s;
+    }
+  }
+  return file->Append(opts, Slice(trailer.data(), trailer.size()));
+}
+
+}  // namespace ROCKSDB_NAMESPACE

db/blob/blob_gen2_format.h

@@ -0,0 +1,92 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+#pragma once
+
+#include <cstddef>
+#include <cstdint>
+
+#include "cache/cache_key.h"
+#include "rocksdb/compression_type.h"
+#include "rocksdb/io_status.h"
+#include "rocksdb/rocksdb_namespace.h"
+#include "rocksdb/status.h"
+
+namespace ROCKSDB_NAMESPACE {
+
+struct ReadOptions;
+struct WriteOptions;
+class RandomAccessFileReader;
+class WritableFileWriter;
+class Slice;
+enum ChecksumType : char;
+
+// "SimpleGen2Blob" is the second-generation on-disk format for a blob payload
+// referenced by a same-file / context-relative blob index. A record is just the
+// payload bytes followed by a small block-style trailer:
+//
+//   <payload bytes> <1-byte compression marker> <4-byte checksum>
+//
+// The checksum is a builtin block checksum over (payload + compression marker),
+// context-modified by the record's absolute file offset (see
+// ChecksumModifierForContext / ComputeBuiltinChecksumWithLastByte in
+// table/format.h) so that an identical payload at a different offset gets a
+// different stored checksum. Payloads are currently uncompressed only.
+//
+// This format is deliberately decoupled from any particular file type: the same
+// record shape is read out of SST files (embedded blobs) and is intended to be
+// reused by other blob-bearing files. The only file-type-specific inputs are
+// the byte source (a RandomAccessFileReader) and the checksum context (checksum
+// type + base context checksum), both passed in by the caller.
+inline constexpr size_t kSimpleGen2BlobTrailerSize = 5;
+
+// Cache key for the SimpleGen2Blob record at byte `record_offset` within a file
+// whose base cache key is `base_cache_key`. Delegates to the shared
+// OffsetableCacheKey::WithOffsetForMinSizeRecord scheme -- the exact same
+// scheme block-based SST data blocks use via BlockBasedTable::GetCacheKey --
+// which is valid because a SimpleGen2Blob record always carries a >= 5-byte
+// trailer. As a result the cache key is collision-free with the file's data
+// blocks even when the blob cache and block cache are the same cache, with no
+// separately maintained algorithm to drift out of sync.
+inline CacheKey GetSimpleGen2BlobCacheKey(
+    const OffsetableCacheKey& base_cache_key, uint64_t record_offset) {
+  return base_cache_key.WithOffsetForMinSizeRecord(record_offset);
+}
+
+// Reads a SimpleGen2Blob record of `record_size` bytes (which must equal
+// `payload_size + kSimpleGen2BlobTrailerSize`) from `file` at `record_offset`
+// into `buf` (capacity >= record_size) and verifies its compression marker and,
+// when read_options.verify_checksums is set, its context-modified checksum.
+//
+// `checksum_type` and `base_context_checksum` are the file's checksum context
+// (e.g. from the SST footer). `expected_compression` is the compression type
+// the caller expects the record to carry (from the blob index); it must match
+// the record's marker, and currently must be kNoCompression.
+//
+// On success, buf[0, payload_size) holds the verified, uncompressed payload
+// (the trailer remains at the tail of `buf` and can be ignored).
+Status ReadAndVerifySimpleGen2BlobRecord(
+    const ReadOptions& read_options, RandomAccessFileReader* file,
+    uint64_t record_offset, size_t payload_size, size_t record_size,
+    ChecksumType checksum_type, uint32_t base_context_checksum,
+    CompressionType expected_compression, char* buf);
+
+// Writes a SimpleGen2Blob record for `payload` at the current end of `file`,
+// which the caller asserts is byte offset `record_offset`. Appends the payload
+// bytes followed by the 5-byte trailer (compression marker + context-modified
+// builtin checksum), mirroring ReadAndVerifySimpleGen2BlobRecord so the on-disk
+// record format lives in one module.
+//
+// `checksum_type` and `base_context_checksum` are the file's checksum context
+// (e.g. from the SST footer). `compression` is the payload's compression type,
+// which currently must be kNoCompression.
+IOStatus WriteSimpleGen2BlobRecord(WritableFileWriter* file,
+                                   const WriteOptions& write_options,
+                                   ChecksumType checksum_type,
+                                   uint32_t base_context_checksum,
+                                   uint64_t record_offset, const Slice& payload,
+                                   CompressionType compression);
+
+}  // namespace ROCKSDB_NAMESPACE

db/blob/blob_index.h

@@ -7,6 +7,7 @@
 #include <sstream>
 #include <string>
 
+#include "db/blob/blob_constants.h"
 #include "rocksdb/compression_type.h"
 #include "util/coding.h"
 #include "util/compression.h"
@@ -57,6 +58,14 @@ class BlobIndex {
 
   bool IsInlined() const { return type_ == Type::kInlinedTTL; }
 
+  // True for a blob record stored in the same physical file as the table entry.
+  // Only embedded-blob SST reader/writer paths should interpret file number
+  // zero this way; generic blob metadata treats it as invalid.
+  bool IsSameFile() const {
+    return (type_ == Type::kBlob || type_ == Type::kBlobTTL) &&
+           file_number_ == kCurrentFileBlobIndexFileNumber;
+  }
+
   bool HasTTL() const {
     return type_ == Type::kInlinedTTL || type_ == Type::kBlobTTL;
   }
@@ -125,8 +134,13 @@ class BlobIndex {
     if (IsInlined()) {
       oss << "[inlined blob] value:" << value_.ToString(output_hex);
     } else {
-      oss << "[blob ref] file:" << file_number_ << " offset:" << offset_
-          << " size:" << size_
+      oss << "[blob ref] file:";
+      if (IsSameFile()) {
+        oss << "same";
+      } else {
+        oss << file_number_;
+      }
+      oss << " offset:" << offset_ << " size:" << size_
           << " compression: " << CompressionTypeToString(compression_);
     }
 

db/blob/blob_source.cc

@@ -12,7 +12,10 @@
 #include "cache/charged_cache.h"
 #include "db/blob/blob_contents.h"
 #include "db/blob/blob_file_reader.h"
+#include "db/blob/blob_gen2_format.h"
 #include "db/blob/blob_log_format.h"
+#include "file/random_access_file_reader.h"
+#include "memory/memory_allocator_impl.h"
 #include "monitoring/statistics_impl.h"
 #include "options/cf_options.h"
 #include "table/get_context.h"
@@ -305,6 +308,96 @@ Status BlobSource::GetBlob(const ReadOptions& read_options,
   return s;
 }
 
+Status BlobSource::GetSimpleGen2Blob(
+    const ReadOptions& read_options, const OffsetableCacheKey& base_cache_key,
+    RandomAccessFileReader* file, uint64_t record_offset, uint64_t payload_size,
+    ChecksumType checksum_type, uint32_t base_context_checksum,
+    CompressionType expected_compression, PinnableSlice* value,
+    uint64_t* bytes_read) {
+  assert(value);
+  assert(file);
+
+  const uint64_t record_size = payload_size + kSimpleGen2BlobTrailerSize;
+
+  // The cache key is derived from the SimpleGen2Blob format (shared scheme with
+  // block-based SST blocks); see GetSimpleGen2BlobCacheKey.
+  const CacheKey cache_key =
+      GetSimpleGen2BlobCacheKey(base_cache_key, record_offset);
+
+  Status s;
+
+  CacheHandleGuard<BlobContents> blob_handle;
+
+  // First, try to get the blob from the cache.
+  if (blob_cache_) {
+    Slice key = cache_key.AsSlice();
+    s = GetBlobFromCache(key, &blob_handle);
+    if (s.ok()) {
+      PinCachedBlob(&blob_handle, value);
+
+      // For consistency, the on-disk record size is assigned to bytes_read on
+      // both cache hits and misses.
+      if (bytes_read) {
+        *bytes_read = record_size;
+      }
+      return s;
+    }
+  }
+
+  assert(blob_handle.IsEmpty());
+
+  const bool no_io = read_options.read_tier == kBlockCacheTier;
+  if (no_io) {
+    return Status::Incomplete("Cannot read blob(s): no disk I/O allowed");
+  }
+
+  // Cache miss (or no cache configured). Read the record into a buffer
+  // allocated from the blob cache's memory allocator when we intend to insert
+  // it, exposing the uncompressed payload as BlobContents (the trailer just
+  // sits unused at the tail of the buffer).
+  MemoryAllocator* const allocator = (blob_cache_ && read_options.fill_cache)
+                                         ? blob_cache_.get()->memory_allocator()
+                                         : nullptr;
+
+  CacheAllocationPtr buf =
+      AllocateBlock(static_cast<size_t>(record_size), allocator);
+  s = ReadAndVerifySimpleGen2BlobRecord(
+      read_options, file, record_offset, static_cast<size_t>(payload_size),
+      static_cast<size_t>(record_size), checksum_type, base_context_checksum,
+      expected_compression, buf.get());
+  if (!s.ok()) {
+    return s;
+  }
+
+  std::unique_ptr<BlobContents> blob_contents(
+      new BlobContents(std::move(buf), static_cast<size_t>(payload_size)));
+
+  // Record the per-read statistics (mirrors BlobFileReader::GetBlob).
+  RecordTick(statistics_, BLOB_DB_BLOB_FILE_BYTES_READ, record_size);
+  PERF_COUNTER_ADD(blob_read_count, 1);
+  PERF_COUNTER_ADD(blob_read_byte, record_size);
+  if (bytes_read) {
+    *bytes_read = record_size;
+  }
+
+  if (blob_cache_ && read_options.fill_cache) {
+    // If filling cache is allowed and a cache is configured, try to put the
+    // blob into the cache.
+    Slice key = cache_key.AsSlice();
+    s = PutBlobIntoCache(key, &blob_contents, &blob_handle);
+    if (!s.ok()) {
+      return s;
+    }
+
+    PinCachedBlob(&blob_handle, value);
+  } else {
+    PinOwnedBlob(&blob_contents, value);
+  }
+
+  assert(s.ok());
+  return s;
+}
+
 void BlobSource::MultiGetBlob(const ReadOptions& read_options,
                               autovector<BlobFileReadRequests>& blob_reqs,
                               uint64_t* bytes_read) {

db/blob/blob_source.h

@@ -25,6 +25,8 @@ struct MutableCFOptions;
 class Status;
 class FilePrefetchBuffer;
 class Slice;
+class RandomAccessFileReader;
+enum ChecksumType : char;
 
 // BlobSource is a class that provides universal access to blobs, regardless of
 // whether they are in the blob cache, secondary cache, or (remote) storage.
@@ -59,6 +61,41 @@ class BlobSource {
                  FilePrefetchBuffer* prefetch_buffer, PinnableSlice* value,
                  uint64_t* bytes_read);
 
+  // Reads a SimpleGen2Blob payload (see db/blob/blob_gen2_format.h) through the
+  // blob value cache and BLOB_DB_* statistics. This is the counterpart to
+  // GetBlob() for the second-generation blob record format, which is read
+  // directly from a RandomAccessFileReader rather than from a traditional blob
+  // file.
+  //
+  // The cache key is derived from the SimpleGen2Blob format itself, not chosen
+  // by the caller: it is GetSimpleGen2BlobCacheKey(base_cache_key,
+  // record_offset), the same offset scheme block-based SST blocks use. The
+  // caller supplies only its file's `base_cache_key` (db_id / db_session_id /
+  // file_number). This keeps blob records collision-free with the file's data
+  // blocks even when the blob cache and block cache are the same cache.
+  //
+  // `file`, `record_offset`, `payload_size`, `checksum_type`,
+  // `base_context_checksum`, and `expected_compression` are the inputs to the
+  // SimpleGen2Blob reader used on a cache miss (see
+  // ReadAndVerifySimpleGen2BlobRecord). The on-disk record size (payload +
+  // trailer) is reported via `*bytes_read` (when non-null) and the
+  // BLOB_DB_BLOB_FILE_BYTES_READ / blob_read_byte counters, consistently on
+  // both cache hits and misses.
+  //
+  // On a cache hit, pins the cached value into `*value` (no copy). On a miss,
+  // reads + verifies the record into a cache-allocator buffer, records the
+  // per-read stats, inserts it into the cache (when configured and fill_cache
+  // is set), and pins it into `*value`. If blob_cache_ is not configured, the
+  // record is still read and read stats recorded, just without a cache
+  // lookup/insert.
+  Status GetSimpleGen2Blob(const ReadOptions& read_options,
+                           const OffsetableCacheKey& base_cache_key,
+                           RandomAccessFileReader* file, uint64_t record_offset,
+                           uint64_t payload_size, ChecksumType checksum_type,
+                           uint32_t base_context_checksum,
+                           CompressionType expected_compression,
+                           PinnableSlice* value, uint64_t* bytes_read);
+
   // Read multiple blobs from the underlying cache or blob file(s).
   //
   // If successful, returns ok and sets "result" in the elements of "blob_reqs"

db/blob/blob_write_batch_transformer.cc

@@ -215,11 +215,10 @@ Status BlobWriteBatchTransformer::PutEntityCF(uint32_t column_family_id,
         output_batch_, column_family_id, key, entity);
   }
 
-  Slice entity_ref = entity;
   std::vector<WideColumn> columns;
   std::vector<std::pair<size_t, BlobIndex>> existing_blob_columns;
-  Status status = WideColumnSerialization::DeserializeV2(entity_ref, columns,
-                                                         existing_blob_columns);
+  Status status = WideColumnSerialization::Deserialize(entity, columns,
+                                                       &existing_blob_columns);
   if (status.ok()) {
     // Reject pre-serialized entities that already contain blob references.
     // Passing them through would preserve references to blob files outside this

db/blob/db_blob_direct_write_test.cc

@@ -10,11 +10,14 @@
 
 #include <algorithm>
 #include <array>
+#include <functional>
 #include <limits>
 #include <mutex>
 #include <set>
 #include <sstream>
 #include <string>
+#include <unordered_set>
+#include <utility>
 #include <vector>
 
 #include "cache/compressed_secondary_cache.h"
@@ -26,10 +29,12 @@
 #include "db/db_test_util.h"
 #include "db/db_with_timestamp_test_util.h"
 #include "db/wide/wide_column_test_util.h"
+#include "env/composite_env_wrapper.h"
 #include "file/filename.h"
 #include "file/random_access_file_reader.h"
 #include "port/stack_trace.h"
 #include "rocksdb/convenience.h"
+#include "rocksdb/file_system.h"
 #include "rocksdb/trace_reader_writer.h"
 #include "rocksdb/trace_record.h"
 #include "rocksdb/utilities/replayer.h"
@@ -101,6 +106,206 @@ class RecordingBlobDirectWritePartitionStrategy
   std::vector<Call> calls_;
 };
 
+namespace {
+
+// Matches BlobDB file names so the test filesystem can special-case only active
+// blob files without changing unrelated file opens.
+bool IsBlobFilePath(const std::string& fname) {
+  const size_t basename_pos = fname.find_last_of("/\\");
+  const std::string basename = basename_pos == std::string::npos
+                                   ? fname
+                                   : fname.substr(basename_pos + 1);
+  uint64_t file_number = 0;
+  FileType file_type = kWalFile;
+  return ParseFileName(basename, &file_number, &file_type) &&
+         file_type == kBlobFile;
+}
+
+// Keeps track of when an active blob file stops being writer-visible so the
+// test filesystem can model current-size visibility only while the file remains
+// active.
+class ActiveBlobVisibilityWritableFile : public FSWritableFileOwnerWrapper {
+ public:
+  ActiveBlobVisibilityWritableFile(std::unique_ptr<FSWritableFile>&& target,
+                                   std::function<void()> on_close)
+      : FSWritableFileOwnerWrapper(std::move(target)),
+        on_close_(std::move(on_close)) {}
+
+  ~ActiveBlobVisibilityWritableFile() override { DeactivateOnce(); }
+
+  IOStatus Close(const IOOptions& options, IODebugContext* dbg) override {
+    IOStatus s = target()->Close(options, dbg);
+    DeactivateOnce();
+    return s;
+  }
+
+ private:
+  void DeactivateOnce() {
+    if (on_close_) {
+      on_close_();
+      on_close_ = nullptr;
+    }
+  }
+
+  std::function<void()> on_close_;
+};
+
+// Simulates a remote readable file whose GetFileSize() either exposes the live
+// on-disk size for default-contract active blobs or hides it for
+// append-only-no-readers files.
+class ActiveBlobVisibilityRandomAccessFile
+    : public FSRandomAccessFileOwnerWrapper {
+ public:
+  ActiveBlobVisibilityRandomAccessFile(
+      std::unique_ptr<FSRandomAccessFile>&& target, bool expose_current_size,
+      std::shared_ptr<FileSystem> underlying_fs, std::string fname)
+      : FSRandomAccessFileOwnerWrapper(std::move(target)),
+        expose_current_size_(expose_current_size),
+        underlying_fs_(std::move(underlying_fs)),
+        fname_(std::move(fname)) {}
+
+  IOStatus GetFileSize(uint64_t* result) override {
+    if (!expose_current_size_) {
+      *result = 0;
+      return IOStatus::OK();
+    }
+    // Delegate to the underlying FS path-level GetFileSize (bypassing the
+    // wrapper that reports 0 for active blobs) to simulate a remote FS that
+    // exposes the current file size through the open handle.
+    return underlying_fs_->GetFileSize(fname_, IOOptions(), result,
+                                       /*dbg=*/nullptr);
+  }
+
+ private:
+  const bool expose_current_size_;
+  std::shared_ptr<FileSystem> underlying_fs_;
+  std::string fname_;
+};
+
+// Models a remote filesystem where active direct-write blobs are only
+// reader-visible when the writer does not promise no concurrent readers.
+class RemoteBlobVisibilityFileSystem : public FileSystemWrapper {
+ public:
+  explicit RemoteBlobVisibilityFileSystem(const std::shared_ptr<FileSystem>& fs)
+      : FileSystemWrapper(fs) {}
+
+  static const char* kClassName() { return "RemoteBlobVisibilityFileSystem"; }
+  const char* Name() const override { return kClassName(); }
+
+  IOStatus NewWritableFile(const std::string& fname, const FileOptions& opts,
+                           std::unique_ptr<FSWritableFile>* result,
+                           IODebugContext* dbg) override {
+    std::unique_ptr<FSWritableFile> file;
+    IOStatus s = target()->NewWritableFile(fname, opts, &file, dbg);
+    if (!s.ok() || !IsBlobFilePath(fname)) {
+      *result = std::move(file);
+      return IOStatus(std::move(s));
+    }
+
+    const bool has_no_reopen_for_write_contract = HasFileOpenContract(
+        opts.open_contract, FileOpenContract::kNoReopenForWrite);
+    const bool has_no_readers_while_open_for_write_contract =
+        HasFileOpenContract(opts.open_contract,
+                            FileOpenContract::kNoReadersWhileOpenForWrite);
+    {
+      std::lock_guard<std::mutex> lock(mu_);
+      saw_no_reopen_for_write_writer_contract_ |=
+          has_no_reopen_for_write_contract;
+      saw_no_readers_while_open_for_write_writer_contract_ |=
+          has_no_readers_while_open_for_write_contract;
+      if (!has_no_readers_while_open_for_write_contract) {
+        active_blob_paths_.insert(fname);
+      }
+    }
+
+    if (has_no_readers_while_open_for_write_contract) {
+      *result = std::move(file);
+      return IOStatus::OK();
+    }
+
+    result->reset(new ActiveBlobVisibilityWritableFile(
+        std::move(file), [this, tracked_path = std::string(fname)]() {
+          DeactivateBlobPath(tracked_path);
+        }));
+    return IOStatus::OK();
+  }
+
+  IOStatus NewRandomAccessFile(const std::string& fname,
+                               const FileOptions& opts,
+                               std::unique_ptr<FSRandomAccessFile>* result,
+                               IODebugContext* dbg) override {
+    std::unique_ptr<FSRandomAccessFile> file;
+    IOStatus s = target()->NewRandomAccessFile(fname, opts, &file, dbg);
+    if (!s.ok() || !IsBlobFilePath(fname)) {
+      *result = std::move(file);
+      return IOStatus(std::move(s));
+    }
+
+    const bool has_no_readers_while_open_for_write_contract =
+        HasFileOpenContract(opts.open_contract,
+                            FileOpenContract::kNoReadersWhileOpenForWrite);
+    const bool active_blob = IsActiveBlobPath(fname);
+    {
+      std::lock_guard<std::mutex> lock(mu_);
+      saw_no_readers_while_open_for_write_reader_contract_ |=
+          has_no_readers_while_open_for_write_contract;
+    }
+
+    if (!active_blob) {
+      *result = std::move(file);
+      return IOStatus::OK();
+    }
+
+    result->reset(new ActiveBlobVisibilityRandomAccessFile(
+        std::move(file), !has_no_readers_while_open_for_write_contract, target_,
+        fname));
+    return IOStatus::OK();
+  }
+
+  IOStatus GetFileSize(const std::string& fname, const IOOptions& options,
+                       uint64_t* file_size, IODebugContext* dbg) override {
+    if (IsActiveBlobPath(fname)) {
+      *file_size = 0;
+      return IOStatus::OK();
+    }
+    return target()->GetFileSize(fname, options, file_size, dbg);
+  }
+
+  bool SawNoReopenForWriteWriterContract() const {
+    std::lock_guard<std::mutex> lock(mu_);
+    return saw_no_reopen_for_write_writer_contract_;
+  }
+
+  bool SawNoReadersWhileOpenForWriteWriterContract() const {
+    std::lock_guard<std::mutex> lock(mu_);
+    return saw_no_readers_while_open_for_write_writer_contract_;
+  }
+
+  bool SawNoReadersWhileOpenForWriteReaderContract() const {
+    std::lock_guard<std::mutex> lock(mu_);
+    return saw_no_readers_while_open_for_write_reader_contract_;
+  }
+
+ private:
+  void DeactivateBlobPath(const std::string& fname) {
+    std::lock_guard<std::mutex> lock(mu_);
+    active_blob_paths_.erase(fname);
+  }
+
+  bool IsActiveBlobPath(const std::string& fname) const {
+    std::lock_guard<std::mutex> lock(mu_);
+    return active_blob_paths_.find(fname) != active_blob_paths_.end();
+  }
+
+  mutable std::mutex mu_;
+  bool saw_no_reopen_for_write_writer_contract_ = false;
+  bool saw_no_readers_while_open_for_write_writer_contract_ = false;
+  bool saw_no_readers_while_open_for_write_reader_contract_ = false;
+  std::unordered_set<std::string> active_blob_paths_;
+};
+
+}  // namespace
+
 class DBBlobDirectWriteTest : public DBTestBase {
  protected:
   DBBlobDirectWriteTest()
@@ -816,6 +1021,30 @@ TEST_F(DBBlobDirectWriteTest, DirectWriteImmutableMemtableRead) {
   verify_reads();
 }
 
+// Verifies active blob direct-write files promise no write reopen without also
+// promising no concurrent readers.
+TEST_F(DBBlobDirectWriteTest, DirectWriteUsesReadableContractForRemoteFile) {
+  auto remote_fs =
+      std::make_shared<RemoteBlobVisibilityFileSystem>(env_->GetFileSystem());
+  std::unique_ptr<Env> remote_env(new CompositeEnvWrapper(env_, remote_fs));
+
+  Options options = GetDirectWriteOptions();
+  options.env = remote_env.get();
+  Reopen(options);
+
+  const std::string key = "remote_blob_key";
+  const std::string value(128, 'w');
+
+  ASSERT_OK(Put(key, value));
+  ASSERT_TRUE(remote_fs->SawNoReopenForWriteWriterContract());
+  ASSERT_FALSE(remote_fs->SawNoReadersWhileOpenForWriteWriterContract());
+
+  ASSERT_EQ(Get(key), value);
+  ASSERT_FALSE(remote_fs->SawNoReadersWhileOpenForWriteReaderContract());
+
+  Close();
+}
+
 TEST_F(DBBlobDirectWriteTest, DirectWriteRefreshesReaderAfterFlush) {
   Options options = GetDirectWriteOptions();
   options.blob_direct_write_partitions = 1;

db/blob/db_blob_index_test.cc

@@ -25,6 +25,9 @@
 #include "file/filename.h"
 #include "port/port.h"
 #include "port/stack_trace.h"
+#include "rocksdb/perf_context.h"
+#include "rocksdb/sst_file_writer.h"
+#include "rocksdb/table.h"
 #include "util/string_util.h"
 #include "utilities/merge_operators.h"
 
@@ -41,6 +44,18 @@ void CorruptPinnedBlobIndexOnCleanup(void* arg1, void* /*arg2*/) {
 
 }  // namespace
 
+TEST(BlobIndexTest, SameFileBlobIndex) {
+  std::string encoded;
+  BlobIndex::EncodeBlob(&encoded, kCurrentFileBlobIndexFileNumber,
+                        /*offset=*/123, /*size=*/456, kNoCompression);
+
+  BlobIndex blob_index;
+  ASSERT_OK(blob_index.DecodeFrom(encoded));
+  EXPECT_TRUE(blob_index.IsSameFile());
+  EXPECT_EQ(blob_index.file_number(), kCurrentFileBlobIndexFileNumber);
+  EXPECT_NE(blob_index.DebugString(false).find("file:same"), std::string::npos);
+}
+
 // kTypeBlobIndex is a value type used by BlobDB only. The base rocksdb
 // should accept the value type on write, and report not supported value
 // for reads, unless caller request for it explicitly. The base rocksdb
@@ -128,14 +143,24 @@ class DBBlobIndexTest : public DBTestBase {
         columns, s, is_blob_index, value_found);
   }
 
-  bool MaybeResolveMemtableBlobValueForTest(const Slice& key,
-                                            const BlobFetcher* blob_fetcher,
-                                            PinnableSlice* value,
-                                            PinnableWideColumns* columns,
-                                            Status* s, bool* is_blob_index,
-                                            bool* value_found = nullptr) {
-    return DBImpl::MaybeResolveMemtableBlobValue(
-        key, blob_fetcher, value, columns, s, is_blob_index, value_found);
+  Status MaybeResolveMemtableBlobValueForTest(
+      const Slice& key, const BlobFetcher* blob_fetcher, PinnableSlice* value,
+      PinnableWideColumns* columns, bool* did_resolve, bool* is_blob_index,
+      bool* value_found = nullptr) {
+    return DBImpl::MaybeResolveMemtableBlobValue(key, blob_fetcher, value,
+                                                 columns, did_resolve,
+                                                 is_blob_index, value_found);
+  }
+
+  Status PostprocessMemtableValueReadForTest(
+      const Slice& key, bool resolve_blob_backed_memtable_value,
+      const BlobFetcher* blob_fetcher, PinnableSlice* value,
+      PinnableWideColumns* columns, Status memtable_read_status,
+      bool* is_blob_index, bool* value_found = nullptr) {
+    return DBImpl::PostprocessMemtableValueRead(
+        key, /*timestamp=*/nullptr, resolve_blob_backed_memtable_value,
+        blob_fetcher, value, columns, std::move(memtable_read_status),
+        is_blob_index, value_found);
   }
 
   Options GetTestOptions() {
@@ -294,17 +319,65 @@ TEST_F(DBBlobIndexTest,
   value.GetSelf()->assign(blob_index.data(), blob_index.size());
   value.PinSelf();
 
-  Status s = Status::OK();
+  bool did_resolve = false;
   bool is_blob_index = true;
-  ASSERT_TRUE(MaybeResolveMemtableBlobValueForTest(
-      Slice("key"), /*blob_fetcher=*/nullptr, &value, /*columns=*/nullptr, &s,
-      &is_blob_index));
+  Status s = MaybeResolveMemtableBlobValueForTest(
+      Slice("key"), /*blob_fetcher=*/nullptr, &value, /*columns=*/nullptr,
+      &did_resolve, &is_blob_index);
 
-  ASSERT_TRUE(s.IsNotSupported()) << s.ToString();
+  ASSERT_TRUE(did_resolve);
+  ASSERT_TRUE(s.IsCorruption()) << s.ToString();
   ASSERT_TRUE(value.empty());
   ASSERT_FALSE(is_blob_index);
 }
 
+TEST_F(DBBlobIndexTest, PostprocessMemtableValueReadClearsOutputsOnError) {
+  // Regression test for the "swallowed status" FIXME: on any error,
+  // PostprocessMemtableValueRead must clear the outputs and return the status,
+  // never leaving a half-populated result behind a non-OK status.
+
+  // Case 1: the incoming memtable read status is already non-OK. The resolve
+  // path is skipped; the finalization block clears outputs and returns it.
+  {
+    PinnableSlice value;
+    value.GetSelf()->assign("stale", 5);
+    value.PinSelf();
+
+    bool is_blob_index = false;
+    Status s = PostprocessMemtableValueReadForTest(
+        Slice("key"), /*resolve_blob_backed_memtable_value=*/true,
+        /*blob_fetcher=*/nullptr, &value, /*columns=*/nullptr,
+        Status::Corruption("read failed"), &is_blob_index);
+
+    ASSERT_TRUE(s.IsCorruption()) << s.ToString();
+    ASSERT_TRUE(value.empty());
+  }
+
+  // Case 2: the incoming status is OK but the blob-resolve path itself fails (a
+  // blob-backed value with no fetcher). This is the actual scenario the FIXME
+  // referred to -- previously the non-OK status from the resolve path was
+  // dropped. The failure must propagate and the raw blob-index bytes must be
+  // cleared rather than handed back as if they were the value.
+  {
+    std::string blob_index;
+    BlobIndex::EncodeBlob(&blob_index, /*file_number=*/123, /*offset=*/456,
+                          /*size=*/789, kNoCompression);
+
+    PinnableSlice value;
+    value.GetSelf()->assign(blob_index.data(), blob_index.size());
+    value.PinSelf();
+
+    bool is_blob_index = true;
+    Status s = PostprocessMemtableValueReadForTest(
+        Slice("key"), /*resolve_blob_backed_memtable_value=*/true,
+        /*blob_fetcher=*/nullptr, &value, /*columns=*/nullptr, Status::OK(),
+        &is_blob_index);
+
+    ASSERT_TRUE(s.IsCorruption()) << s.ToString();
+    ASSERT_TRUE(value.empty());
+  }
+}
+
 TEST_F(DBBlobIndexTest, ReadOnlyGetImplReturnsBlobIndexWhenRequested) {
   // Goal: cover the internal read-only GetImpl contract when the caller
   // explicitly asks for raw blob-index bytes via `is_blob_index`. Recovery
@@ -330,6 +403,671 @@ TEST_F(DBBlobIndexTest, ReadOnlyGetImplReturnsBlobIndexWhenRequested) {
   ASSERT_TRUE(is_blob_index);
 }
 
+// Regression test for a same-file (embedded) BlobIndex being exposed
+// unresolved through an iterator. With index_type=kBinarySearchWithFirstKey
+// and allow_unprepared_value (the default for DB iterators), the block-based
+// table iterator can sit in the is_at_first_key_from_index_ state and return
+// the raw kTypeBlobIndex internal key from the index, while value() resolves
+// the same-file blob to its plain payload. DBIter then reads the stale
+// kTypeBlobIndex type and routes the already-resolved plain value through the
+// blob-index path, corrupting/misreading it.
+//
+// Forcing one entry per data block makes every embedded blob the first key of
+// a block, reliably triggering the deferred-first-key state on both SeekToFirst
+// and forward block transitions.
+TEST_F(DBBlobIndexTest, EmbeddedBlobIteratorWithFirstKeyIndex) {
+  Options options = GetTestOptions();
+  options.create_if_missing = true;
+  BlockBasedTableOptions bbto;
+  bbto.index_type =
+      BlockBasedTableOptions::IndexType::kBinarySearchWithFirstKey;
+  // Soft limit of 1 byte starts a new data block after every entry.
+  bbto.block_size = 1;
+  options.table_factory.reset(NewBlockBasedTableFactory(bbto));
+  DestroyAndReopen(options);
+
+  // Build an embedded-blob SST whose large values are stored as same-file blob
+  // records (table entries become same-file BlobIndex references).
+  const std::string sst_path = dbname_ + "/embedded_first_key.sst";
+  SstFileWriterEmbeddedBlobOptions embedded_blob_options;
+  embedded_blob_options.min_blob_size = 8;
+
+  const std::string big1(1024, 'x');
+  const std::string big2(1024, 'y');
+  const std::string big3(1024, 'z');
+
+  SstFileWriter writer(EnvOptions(), options);
+  ASSERT_OK(writer.OpenWithEmbeddedBlobs(sst_path, embedded_blob_options));
+  ASSERT_OK(writer.Put("k1", big1));
+  ASSERT_OK(writer.Put("k2", big2));
+  ASSERT_OK(writer.Put("k3", big3));
+  ASSERT_OK(writer.Finish());
+
+  ASSERT_OK(db_->IngestExternalFile({sst_path}, IngestExternalFileOptions()));
+
+  // Point lookups resolve same-file blobs before exposing them, so Get works.
+  ASSERT_EQ(Get("k1"), big1);
+  ASSERT_EQ(Get("k2"), big2);
+  ASSERT_EQ(Get("k3"), big3);
+
+  // Iterator path: the bug manifests here.
+  std::unique_ptr<Iterator> iter(db_->NewIterator(ReadOptions()));
+  iter->SeekToFirst();
+  ASSERT_TRUE(iter->Valid());
+  ASSERT_OK(iter->status());
+  EXPECT_EQ(iter->key(), "k1");
+  EXPECT_EQ(iter->value(), big1);
+
+  iter->Next();
+  ASSERT_TRUE(iter->Valid());
+  ASSERT_OK(iter->status());
+  EXPECT_EQ(iter->key(), "k2");
+  EXPECT_EQ(iter->value(), big2);
+
+  iter->Next();
+  ASSERT_TRUE(iter->Valid());
+  ASSERT_OK(iter->status());
+  EXPECT_EQ(iter->key(), "k3");
+  EXPECT_EQ(iter->value(), big3);
+
+  iter->Next();
+  ASSERT_FALSE(iter->Valid());
+  ASSERT_OK(iter->status());
+
+  // Seek directly to a block whose first key is an embedded blob.
+  iter->Seek("k2");
+  ASSERT_TRUE(iter->Valid());
+  ASSERT_OK(iter->status());
+  EXPECT_EQ(iter->key(), "k2");
+  EXPECT_EQ(iter->value(), big2);
+}
+
+// Backward iteration (Prev / SeekForPrev) over an embedded-blob SST must work,
+// including for wide-column entities whose same-file blob columns are resolved
+// into an iterator-owned buffer. DBIter requires the underlying iterator's
+// value to be pinned for backward iteration; EmbeddedBlobResolvingIterator
+// therefore pins resolved values -- both whole-value blob payloads and rebuilt
+// wide-column values -- and hands their cleanup to the PinnedIteratorsManager
+// across repositioning. Before that, backward iteration over the wide-column
+// entity returned NotSupported (or tripped the IsValuePinned() assertion).
+TEST_F(DBBlobIndexTest, EmbeddedBlobBackwardIteration) {
+  Options options = GetTestOptions();
+  options.create_if_missing = true;
+  DestroyAndReopen(options);
+
+  const std::string sst_path = dbname_ + "/embedded_backward.sst";
+  SstFileWriterEmbeddedBlobOptions embedded_blob_options;
+  embedded_blob_options.min_blob_size = 8;
+
+  // Whole-value same-file blobs.
+  const std::string big1(1024, 'x');
+  const std::string big3(1024, 'z');
+  // Wide-column entity: a small (inline) default column plus a large
+  // (embedded, same-file blob) non-default column -- the mixed wide-column
+  // path whose resolved value lands in an iterator-owned buffer.
+  const std::string k2_default(2, 'd');
+  const std::string k2_big_col(1024, 'c');
+  const WideColumns k2_columns{{kDefaultWideColumnName, k2_default},
+                               {"big", k2_big_col}};
+
+  SstFileWriter writer(EnvOptions(), options);
+  ASSERT_OK(writer.OpenWithEmbeddedBlobs(sst_path, embedded_blob_options));
+  ASSERT_OK(writer.Put("k1", big1));
+  ASSERT_OK(writer.PutEntity("k2", k2_columns));
+  ASSERT_OK(writer.Put("k3", big3));
+  ASSERT_OK(writer.Finish());
+
+  ASSERT_OK(db_->IngestExternalFile({sst_path}, IngestExternalFileOptions()));
+
+  std::unique_ptr<Iterator> iter(db_->NewIterator(ReadOptions()));
+
+  // Backward scan via SeekToLast + Prev.
+  iter->SeekToLast();
+  ASSERT_TRUE(iter->Valid());
+  ASSERT_OK(iter->status());
+  EXPECT_EQ(iter->key(), "k3");
+  EXPECT_EQ(iter->value(), big3);
+
+  iter->Prev();
+  ASSERT_TRUE(iter->Valid());
+  ASSERT_OK(iter->status());
+  EXPECT_EQ(iter->key(), "k2");
+  EXPECT_EQ(iter->value(), k2_default);
+  EXPECT_EQ(iter->columns(), k2_columns);
+
+  iter->Prev();
+  ASSERT_TRUE(iter->Valid());
+  ASSERT_OK(iter->status());
+  EXPECT_EQ(iter->key(), "k1");
+  EXPECT_EQ(iter->value(), big1);
+
+  iter->Prev();
+  ASSERT_FALSE(iter->Valid());
+  ASSERT_OK(iter->status());
+
+  // SeekForPrev directly onto the wide-column entity, then step back.
+  iter->SeekForPrev("k2");
+  ASSERT_TRUE(iter->Valid());
+  ASSERT_OK(iter->status());
+  EXPECT_EQ(iter->key(), "k2");
+  EXPECT_EQ(iter->value(), k2_default);
+  EXPECT_EQ(iter->columns(), k2_columns);
+
+  iter->Prev();
+  ASSERT_TRUE(iter->Valid());
+  ASSERT_OK(iter->status());
+  EXPECT_EQ(iter->key(), "k1");
+  EXPECT_EQ(iter->value(), big1);
+}
+
+// Regression test for T277566778. If resolving a same-file ("embedded") blob
+// fails during compaction (e.g. a blob-region read fault), the error must be
+// surfaced as-is. It must NOT be masked by feeding the raw, unresolved
+// same-file BlobIndex downstream, where FileMetaData::UpdateBoundaries would
+// (correctly) reject file number 0 and report a misleading "Invalid blob file
+// number" corruption. The EmbeddedBlobResolvingIterator must not expose an
+// unresolved same-file value on error: for eager (compaction) callers it
+// resolves during positioning so the error is visible via status()/Valid()
+// before value().
+TEST_F(DBBlobIndexTest, EmbeddedBlobResolveErrorDuringCompactionNotMasked) {
+  Options options = GetTestOptions();
+  options.create_if_missing = true;
+  options.disable_auto_compactions = true;
+  DestroyAndReopen(options);
+
+  const std::string sst_path = dbname_ + "/embedded_resolve_err.sst";
+  SstFileWriterEmbeddedBlobOptions embedded_blob_options;
+  embedded_blob_options.min_blob_size = 8;
+
+  // Wide-column entity with a large (embedded, same-file blob) column: the
+  // mixed path whose resolution reads the blob region during compaction.
+  const std::string big_col(1024, 'c');
+  const WideColumns columns{{kDefaultWideColumnName, "d"}, {"big", big_col}};
+
+  SstFileWriter writer(EnvOptions(), options);
+  ASSERT_OK(writer.OpenWithEmbeddedBlobs(sst_path, embedded_blob_options));
+  ASSERT_OK(writer.PutEntity("k2", columns));
+  ASSERT_OK(writer.Finish());
+  ASSERT_OK(db_->IngestExternalFile({sst_path}, IngestExternalFileOptions()));
+
+  // Add keys straddling the entity so a real compaction (not a trivial move)
+  // reads and resolves the embedded entity.
+  ASSERT_OK(Put("k1", "v1"));
+  ASSERT_OK(Put("k3", "v3"));
+  ASSERT_OK(Flush());
+
+  // Inject a resolution failure only during the compaction below, simulating a
+  // blob-region read fault.
+  const Status kInjected = Status::IOError("injected embedded blob read error");
+  SyncPoint::GetInstance()->SetCallBack(
+      "BlockBasedTable::MaybeResolveEmbeddedValue:InjectError",
+      [&](void* arg) { *static_cast<Status*>(arg) = kInjected; });
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  const Status s = db_->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+
+  SyncPoint::GetInstance()->DisableProcessing();
+  SyncPoint::GetInstance()->ClearAllCallBacks();
+
+  // The compaction must fail with the injected read error, NOT a masked
+  // "Invalid blob file number" corruption.
+  ASSERT_NOK(s);
+  EXPECT_TRUE(s.IsIOError()) << s.ToString();
+  EXPECT_EQ(s.ToString().find("Invalid blob file number"), std::string::npos)
+      << s.ToString();
+}
+
+// Regression test for T277310719, the whole-value (kTypeBlobIndex) sibling of
+// the wide-column-entity case above (T277566778). Both share one root cause: on
+// a blob-region resolution error during compaction, the
+// EmbeddedBlobResolvingIterator must not fall back to exposing the raw,
+// unresolved same-file value.
+//
+// The whole-value variant is more dangerous than the entity variant because it
+// escapes the FileMetaData::UpdateBoundaries tripwire: ResolveKeyType()
+// rewrites the key type kTypeBlobIndex -> kTypeValue (no I/O, always succeeds),
+// so on a masked resolution error the emitted entry is {kTypeValue, raw
+// BlobIndex bytes}. UpdateBoundaries only scans kTypeBlobIndex/entity types, so
+// unlike the entity variant it does NOT reject it -- the masked error slips
+// through, a corrupt {kTypeValue, BlobIndex} record is written to the
+// compaction output, and a later point lookup reads those raw BlobIndex bytes
+// back as the value (in db_stress this surfaced as the
+// ExpectedValue::IsValueBaseValid assertion, not "Invalid blob file number").
+// The eager (compaction) resolver must surface the error via status()/Valid()
+// before value() is consumed.
+TEST_F(DBBlobIndexTest,
+       EmbeddedBlobResolveErrorWholeValueDuringCompactionNotMasked) {
+  Options options = GetTestOptions();
+  options.create_if_missing = true;
+  options.disable_auto_compactions = true;
+  DestroyAndReopen(options);
+
+  const std::string sst_path = dbname_ + "/embedded_resolve_err_wholevalue.sst";
+  SstFileWriterEmbeddedBlobOptions embedded_blob_options;
+  embedded_blob_options.min_blob_size = 8;
+
+  // A large whole value is stored as a same-file blob, so the table entry
+  // becomes a same-file (file number 0) BlobIndex (kTypeBlobIndex) -- the
+  // whole-value path whose resolution reads the blob region during compaction.
+  const std::string big_val(1024, 'b');
+
+  SstFileWriter writer(EnvOptions(), options);
+  ASSERT_OK(writer.OpenWithEmbeddedBlobs(sst_path, embedded_blob_options));
+  ASSERT_OK(writer.Put("k2", big_val));
+  ASSERT_OK(writer.Finish());
+  ASSERT_OK(db_->IngestExternalFile({sst_path}, IngestExternalFileOptions()));
+
+  // Add keys straddling the embedded blob so a real compaction (not a trivial
+  // move) reads and resolves it.
+  ASSERT_OK(Put("k1", "v1"));
+  ASSERT_OK(Put("k3", "v3"));
+  ASSERT_OK(Flush());
+
+  // Inject a resolution failure only during the compaction below, simulating a
+  // blob-region read fault.
+  const Status kInjected = Status::IOError("injected embedded blob read error");
+  SyncPoint::GetInstance()->SetCallBack(
+      "BlockBasedTable::MaybeResolveEmbeddedValue:InjectError",
+      [&](void* arg) { *static_cast<Status*>(arg) = kInjected; });
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  const Status s = db_->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+
+  SyncPoint::GetInstance()->DisableProcessing();
+  SyncPoint::GetInstance()->ClearAllCallBacks();
+
+  // The compaction must fail with the injected read error. Before the fix it
+  // instead succeeded (masking the error) and persisted a {kTypeValue, raw
+  // BlobIndex} record; it must not do that, and it must not report a misleading
+  // "Invalid blob file number" corruption either.
+  ASSERT_NOK(s);
+  EXPECT_TRUE(s.IsIOError()) << s.ToString();
+  EXPECT_EQ(s.ToString().find("Invalid blob file number"), std::string::npos)
+      << s.ToString();
+}
+
+// When a blob cache is configured, same-file ("embedded") blob reads should go
+// through BlobSource: the first read misses + inserts + does a disk read, and
+// the second read of the same blob is served from the blob cache. This holds on
+// both the Get() and the iterator paths.
+TEST_F(DBBlobIndexTest, EmbeddedBlobBlobCacheHitMiss) {
+  Options options = GetTestOptions();
+  options.create_if_missing = true;
+  options.statistics = CreateDBStatistics();
+
+  LRUCacheOptions co;
+  co.capacity = 8 << 20;
+  options.blob_cache = NewLRUCache(co);
+
+  DestroyAndReopen(options);
+
+  const std::string sst_path = dbname_ + "/embedded_cache.sst";
+  SstFileWriterEmbeddedBlobOptions embedded_blob_options;
+  embedded_blob_options.min_blob_size = 8;
+
+  const std::string big1(1024, 'x');
+  const std::string big2(1024, 'y');
+
+  SstFileWriter writer(EnvOptions(), options);
+  ASSERT_OK(writer.OpenWithEmbeddedBlobs(sst_path, embedded_blob_options));
+  ASSERT_OK(writer.Put("k1", big1));
+  ASSERT_OK(writer.Put("k2", big2));
+  ASSERT_OK(writer.Finish());
+
+  ASSERT_OK(db_->IngestExternalFile({sst_path}, IngestExternalFileOptions()));
+
+  Statistics* const stats = options.statistics.get();
+
+  // First Get: blob cache miss -> disk read -> cache insert.
+  ASSERT_OK(stats->Reset());
+  get_perf_context()->Reset();
+  SetPerfLevel(kEnableCount);
+  ASSERT_EQ(Get("k1"), big1);
+  EXPECT_EQ(stats->getTickerCount(BLOB_DB_CACHE_MISS), 1);
+  EXPECT_EQ(stats->getTickerCount(BLOB_DB_CACHE_HIT), 0);
+  EXPECT_EQ(stats->getTickerCount(BLOB_DB_CACHE_ADD), 1);
+  EXPECT_GT(stats->getTickerCount(BLOB_DB_BLOB_FILE_BYTES_READ), 0);
+  EXPECT_EQ(get_perf_context()->blob_read_count, 1);
+
+  // Second Get of the same key: blob cache hit, no disk read, no insert.
+  ASSERT_OK(stats->Reset());
+  get_perf_context()->Reset();
+  ASSERT_EQ(Get("k1"), big1);
+  EXPECT_EQ(stats->getTickerCount(BLOB_DB_CACHE_HIT), 1);
+  EXPECT_EQ(stats->getTickerCount(BLOB_DB_CACHE_MISS), 0);
+  EXPECT_EQ(stats->getTickerCount(BLOB_DB_CACHE_ADD), 0);
+  EXPECT_EQ(stats->getTickerCount(BLOB_DB_BLOB_FILE_BYTES_READ), 0);
+  EXPECT_EQ(get_perf_context()->blob_read_count, 0);
+  EXPECT_EQ(get_perf_context()->blob_cache_hit_count, 1);
+
+  // Iterator path: first read of a fresh key misses + inserts.
+  ASSERT_OK(stats->Reset());
+  {
+    std::unique_ptr<Iterator> iter(db_->NewIterator(ReadOptions()));
+    iter->Seek("k2");
+    ASSERT_TRUE(iter->Valid());
+    ASSERT_OK(iter->status());
+    EXPECT_EQ(iter->value(), big2);
+  }
+  EXPECT_EQ(stats->getTickerCount(BLOB_DB_CACHE_MISS), 1);
+  EXPECT_EQ(stats->getTickerCount(BLOB_DB_CACHE_ADD), 1);
+
+  // Iterator path: second read of the same key hits the blob cache.
+  ASSERT_OK(stats->Reset());
+  {
+    std::unique_ptr<Iterator> iter(db_->NewIterator(ReadOptions()));
+    iter->Seek("k2");
+    ASSERT_TRUE(iter->Valid());
+    ASSERT_OK(iter->status());
+    EXPECT_EQ(iter->value(), big2);
+  }
+  EXPECT_EQ(stats->getTickerCount(BLOB_DB_CACHE_HIT), 1);
+  EXPECT_EQ(stats->getTickerCount(BLOB_DB_CACHE_ADD), 0);
+
+  SetPerfLevel(kDisable);
+}
+
+// With the blob cache and block cache pointing at the same Cache, the SST-like
+// cache-key scheme (offset >> 2 on the SST's base key) keeps embedded blob
+// records disjoint from the SST's data blocks. Reading embedded blobs must not
+// alias or evict data blocks, and values must remain correct.
+TEST_F(DBBlobIndexTest, EmbeddedBlobSharedBlockAndBlobCache) {
+  Options options = GetTestOptions();
+  options.create_if_missing = true;
+  options.statistics = CreateDBStatistics();
+
+  LRUCacheOptions co;
+  co.capacity = 16 << 20;
+  auto shared_cache = NewLRUCache(co);
+  options.blob_cache = shared_cache;
+
+  BlockBasedTableOptions bbto;
+  bbto.block_cache = shared_cache;
+  bbto.cache_index_and_filter_blocks = true;
+  options.table_factory.reset(NewBlockBasedTableFactory(bbto));
+
+  DestroyAndReopen(options);
+
+  const std::string sst_path = dbname_ + "/embedded_shared.sst";
+  SstFileWriterEmbeddedBlobOptions embedded_blob_options;
+  embedded_blob_options.min_blob_size = 8;
+
+  const std::string big1(1024, 'x');
+  const std::string big2(1024, 'y');
+
+  SstFileWriter writer(EnvOptions(), options);
+  ASSERT_OK(writer.OpenWithEmbeddedBlobs(sst_path, embedded_blob_options));
+  ASSERT_OK(writer.Put("k1", big1));
+  ASSERT_OK(writer.Put("k2", big2));
+  ASSERT_OK(writer.Finish());
+
+  ASSERT_OK(db_->IngestExternalFile({sst_path}, IngestExternalFileOptions()));
+
+  Statistics* const stats = options.statistics.get();
+
+  // Values are correct with a shared cache, and re-reads hit both the block
+  // cache (data blocks) and the blob cache (embedded payloads) independently.
+  ASSERT_EQ(Get("k1"), big1);
+  ASSERT_EQ(Get("k2"), big2);
+  ASSERT_EQ(Get("k1"), big1);
+  ASSERT_EQ(Get("k2"), big2);
+  EXPECT_GT(stats->getTickerCount(BLOB_DB_CACHE_HIT), 0);
+  EXPECT_GT(stats->getTickerCount(BLOCK_CACHE_HIT), 0);
+
+  // Iterator yields correct values too.
+  std::unique_ptr<Iterator> iter(db_->NewIterator(ReadOptions()));
+  iter->SeekToFirst();
+  ASSERT_TRUE(iter->Valid());
+  ASSERT_OK(iter->status());
+  EXPECT_EQ(iter->key(), "k1");
+  EXPECT_EQ(iter->value(), big1);
+
+  iter->Next();
+  ASSERT_TRUE(iter->Valid());
+  ASSERT_OK(iter->status());
+  EXPECT_EQ(iter->key(), "k2");
+  EXPECT_EQ(iter->value(), big2);
+
+  iter->Next();
+  ASSERT_FALSE(iter->Valid());
+  ASSERT_OK(iter->status());
+}
+
+// Point GetEntity / MultiGetEntity of a wide-column entity whose non-default
+// columns are embedded (same-file) blobs must return the correct columns and
+// resolve them zero-copy: each resolved blob column value points straight at
+// its pinned blob-cache buffer (identical across reads, and in a separate
+// buffer per column), rather than being copied into one re-serialized entity.
+TEST_F(DBBlobIndexTest, EmbeddedBlobWideColumnGetEntityZeroCopy) {
+  Options options = GetTestOptions();
+  options.create_if_missing = true;
+  options.statistics = CreateDBStatistics();
+  LRUCacheOptions co;
+  co.capacity = 8 << 20;
+  options.blob_cache = NewLRUCache(co);
+  DestroyAndReopen(options);
+
+  const std::string sst_path = dbname_ + "/embedded_wc_zerocopy.sst";
+  SstFileWriterEmbeddedBlobOptions embedded_blob_options;
+  embedded_blob_options.min_blob_size = 8;
+
+  const std::string small_default = "d";  // inline default column
+  const std::string big1(1024, 'a');      // same-file blob column
+  const std::string big2(2048, 'b');      // same-file blob column
+  const WideColumns columns{
+      {kDefaultWideColumnName, small_default}, {"big1", big1}, {"big2", big2}};
+
+  SstFileWriter writer(EnvOptions(), options);
+  ASSERT_OK(writer.OpenWithEmbeddedBlobs(sst_path, embedded_blob_options));
+  ASSERT_OK(writer.PutEntity("k", columns));
+  ASSERT_OK(writer.Finish());
+  ASSERT_OK(db_->IngestExternalFile({sst_path}, IngestExternalFileOptions()));
+
+  const char* big1_ptr = nullptr;
+  const char* big2_ptr = nullptr;
+  {
+    PinnableWideColumns result;
+    ASSERT_OK(db_->GetEntity(ReadOptions(), db_->DefaultColumnFamily(), "k",
+                             &result));
+    ASSERT_EQ(result.columns(), columns);
+
+    // columns() is sorted by name -> [default(""), big1, big2].
+    big1_ptr = result.columns()[1].value().data();
+    big2_ptr = result.columns()[2].value().data();
+    // Multi-buffer: each blob column lives in its own backing buffer, so the
+    // two values are not laid out contiguously (as they would be in a single
+    // re-serialized V1 entity).
+    ASSERT_NE(big1_ptr + big1.size(), big2_ptr);
+
+    // Pointer stability across move (address-stable backing nodes).
+    PinnableWideColumns moved(std::move(result));
+    ASSERT_EQ(moved.columns(), columns);
+    ASSERT_EQ(moved.columns()[1].value().data(), big1_ptr);
+    ASSERT_EQ(moved.columns()[2].value().data(), big2_ptr);
+  }
+
+  // A second read serves the same pinned blob-cache buffers (zero-copy): the
+  // resolved column values point at the very same addresses, which a per-read
+  // re-serialization would not.
+  {
+    PinnableWideColumns result;
+    ASSERT_OK(db_->GetEntity(ReadOptions(), db_->DefaultColumnFamily(), "k",
+                             &result));
+    ASSERT_EQ(result.columns(), columns);
+    ASSERT_EQ(result.columns()[1].value().data(), big1_ptr);
+    ASSERT_EQ(result.columns()[2].value().data(), big2_ptr);
+  }
+
+  // MultiGetEntity: same correctness and zero-copy (cache-hit) guarantee.
+  {
+    const std::array<Slice, 1> keys{Slice("k")};
+    std::array<PinnableWideColumns, 1> results;
+    std::array<Status, 1> statuses;
+    db_->MultiGetEntity(ReadOptions(), db_->DefaultColumnFamily(), keys.size(),
+                        keys.data(), results.data(), statuses.data());
+    ASSERT_OK(statuses[0]);
+    ASSERT_EQ(results[0].columns(), columns);
+    ASSERT_EQ(results[0].columns()[1].value().data(), big1_ptr);
+    ASSERT_EQ(results[0].columns()[2].value().data(), big2_ptr);
+  }
+}
+
+// A wide-column entity whose default column is itself an embedded (same-file)
+// blob must resolve correctly on the plain Get path (default-column
+// extraction) as well as GetEntity, matching the values written.
+TEST_F(DBBlobIndexTest, EmbeddedBlobWideColumnGetDefaultColumn) {
+  Options options = GetTestOptions();
+  options.create_if_missing = true;
+  DestroyAndReopen(options);
+
+  const std::string sst_path = dbname_ + "/embedded_wc_default.sst";
+  SstFileWriterEmbeddedBlobOptions embedded_blob_options;
+  embedded_blob_options.min_blob_size = 8;
+
+  const std::string big_default(1024, 'x');  // same-file blob default column
+  const std::string small_attr = "s";        // inline non-default column
+  const WideColumns columns{{kDefaultWideColumnName, big_default},
+                            {"attr", small_attr}};
+
+  SstFileWriter writer(EnvOptions(), options);
+  ASSERT_OK(writer.OpenWithEmbeddedBlobs(sst_path, embedded_blob_options));
+  ASSERT_OK(writer.PutEntity("k", columns));
+  ASSERT_OK(writer.Finish());
+  ASSERT_OK(db_->IngestExternalFile({sst_path}, IngestExternalFileOptions()));
+
+  // Plain Get returns the (blob-backed) default column value.
+  PinnableSlice value;
+  ASSERT_OK(db_->Get(ReadOptions(), db_->DefaultColumnFamily(), "k", &value));
+  ASSERT_EQ(value, big_default);
+
+  // GetEntity returns all columns.
+  PinnableWideColumns result;
+  ASSERT_OK(
+      db_->GetEntity(ReadOptions(), db_->DefaultColumnFamily(), "k", &result));
+  ASSERT_EQ(result.columns(), columns);
+}
+
+// Merge over an embedded-blob wide-column base entity must resolve the base
+// entity's same-file default column before merging, and GetMergeOperands must
+// return the resolved default column as the base operand.
+TEST_F(DBBlobIndexTest, EmbeddedBlobWideColumnMergeBase) {
+  Options options = GetTestOptions();  // merge_operator = StringAppend(',')
+  options.create_if_missing = true;
+  DestroyAndReopen(options);
+
+  const std::string sst_path = dbname_ + "/embedded_wc_merge.sst";
+  SstFileWriterEmbeddedBlobOptions embedded_blob_options;
+  embedded_blob_options.min_blob_size = 8;
+
+  const std::string big_default(1024, 'D');  // same-file blob default column
+  const WideColumns columns{{kDefaultWideColumnName, big_default},
+                            {"attr", "s"}};
+
+  SstFileWriter writer(EnvOptions(), options);
+  ASSERT_OK(writer.OpenWithEmbeddedBlobs(sst_path, embedded_blob_options));
+  ASSERT_OK(writer.PutEntity("k", columns));
+  ASSERT_OK(writer.Finish());
+  ASSERT_OK(db_->IngestExternalFile({sst_path}, IngestExternalFileOptions()));
+
+  // Newer merge operand lives in the memtable; the embedded entity base lives
+  // in the ingested SST, so the merge resolves the base via the same-file path.
+  ASSERT_OK(db_->Merge(WriteOptions(), db_->DefaultColumnFamily(), "k", "op1"));
+
+  PinnableSlice value;
+  ASSERT_OK(db_->Get(ReadOptions(), db_->DefaultColumnFamily(), "k", &value));
+  ASSERT_EQ(value, big_default + "," + "op1");
+
+  // GetMergeOperands returns the resolved base default column plus the operand.
+  std::array<PinnableSlice, 8> operands;
+  GetMergeOperandsOptions get_merge_opts;
+  get_merge_opts.expected_max_number_of_operands =
+      static_cast<int>(operands.size());
+  int num_operands = 0;
+  ASSERT_OK(db_->GetMergeOperands(ReadOptions(), db_->DefaultColumnFamily(),
+                                  "k", operands.data(), &get_merge_opts,
+                                  &num_operands));
+  ASSERT_EQ(num_operands, 2);
+  ASSERT_EQ(operands[0], big_default);
+  ASSERT_EQ(operands[1], "op1");
+}
+
+// With read_tier == kBlockCacheTier, an unresolved same-file blob column of a
+// wide-column entity must surface Incomplete (never a raw same-file BlobIndex).
+TEST_F(DBBlobIndexTest, EmbeddedBlobWideColumnGetEntityBlockCacheTier) {
+  Options options = GetTestOptions();
+  options.create_if_missing = true;
+  DestroyAndReopen(options);
+
+  const std::string sst_path = dbname_ + "/embedded_wc_blockcache.sst";
+  SstFileWriterEmbeddedBlobOptions embedded_blob_options;
+  embedded_blob_options.min_blob_size = 8;
+
+  const std::string big1(1024, 'a');
+  const WideColumns columns{{kDefaultWideColumnName, "d"}, {"big", big1}};
+
+  SstFileWriter writer(EnvOptions(), options);
+  ASSERT_OK(writer.OpenWithEmbeddedBlobs(sst_path, embedded_blob_options));
+  ASSERT_OK(writer.PutEntity("k", columns));
+  ASSERT_OK(writer.Finish());
+  ASSERT_OK(db_->IngestExternalFile({sst_path}, IngestExternalFileOptions()));
+
+  // Warm the block cache so the data block read succeeds under kBlockCacheTier
+  // and the lookup actually reaches same-file blob column resolution.
+  {
+    PinnableWideColumns result;
+    ASSERT_OK(db_->GetEntity(ReadOptions(), db_->DefaultColumnFamily(), "k",
+                             &result));
+    ASSERT_EQ(result.columns(), columns);
+  }
+
+  ReadOptions read_options;
+  read_options.read_tier = kBlockCacheTier;
+  PinnableWideColumns result;
+  Status s =
+      db_->GetEntity(read_options, db_->DefaultColumnFamily(), "k", &result);
+  ASSERT_TRUE(s.IsIncomplete()) << s.ToString();
+}
+
+// A corrupt embedded blob record backing a wide-column column must surface an
+// error on GetEntity, not expose the raw same-file BlobIndex.
+// A failure resolving an embedded blob record backing a wide-column column
+// (e.g. a bad record or blob-region read fault) must surface on GetEntity, not
+// expose the raw same-file BlobIndex. The failure is injected via a sync point
+// so the test is independent of the on-disk byte layout (e.g. encrypted env).
+TEST_F(DBBlobIndexTest, EmbeddedBlobWideColumnGetEntityCorruptionSurfaced) {
+  Options options = GetTestOptions();
+  options.create_if_missing = true;
+  DestroyAndReopen(options);
+
+  const std::string sst_path = dbname_ + "/embedded_wc_corrupt.sst";
+  SstFileWriterEmbeddedBlobOptions embedded_blob_options;
+  embedded_blob_options.min_blob_size = 8;
+
+  const std::string big_col(1024, 'Z');
+  const WideColumns columns{{kDefaultWideColumnName, "d"}, {"big", big_col}};
+
+  SstFileWriter writer(EnvOptions(), options);
+  ASSERT_OK(writer.OpenWithEmbeddedBlobs(sst_path, embedded_blob_options));
+  ASSERT_OK(writer.PutEntity("k", columns));
+  ASSERT_OK(writer.Finish());
+  ASSERT_OK(db_->IngestExternalFile({sst_path}, IngestExternalFileOptions()));
+
+  const Status kInjected = Status::Corruption("injected embedded blob error");
+  SyncPoint::GetInstance()->SetCallBack(
+      "BlockBasedTable::MaybeResolveEmbeddedValue:InjectError",
+      [&](void* arg) { *static_cast<Status*>(arg) = kInjected; });
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  PinnableWideColumns result;
+  const Status s =
+      db_->GetEntity(ReadOptions(), db_->DefaultColumnFamily(), "k", &result);
+
+  SyncPoint::GetInstance()->DisableProcessing();
+  SyncPoint::GetInstance()->ClearAllCallBacks();
+
+  ASSERT_TRUE(s.IsCorruption()) << s.ToString();
+}
+
 class PlainBlobValueFilterV3 : public CompactionFilter {
  public:
   PlainBlobValueFilterV3(std::atomic<int>* filter_call_count,

db/blob/same_file_blob_reader.h

@@ -0,0 +1,39 @@
+//  Copyright (c) Meta Platforms, Inc. and affiliates.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+#pragma once
+
+#include "rocksdb/rocksdb_namespace.h"
+#include "rocksdb/status.h"
+
+namespace ROCKSDB_NAMESPACE {
+
+class BlobIndex;
+class PinnableSlice;
+struct ReadOptions;
+
+// Narrow interface for resolving a same-file ("embedded") blob reference
+// against the physical file that currently holds the table entry. Implemented
+// by BlockBasedTable, which owns the reader context (file, footer, blob cache)
+// needed to read an embedded blob record.
+//
+// Kept in its own low-level header, free of any table-layer dependency, so that
+// db/blob (EmbeddedAwareBlobFetcher) and table/get_context can depend on just
+// this interface without pulling in BlockBasedTable. ReadPathBlobResolver (the
+// lazy-resolution read path) can likewise hold a const SameFileBlobReader*.
+class SameFileBlobReader {
+ public:
+  virtual ~SameFileBlobReader() = default;
+
+  // Resolves a same-file blob reference (one for which blob_index.IsSameFile()
+  // is true) into `value`, pinning the payload zero-copy (from the blob cache
+  // or an owned buffer). read_tier == kBlockCacheTier without a cached record
+  // yields Status::Incomplete; a bad record yields Status::Corruption.
+  virtual Status GetSameFileBlob(const ReadOptions& read_options,
+                                 const BlobIndex& blob_index,
+                                 PinnableSlice* value) const = 0;
+};
+
+}  // namespace ROCKSDB_NAMESPACE

db/builder.cc

@@ -179,6 +179,8 @@ Status BuildTable(
       TEST_SYNC_POINT_CALLBACK("BuildTable:create_file", &use_direct_writes);
 #endif  // !NDEBUG
       FileOptions fo_copy = file_options;
+      fo_copy.open_contract = FileOpenContract::kNoReopenForWrite |
+                              FileOpenContract::kNoReadersWhileOpenForWrite;
       fo_copy.write_hint = write_hint;
       IOStatus io_s = NewWritableFile(fs, fname, &file, fo_copy);
       assert(s.ok());

db/column_family.cc

@@ -678,6 +678,9 @@ ColumnFamilyData::ColumnFamilyData(
                           internal_stats_->GetBlobFileReadHist(), io_tracer));
     blob_source_.reset(new BlobSource(ioptions_, mutable_cf_options_, db_id,
                                       db_session_id, blob_file_cache_.get()));
+    // Let the table cache route same-file ("embedded") blob reads through the
+    // blob value cache + stats. Both objects share this CFD's lifetime.
+    table_cache_->SetBlobSource(blob_source_.get());
 
     if (ioptions_.compaction_style == kCompactionStyleLevel) {
       compaction_picker_.reset(

db/compaction/compaction_iterator.cc

@@ -251,8 +251,9 @@ void CompactionIterator::SetBlobFetcher(const Version* version,
   read_options.io_activity = io_activity;
   read_options.fill_cache = false;
 
-  blob_fetcher_ = std::make_unique<BlobFetcher>(
-      version, read_options, blob_file_cache, allow_write_path_fallback);
+  blob_fetcher_ = std::make_unique<OwningVersionBlobFetcher>(
+      version, std::move(read_options), blob_file_cache,
+      allow_write_path_fallback);
   blob_resolver_.Init(blob_fetcher_.get(), prefetch_buffers_.get(),
                       &iter_stats_);
 }
@@ -467,14 +468,14 @@ bool CompactionIterator::InvokeFilterIfNeeded(bool* need_skip,
           }
         }
         if (UNLIKELY(has_blob_columns)) {
-          // Entity has blob columns. DeserializeV2() populates
+          // Entity has blob columns. Deserialize() populates
           // entity_columns_ with the full column set; blob-backed entries still
           // carry the serialized BlobIndex bytes in value(). The companion
           // entity_blob_columns_ side list identifies which column indexes are
           // blob references and provides the decoded BlobIndex objects.
           Slice input_copy = value_;
-          Status s = WideColumnSerialization::DeserializeV2(
-              input_copy, entity_columns_, entity_blob_columns_);
+          Status s = WideColumnSerialization::Deserialize(
+              input_copy, entity_columns_, &entity_blob_columns_);
           if (!s.ok()) {
             status_ = s;
             validity_info_.Invalidate();
@@ -541,7 +542,7 @@ bool CompactionIterator::InvokeFilterIfNeeded(bool* need_skip,
         } else {
           // No blob columns, use fast path
           Slice value_copy = value_;
-          const Status s = WideColumnSerialization::Deserialize(
+          const Status s = WideColumnSerialization::DeserializeSimple(
               value_copy, filter_existing_columns_);
 
           if (!s.ok()) {
@@ -1550,8 +1551,8 @@ void CompactionIterator::ExtractLargeColumnValuesIfNeeded() {
     Slice entity_slice = value_;
     entity_columns_.clear();
     entity_blob_columns_.clear();
-    Status s = WideColumnSerialization::DeserializeV2(
-        entity_slice, entity_columns_, entity_blob_columns_);
+    Status s = WideColumnSerialization::Deserialize(
+        entity_slice, entity_columns_, &entity_blob_columns_);
     if (!s.ok()) {
       status_ = s;
       validity_info_.Invalidate();
@@ -1669,7 +1670,7 @@ bool CompactionIterator::FetchBlobsNeedingGC(
     Status s = blob_fetcher_->FetchBlob(user_key(), blob_index, prefetch_buffer,
                                         &blob_value, &bytes_read);
     if (!s.ok()) {
-      status_ = s;
+      status_ = std::move(s);
       validity_info_.Invalidate();
       return false;
     }
@@ -1817,8 +1818,8 @@ void CompactionIterator::GarbageCollectEntityBlobsIfNeeded() {
     Slice entity_slice = value_;
     entity_columns_.clear();
     entity_blob_columns_.clear();
-    Status s = WideColumnSerialization::DeserializeV2(
-        entity_slice, entity_columns_, entity_blob_columns_);
+    Status s = WideColumnSerialization::Deserialize(
+        entity_slice, entity_columns_, &entity_blob_columns_);
     if (!s.ok()) {
       status_ = s;
       validity_info_.Invalidate();
@@ -1869,13 +1870,13 @@ void CompactionIterator::PrepareOutput() {
               "CompactionIterator::PrepareOutput:DeserializeEntity");
           // Deserialize entity once into member variables, then decide between
           // blob GC and extraction based on whether blob columns exist.
-          // This avoids the double parse of HasBlobColumns() + DeserializeV2().
+          // This avoids the double parse of HasBlobColumns() + Deserialize().
           entity_columns_.clear();
           entity_blob_columns_.clear();
           Slice entity_slice = value_;
           {
-            Status s_deser = WideColumnSerialization::DeserializeV2(
-                entity_slice, entity_columns_, entity_blob_columns_);
+            Status s_deser = WideColumnSerialization::Deserialize(
+                entity_slice, entity_columns_, &entity_blob_columns_);
             if (!s_deser.ok()) {
               status_ = s_deser;
               validity_info_.Invalidate();
@@ -2064,8 +2065,9 @@ std::unique_ptr<BlobFetcher> CompactionIterator::CreateBlobFetcherIfNeeded(
     }
   }
 
-  return std::unique_ptr<BlobFetcher>(new BlobFetcher(
-      version, read_options, blob_file_cache, allow_write_path_fallback));
+  return std::unique_ptr<BlobFetcher>(
+      new OwningVersionBlobFetcher(version, std::move(read_options),
+                                   blob_file_cache, allow_write_path_fallback));
 }
 
 std::unique_ptr<PrefetchBufferCollection>

db/compaction/compaction_iterator_test.cc

@@ -1896,8 +1896,8 @@ TEST_F(WideColumnEntityBlobExtractionTest,
   Slice value_slice(output_values[0]);
   std::vector<WideColumn> output_columns;
   std::vector<std::pair<size_t, BlobIndex>> blob_columns;
-  ASSERT_OK(WideColumnSerialization::DeserializeV2(value_slice, output_columns,
-                                                   blob_columns));
+  ASSERT_OK(WideColumnSerialization::Deserialize(value_slice, output_columns,
+                                                 &blob_columns));
 
   // The large column should now be a blob reference
   ASSERT_EQ(1, blob_columns.size());
@@ -1941,8 +1941,8 @@ TEST_F(WideColumnEntityBlobExtractionTest,
   Slice value_slice(output_values[0]);
   std::vector<WideColumn> output_columns;
   std::vector<std::pair<size_t, BlobIndex>> blob_columns;
-  ASSERT_OK(WideColumnSerialization::DeserializeV2(value_slice, output_columns,
-                                                   blob_columns));
+  ASSERT_OK(WideColumnSerialization::Deserialize(value_slice, output_columns,
+                                                 &blob_columns));
 
   // All three columns should be blob references
   ASSERT_EQ(3, blob_columns.size());
@@ -2014,8 +2014,8 @@ TEST_F(WideColumnEntityBlobExtractionTest, WideColumnEntityMixedSizes) {
   Slice value_slice(output_values[0]);
   std::vector<WideColumn> output_columns;
   std::vector<std::pair<size_t, BlobIndex>> blob_columns;
-  ASSERT_OK(WideColumnSerialization::DeserializeV2(value_slice, output_columns,
-                                                   blob_columns));
+  ASSERT_OK(WideColumnSerialization::Deserialize(value_slice, output_columns,
+                                                 &blob_columns));
 
   // Only the large column should be a blob reference
   ASSERT_EQ(1, blob_columns.size());
@@ -2058,8 +2058,8 @@ TEST_F(WideColumnEntityBlobExtractionTest,
   Slice value_slice(output_values[0]);
   std::vector<WideColumn> output_columns;
   std::vector<std::pair<size_t, BlobIndex>> blob_columns;
-  ASSERT_OK(WideColumnSerialization::DeserializeV2(value_slice, output_columns,
-                                                   blob_columns));
+  ASSERT_OK(WideColumnSerialization::Deserialize(value_slice, output_columns,
+                                                 &blob_columns));
 
   // The default column should be a blob reference
   ASSERT_EQ(1, blob_columns.size());
@@ -2078,11 +2078,14 @@ TEST_F(WideColumnEntityBlobExtractionTest,
 // Mock BlobFetcher for testing blob GC without a real Version
 class MockBlobFetcher : public BlobFetcher {
  public:
-  MockBlobFetcher() : BlobFetcher(nullptr /* version */, ReadOptions()) {}
+  MockBlobFetcher() = default;
+
+  using BlobFetcher::FetchBlob;
 
   Status FetchBlob(const Slice& /* user_key */, const BlobIndex& blob_index,
                    FilePrefetchBuffer* /* prefetch_buffer */,
-                   PinnableSlice* blob_value, uint64_t* bytes_read) const {
+                   PinnableSlice* blob_value,
+                   uint64_t* bytes_read) const override {
     auto it = blob_data_.find(
         std::make_pair(blob_index.file_number(), blob_index.offset()));
     if (it == blob_data_.end()) {
@@ -2096,6 +2099,8 @@ class MockBlobFetcher : public BlobFetcher {
     return Status::OK();
   }
 
+  const ReadOptions& read_options() const override { return read_options_; }
+
   void AddBlob(uint64_t file_number, uint64_t offset,
                const std::string& value) {
     blob_data_[std::make_pair(file_number, offset)] = value;
@@ -2103,6 +2108,7 @@ class MockBlobFetcher : public BlobFetcher {
 
  private:
   std::map<std::pair<uint64_t, uint64_t>, std::string> blob_data_;
+  ReadOptions read_options_;
 };
 
 // FakeCompaction that supports blob garbage collection
@@ -2372,8 +2378,8 @@ TEST_F(WideColumnEntityBlobGCTest, WideColumnEntityBlobGCRelocate) {
   Slice value_slice(output_values[0]);
   std::vector<WideColumn> output_columns;
   std::vector<std::pair<size_t, BlobIndex>> blob_columns_out;
-  ASSERT_OK(WideColumnSerialization::DeserializeV2(value_slice, output_columns,
-                                                   blob_columns_out));
+  ASSERT_OK(WideColumnSerialization::Deserialize(value_slice, output_columns,
+                                                 &blob_columns_out));
 
   // Entity structure should be preserved
   ASSERT_EQ(2, output_columns.size());
@@ -2431,8 +2437,8 @@ TEST_F(WideColumnEntityBlobGCTest, WideColumnEntityBlobGCNoRelocateNewerFile) {
   Slice value_slice(output_values[0]);
   std::vector<WideColumn> output_columns;
   std::vector<std::pair<size_t, BlobIndex>> blob_columns_out;
-  ASSERT_OK(WideColumnSerialization::DeserializeV2(value_slice, output_columns,
-                                                   blob_columns_out));
+  ASSERT_OK(WideColumnSerialization::Deserialize(value_slice, output_columns,
+                                                 &blob_columns_out));
 
   // Blob column should still exist and reference the same file
   ASSERT_EQ(1, blob_columns_out.size());
@@ -2491,8 +2497,8 @@ TEST_F(WideColumnEntityBlobGCTest, WideColumnEntityBlobGCPartialRelocation) {
   Slice value_slice(output_values[0]);
   std::vector<WideColumn> output_columns;
   std::vector<std::pair<size_t, BlobIndex>> blob_columns_out;
-  ASSERT_OK(WideColumnSerialization::DeserializeV2(value_slice, output_columns,
-                                                   blob_columns_out));
+  ASSERT_OK(WideColumnSerialization::Deserialize(value_slice, output_columns,
+                                                 &blob_columns_out));
 
   // All three columns should exist
   ASSERT_EQ(3, output_columns.size());
@@ -2563,8 +2569,8 @@ TEST_F(WideColumnEntityBlobGCTest, WideColumnEntityBlobGCToInline) {
   Slice value_slice(output_values[0]);
   std::vector<WideColumn> output_columns;
   std::vector<std::pair<size_t, BlobIndex>> blob_columns_out;
-  ASSERT_OK(WideColumnSerialization::DeserializeV2(value_slice, output_columns,
-                                                   blob_columns_out));
+  ASSERT_OK(WideColumnSerialization::Deserialize(value_slice, output_columns,
+                                                 &blob_columns_out));
 
   // Both columns should exist
   ASSERT_EQ(2, output_columns.size());
@@ -2788,8 +2794,8 @@ TEST_F(WideColumnEntityBlobGCTest,
   Slice value_slice(output_values[0]);
   std::vector<WideColumn> output_columns;
   std::vector<std::pair<size_t, BlobIndex>> output_blob_columns;
-  ASSERT_OK(WideColumnSerialization::DeserializeV2(value_slice, output_columns,
-                                                   output_blob_columns));
+  ASSERT_OK(WideColumnSerialization::Deserialize(value_slice, output_columns,
+                                                 &output_blob_columns));
 
   ASSERT_EQ(2, output_columns.size());
   ASSERT_EQ("alpha", output_columns[0].name().ToString());

db/compaction/compaction_job.cc

@@ -173,6 +173,7 @@ CompactionJob::CompactionJob(
       fs_(db_options.fs, io_tracer),
       file_options_for_read_(
           fs_->OptimizeForCompactionTableRead(file_options, db_options_)),
+      file_options_for_compaction_input_read_(file_options_for_read_),
       versions_(versions),
       shutting_down_(shutting_down),
       manual_compaction_canceled_(manual_compaction_canceled),
@@ -203,6 +204,11 @@ CompactionJob::CompactionJob(
   assert(job_context);
   assert(job_context->snapshot_context_initialized);
 
+  if (db_options_.use_direct_io_for_compaction_reads &&
+      !db_options_.use_direct_reads) {
+    file_options_for_compaction_input_read_.use_direct_reads = true;
+  }
+
   const auto* cfd = compact_->compaction->column_family_data();
   ThreadStatusUtil::SetEnableTracking(db_options_.enable_thread_tracking);
   ThreadStatusUtil::SetColumnFamily(cfd);
@@ -1536,10 +1542,20 @@ InternalIterator* CompactionJob::CreateInputIterator(
 
   // Although the v2 aggregator is what the level iterator(s) know about,
   // the AddTombstones calls will be propagated down to the v1 aggregator.
+  const bool open_ephemeral_table_reader =
+      db_options_.use_direct_io_for_compaction_reads &&
+      !db_options_.use_direct_reads;
+  FileOptions& input_file_options =
+      open_ephemeral_table_reader ? file_options_for_compaction_input_read_
+                                  : file_options_for_read_;
+  TEST_SYNC_POINT_CALLBACK(
+      "CompactionJob::CreateInputIterator:InputFileOptions",
+      &input_file_options);
   iterators.raw_input =
       std::unique_ptr<InternalIterator>(versions_->MakeInputIterator(
           read_options, sub_compact->compaction, sub_compact->RangeDelAgg(),
-          file_options_for_read_, boundaries.start, boundaries.end));
+          input_file_options, boundaries.start, boundaries.end,
+          open_ephemeral_table_reader));
   InternalIterator* input = iterators.raw_input.get();
 
   if (boundaries.start.has_value() || boundaries.end.has_value()) {
@@ -2470,6 +2486,8 @@ Status CompactionJob::OpenCompactionOutputFile(SubcompactionState* sub_compact,
   auto temperature =
       sub_compact->compaction->GetOutputTemperature(outputs.IsProximalLevel());
   fo_copy.temperature = temperature;
+  fo_copy.open_contract = FileOpenContract::kNoReopenForWrite |
+                          FileOpenContract::kNoReadersWhileOpenForWrite;
   fo_copy.write_hint = write_hint_;
 
   Status s;

db/compaction/compaction_job.h

@@ -462,6 +462,8 @@ class CompactionJob {
   FileSystemPtr fs_;
   // env_option optimized for compaction table reads
   FileOptions file_options_for_read_;
+  // file_options_for_read_ plus compaction-input-only overrides.
+  FileOptions file_options_for_compaction_input_read_;
   VersionSet* versions_;
   const std::atomic<bool>* shutting_down_;
   const std::atomic<bool>& manual_compaction_canceled_;

db/compaction/compaction_service_job.cc

@@ -431,6 +431,14 @@ Status CompactionServiceCompactionJob::Run() {
   // Please note that input stats will be updated by primary host when all
   // subcompactions are finished
   UpdateCompactionJobOutputStatsFromInternalStats(status, internal_stats_);
+  // Whole-file filtering changes which keys are fed to the iterator, so the
+  // iterator-based input record count should not be treated as exact.
+  for (size_t level = 1; level < c->num_input_levels(); ++level) {
+    if (!c->filtered_input_levels(level).empty()) {
+      compaction_result_->stats.has_accurate_num_input_records = false;
+      break;
+    }
+  }
   // and set fields that are not propagated as part of the update
   compaction_result_->stats.is_manual_compaction = c->is_manual_compaction();
   compaction_result_->stats.is_full_compaction = c->is_full_compaction();
@@ -769,6 +777,40 @@ static std::unordered_map<std::string, OptionTypeInfo>
           OptionTypeFlags::kNone}},
 };
 
+namespace {
+
+// Like OptionTypeInfo::Array<T, kSize> but deserialization tolerates
+// size mismatches: extra elements beyond kSize are silently ignored,
+// missing elements are zero-filled. Serialization and comparison are
+// inherited from Array<T, kSize> unchanged; only the parse
+// (deserialize) callback is overridden.
+template <typename T, size_t kSize>
+OptionTypeInfo SizeTolerantDeserializeArray(int offset,
+                                            OptionVerificationType verification,
+                                            OptionTypeFlags flags,
+                                            const OptionTypeInfo& elem_info,
+                                            char separator = ':') {
+  return OptionTypeInfo::Array<T, kSize>(offset, verification, flags, elem_info,
+                                         separator)
+      .SetParseFunc([elem_info, separator](
+                        const ConfigOptions& opts, const std::string& name,
+                        const std::string& value, void* addr) {
+        std::vector<T> parsed;
+        Status s =
+            ParseVector<T>(opts, elem_info, separator, name, value, &parsed);
+        if (!s.ok()) {
+          return Status(std::move(s));
+        }
+        auto* arr = static_cast<std::array<T, kSize>*>(addr);
+        for (size_t i = 0; i < kSize; ++i) {
+          (*arr)[i] = (i < parsed.size()) ? parsed[i] : T{};
+        }
+        return Status::OK();
+      });
+}
+
+}  // namespace
+
 static std::unordered_map<std::string, OptionTypeInfo>
     compaction_stats_type_info = {
         {"micros",
@@ -860,12 +902,20 @@ static std::unordered_map<std::string, OptionTypeInfo>
          {offsetof(struct InternalStats::CompactionStats, count),
           OptionType::kUInt64T, OptionVerificationType::kNormal,
           OptionTypeFlags::kNone}},
+        // The remote worker may serialize a different number of
+        // CompactionReason counts than this primary expects (RocksDB versions
+        // can disagree on kNumOfReasons). Deserialization tolerates the
+        // mismatch: extra trailing elements are dropped, missing ones are
+        // zero-filled.
+        //
+        // The serialized width is intentionally kept at kNumOfReasons - 1 (the
+        // reduced width) so this change does NOT alter what is written -- only
+        // the read path becomes tolerant. Widening the serialized width to the
+        // full kNumOfReasons is a separate, later change that must not land
+        // until this tolerant read has been deployed to every primary,
+        // including rollback targets.
         {"counts",
-         OptionTypeInfo::Array<
-             /* In release 11.2, a new compaction reason was added. This broken
-              * the reader and writer. To unblock release 11.1, we temporarily
-              * reduce the count array size to the old one. TODO add a proper
-              * serialization and deserialization method. */
+         SizeTolerantDeserializeArray<
              int, static_cast<int>(CompactionReason::kNumOfReasons) - 1>(
              offsetof(struct InternalStats::CompactionStats, counts),
              OptionVerificationType::kNormal, OptionTypeFlags::kNone,

db/compaction/compaction_service_test.cc

@@ -3,6 +3,8 @@
 //  COPYING file in the root directory) and Apache 2.0 License
 //  (found in the LICENSE.Apache file in the root directory).
 
+#include <memory>
+
 #include "db/db_test_util.h"
 #include "file/file_util.h"
 #include "port/stack_trace.h"
@@ -12,6 +14,71 @@
 
 namespace ROCKSDB_NAMESPACE {
 
+namespace {
+
+// Append ":extra_value" to every "counts=...;" field in a serialized
+// CompactionServiceResult string to simulate a worker with more
+// CompactionReason enums than the reader. Loops because the serialized
+// result contains multiple CompactionStats structs (output_level_stats,
+// proximal_level_stats), each with its own "counts=" field.
+void AppendToCountsFields(std::string* s, const std::string& extra_value) {
+  const std::string prefix = "counts=";
+  size_t pos = s->find(prefix);
+  while (pos != std::string::npos) {
+    // Find the semicolon that terminates this field's value.
+    size_t semi = s->find(';', pos + prefix.size());
+    assert(semi != std::string::npos);
+    // Insert ":extra_value" just before the semicolon.
+    s->insert(semi, ":" + extra_value);
+    // Advance past the inserted text to find the next occurrence.
+    pos = s->find(prefix, semi + extra_value.size() + 2);
+  }
+}
+
+// Remove the last colon-separated element from every "counts=...;" field
+// to simulate a worker with fewer CompactionReason enums than the reader.
+// Loops for the same reason as AppendToCountsFields (multiple counts fields).
+void RemoveLastFromCountsFields(std::string* s) {
+  const std::string prefix = "counts=";
+  size_t pos = s->find(prefix);
+  while (pos != std::string::npos) {
+    size_t val_begin = pos + prefix.size();
+    size_t semi = s->find(';', val_begin);
+    assert(semi != std::string::npos);
+    // Find the last ':' separator within the value.
+    size_t last_sep = s->rfind(':', semi - 1);
+    assert(last_sep != std::string::npos && last_sep >= val_begin);
+    // Erase from the last ':' up to (not including) the ';'.
+    // e.g. "counts=0:3:5;" -> erase ":5" -> "counts=0:3;"
+    s->erase(last_sep, semi - last_sep);
+    pos = s->find(prefix, last_sep);
+  }
+}
+
+// Count the colon-separated elements in the first "counts=...;" field of a
+// serialized CompactionServiceResult. Used to assert the serialized width
+// (how many CompactionReason counts this binary writes).
+size_t CountFirstCountsFieldElements(const std::string& s) {
+  const std::string prefix = "counts=";
+  size_t pos = s.find(prefix);
+  assert(pos != std::string::npos);
+  size_t val_begin = pos + prefix.size();
+  size_t semi = s.find(';', val_begin);
+  assert(semi != std::string::npos);
+  if (semi == val_begin) {
+    return 0;
+  }
+  size_t count = 1;
+  for (size_t i = val_begin; i < semi; ++i) {
+    if (s[i] == ':') {
+      ++count;
+    }
+  }
+  return count;
+}
+
+}  // namespace
+
 class MyTestCompactionService : public CompactionService {
  public:
   MyTestCompactionService(
@@ -611,6 +678,101 @@ TEST_F(CompactionServiceTest, StandaloneDeleteRangeTombstoneOptimization) {
   SyncPoint::GetInstance()->DisableProcessing();
 }
 
+TEST_F(CompactionServiceTest,
+       StandaloneDeleteRangeTombstoneMakesInputCountInaccurate) {
+  Options options = CurrentOptions();
+  options.compaction_style = CompactionStyle::kCompactionStyleUniversal;
+  options.compaction_verify_record_count = true;
+  ReopenWithCompactionService(&options);
+
+  std::vector<std::string> files;
+  {
+    SstFileWriter sst_file_writer(EnvOptions(), options);
+    std::string file1 = dbname_ + "file1.sst";
+    ASSERT_OK(sst_file_writer.Open(file1));
+    ASSERT_OK(sst_file_writer.Put("a", "a1"));
+    ASSERT_OK(sst_file_writer.Put("b", "b1"));
+    ExternalSstFileInfo file1_info;
+    ASSERT_OK(sst_file_writer.Finish(&file1_info));
+    files.push_back(std::move(file1));
+
+    std::string file2 = dbname_ + "file2.sst";
+    ASSERT_OK(sst_file_writer.Open(file2));
+    ASSERT_OK(sst_file_writer.Put("x", "x1"));
+    ASSERT_OK(sst_file_writer.Put("y", "y1"));
+    ExternalSstFileInfo file2_info;
+    ASSERT_OK(sst_file_writer.Finish(&file2_info));
+    files.push_back(std::move(file2));
+  }
+
+  IngestExternalFileOptions ifo;
+  ASSERT_OK(db_->IngestExternalFile(files, ifo));
+  ASSERT_EQ(Get("a"), "a1");
+  ASSERT_EQ(Get("b"), "b1");
+  ASSERT_EQ(Get("x"), "x1");
+  ASSERT_EQ(Get("y"), "y1");
+  ASSERT_EQ(2, NumTableFilesAtLevel(6));
+
+  auto my_cs = GetCompactionService();
+  uint64_t comp_num = my_cs->GetCompactionNum();
+
+  size_t num_files_after_filtered = 0;
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "VersionSet::MakeInputIterator:NewCompactionMergingIterator",
+      [&](void* arg) {
+        num_files_after_filtered = *static_cast<size_t*>(arg);
+      });
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  {
+    // The standalone range tombstone can fully cover the old versioned files,
+    // so universal compaction may filter those files before iteration.
+    files.clear();
+    SstFileWriter sst_file_writer(EnvOptions(), options);
+    std::string file2 = dbname_ + "file2.sst";
+    ASSERT_OK(sst_file_writer.Open(file2));
+    ASSERT_OK(sst_file_writer.DeleteRange("a", "z"));
+    ExternalSstFileInfo file2_info;
+    ASSERT_OK(sst_file_writer.Finish(&file2_info));
+    files.push_back(std::move(file2));
+
+    std::string file3 = dbname_ + "file3.sst";
+    ASSERT_OK(sst_file_writer.Open(file3));
+    ASSERT_OK(sst_file_writer.Put("a", "a2"));
+    ASSERT_OK(sst_file_writer.Put("b", "b2"));
+    ExternalSstFileInfo file3_info;
+    ASSERT_OK(sst_file_writer.Finish(&file3_info));
+    files.push_back(std::move(file3));
+
+    std::string file4 = dbname_ + "file4.sst";
+    ASSERT_OK(sst_file_writer.Open(file4));
+    ASSERT_OK(sst_file_writer.Put("x", "x2"));
+    ASSERT_OK(sst_file_writer.Put("y", "y2"));
+    ExternalSstFileInfo file4_info;
+    ASSERT_OK(sst_file_writer.Finish(&file4_info));
+    files.push_back(std::move(file4));
+  }
+
+  ASSERT_OK(db_->IngestExternalFile(files, ifo));
+  ASSERT_OK(db_->WaitForCompact(WaitForCompactOptions()));
+  ASSERT_GE(my_cs->GetCompactionNum(), comp_num + 1);
+
+  CompactionServiceResult result;
+  my_cs->GetResult(&result);
+  ASSERT_OK(result.status);
+  ASSERT_TRUE(result.stats.is_remote_compaction);
+  ASSERT_FALSE(result.stats.has_accurate_num_input_records);
+  ASSERT_EQ(1, num_files_after_filtered);
+
+  ASSERT_EQ(Get("a"), "a2");
+  ASSERT_EQ(Get("b"), "b2");
+  ASSERT_EQ(Get("x"), "x2");
+  ASSERT_EQ(Get("y"), "y2");
+
+  SyncPoint::GetInstance()->DisableProcessing();
+  SyncPoint::GetInstance()->ClearAllCallBacks();
+}
+
 TEST_F(CompactionServiceTest, CompactionOutputFileIOError) {
   Options options = CurrentOptions();
   options.disable_auto_compactions = true;
@@ -1316,6 +1478,10 @@ TEST_F(CompactionServiceTest, TruncatedOutput) {
 
 TEST_F(CompactionServiceTest, CustomFileChecksum) {
   Options options = CurrentOptions();
+  // Pin compression so the auto-compacted LSM shape (and thus whether the
+  // manual CompactRange below schedules a remote compaction) doesn't depend on
+  // the default compression type. kNoCompression is always available.
+  options.compression = kNoCompression;
   options.file_checksum_gen_factory = GetFileChecksumGenCrc32cFactory();
   ReopenWithCompactionService(&options);
   GenerateTestData();
@@ -1510,6 +1676,68 @@ TEST_F(CompactionServiceTest, InvalidResultFallsBackToLocal) {
             my_cs->GetFinalCompactionServiceJobStatus());
 }
 
+TEST_F(CompactionServiceTest, CompatCheckCountsWidthTolerance) {
+  // Number of per-CompactionReason counts this binary serializes and expects.
+  constexpr size_t kExpected =
+      static_cast<size_t>(CompactionReason::kNumOfReasons) - 1;
+  constexpr size_t kLastIndex = kExpected - 1;
+  constexpr int kFirstCount = 3;
+  constexpr int kLastCount = 5;
+
+  CompactionServiceResult result;
+  result.status = Status::OK();
+  result.output_level = 2;
+  result.output_path = "/tmp/output";
+  result.internal_stats.output_level_stats.counts[0] = kFirstCount;
+  result.internal_stats.output_level_stats.counts[kLastIndex] = kLastCount;
+  std::string serialized;
+  ASSERT_OK(result.Write(&serialized));
+
+  // Guard the produced format: this binary writes exactly kExpected counts.
+  // That element count is what a remote worker hands to the primary to read,
+  // so pinning it catches an accidental change to what is written.
+  ASSERT_EQ(CountFirstCountsFieldElements(serialized), kExpected);
+
+  // Peer agrees on kNumOfReasons: every written count is read back unchanged.
+  {
+    CompactionServiceResult parsed;
+    ASSERT_OK(CompactionServiceResult::Read(serialized, &parsed));
+    ASSERT_OK(parsed.status);
+    const auto& counts = parsed.internal_stats.output_level_stats.counts;
+    ASSERT_EQ(counts[0], kFirstCount);
+    ASSERT_EQ(counts[kLastIndex], kLastCount);
+  }
+
+  // Peer serialized one more count than this binary knows (a worker with an
+  // extra CompactionReason): the surplus trailing count is ignored, and the
+  // counts this binary understands are read unchanged.
+  {
+    std::string from_wider_peer = serialized;
+    AppendToCountsFields(&from_wider_peer, "7");
+    ASSERT_EQ(CountFirstCountsFieldElements(from_wider_peer), kExpected + 1);
+    CompactionServiceResult parsed;
+    ASSERT_OK(CompactionServiceResult::Read(from_wider_peer, &parsed));
+    ASSERT_OK(parsed.status);
+    const auto& counts = parsed.internal_stats.output_level_stats.counts;
+    ASSERT_EQ(counts[0], kFirstCount);
+    ASSERT_EQ(counts[kLastIndex], kLastCount);
+  }
+
+  // Peer serialized one fewer count than this binary knows (a worker missing
+  // the newest CompactionReason): the absent trailing count reads as zero.
+  {
+    std::string from_narrower_peer = serialized;
+    RemoveLastFromCountsFields(&from_narrower_peer);
+    ASSERT_EQ(CountFirstCountsFieldElements(from_narrower_peer), kExpected - 1);
+    CompactionServiceResult parsed;
+    ASSERT_OK(CompactionServiceResult::Read(from_narrower_peer, &parsed));
+    ASSERT_OK(parsed.status);
+    const auto& counts = parsed.internal_stats.output_level_stats.counts;
+    ASSERT_EQ(counts[0], kFirstCount);
+    ASSERT_EQ(counts[kLastIndex], 0);
+  }
+}
+
 TEST_F(CompactionServiceTest, SubCompaction) {
   Options options = CurrentOptions();
   options.max_subcompactions = 10;

db/db_basic_test.cc

@@ -10,6 +10,9 @@
 #include <cstring>
 
 #include "db/db_test_util.h"
+#include "db/log_writer.h"
+#include "db/version_edit.h"
+#include "file/writable_file_writer.h"
 #include "options/options_helper.h"
 #include "port/stack_trace.h"
 #include "rocksdb/filter_policy.h"
@@ -78,6 +81,7 @@ class MyFlushBlockPolicyFactory : public FlushBlockPolicyFactory {
  private:
   const int num_keys_in_block_;
 };
+
 }  // namespace
 
 static bool enable_io_uring = true;
@@ -88,6 +92,15 @@ class DBBasicTest : public DBTestBase {
   DBBasicTest() : DBTestBase("db_basic_test", /*env_do_fsync=*/false) {}
 };
 
+class DBBasicGetWithParam : public DBBasicTest,
+                            public testing::WithParamInterface<bool> {
+ public:
+  DBBasicGetWithParam() : DBBasicTest(), use_coroutine_(GetParam()) {}
+
+ protected:
+  const bool use_coroutine_;
+};
+
 TEST_F(DBBasicTest, OpenWhenOpen) {
   Options options = CurrentOptions();
   options.env = env_;
@@ -207,8 +220,9 @@ TEST_F(DBBasicTest, OptimizeManifestForRecoveryEmitsPerCFWhenLogAdvances) {
   ASSERT_EQ("v", Get("k"));
 }
 
-// With track_and_verify_wals_in_manifest=true, the wal_deletion edit
-// must STILL be emitted (the DeleteWalsBefore record is required).
+// With track_and_verify_wals_in_manifest=true and a clean close, the
+// close-time code already advances WalSet markers. Recovery's
+// DeleteWalsBefore is redundant and should be skipped.
 TEST_F(DBBasicTest, OptimizeManifestForRecoveryPreservesWalTracking) {
   Options options = CurrentOptions();
   options.create_if_missing = true;
@@ -224,7 +238,7 @@ TEST_F(DBBasicTest, OptimizeManifestForRecoveryPreservesWalTracking) {
   Reopen(options);
   counters.Uninstall();
 
-  ASSERT_EQ(0, counters.wal_deletion.load());
+  ASSERT_EQ(1, counters.wal_deletion.load());
 }
 
 // best_efforts_recovery requires a fresh MANIFEST + CURRENT to be
@@ -497,11 +511,60 @@ TEST_F(DBBasicTest, OptimizeManifestForRecoveryEmitsWalDeletionWhenTracking) {
   counters.Uninstall();
 
   ASSERT_GT(counters.per_cf.load(), 0);
-  ASSERT_EQ(1, records.load());
+  ASSERT_EQ(0, records.load());
   ASSERT_GT(dbfull()->GetVersionSet()->GetWalSet().GetMinWalNumberToKeep(), 0u);
   ASSERT_EQ("v", Get("k"));
 }
 
+// Zero MANIFEST growth on clean reopen with optimize_manifest_for_recovery=1,
+// reuse_manifest_on_open=1, and track_and_verify_wals_in_manifest=1.
+// Regression test for the stress test failure where a redundant
+// DeleteWalsBefore edit was written during recovery.
+TEST_F(DBBasicTest, OptimizeManifestForRecoveryZeroGrowthOnCleanReopen) {
+  Options options = CurrentOptions();
+  options.create_if_missing = true;
+  options.optimize_manifest_for_recovery = true;
+  options.reuse_manifest_on_open = true;
+  options.track_and_verify_wals_in_manifest = true;
+  options.avoid_flush_during_recovery = true;
+  options.write_dbid_to_manifest = false;
+  DestroyAndReopen(options);
+  ASSERT_OK(Put("k", "v"));
+  ASSERT_OK(Flush());
+  Close();
+
+  // Get MANIFEST size after close
+  std::vector<std::string> files;
+  ASSERT_OK(env_->GetChildren(dbname_, &files));
+  uint64_t manifest_size_before = 0;
+  for (const auto& f : files) {
+    if (f.find("MANIFEST-") == 0) {
+      ASSERT_OK(env_->GetFileSize(dbname_ + "/" + f, &manifest_size_before));
+      break;
+    }
+  }
+  ASSERT_GT(manifest_size_before, 0u);
+
+  // Reopen - should not grow MANIFEST at all
+  Reopen(options);
+
+  // Get MANIFEST size after reopen
+  files.clear();
+  ASSERT_OK(env_->GetChildren(dbname_, &files));
+  uint64_t manifest_size_after = 0;
+  for (const auto& f : files) {
+    if (f.find("MANIFEST-") == 0) {
+      ASSERT_OK(env_->GetFileSize(dbname_ + "/" + f, &manifest_size_after));
+      break;
+    }
+  }
+
+  ASSERT_EQ(manifest_size_before, manifest_size_after)
+      << "MANIFEST grew by " << (manifest_size_after - manifest_size_before)
+      << " bytes on clean reopen; expected zero growth";
+  ASSERT_EQ("v", Get("k"));
+}
+
 // Regression for the close-time marker path: if a new WAL is created while an
 // otherwise-empty user CF stays empty, Close() must reserve the next file
 // number before persisting SetLogNumber(cur_wal + 1) for that CF.
@@ -958,6 +1021,92 @@ TEST_F(DBBasicTest, ReuseManifestOnOpenSkipsOnTailCorruption) {
   ASSERT_EQ("v", Get("k"));
 }
 
+// Regression test for corrupted atomic group when reuse_manifest_on_open
+// appends after an incomplete atomic group at the MANIFEST tail.
+// Before the fix, last_valid_record_end_ included records from an incomplete
+// trailing atomic group, causing ReopenManifestForAppend to append new records
+// after them. On subsequent recovery the incomplete group followed by new
+// records triggers "corrupted atomic group".
+TEST_F(DBBasicTest, ReuseManifestOnOpenIncompleteAtomicGroupAtTail) {
+  Options options = CurrentOptions();
+  options.create_if_missing = true;
+  options.reuse_manifest_on_open = true;
+  DestroyAndReopen(options);
+  ASSERT_OK(Put("k", "v"));
+  ASSERT_OK(Flush());
+  Close();
+
+  // Find the MANIFEST file.
+  std::string manifest_path;
+  {
+    std::vector<std::string> files;
+    ASSERT_OK(env_->GetChildren(dbname_, &files));
+    for (const auto& f : files) {
+      uint64_t number;
+      FileType type;
+      if (ParseFileName(f, &number, &type) && type == kDescriptorFile) {
+        manifest_path = dbname_ + "/" + f;
+        break;
+      }
+    }
+  }
+  ASSERT_FALSE(manifest_path.empty());
+
+  // Append an incomplete atomic group (2 records of a 3-member group) to the
+  // MANIFEST. These are valid log records with correct CRCs but the atomic
+  // group is not complete (missing the third record with remaining_entries=0).
+  {
+    uint64_t file_size;
+    ASSERT_OK(env_->GetFileSize(manifest_path, &file_size));
+    const auto& fs = env_->GetFileSystem();
+    std::unique_ptr<FSWritableFile> fs_file;
+    ASSERT_OK(fs->ReopenWritableFile(manifest_path, FileOptions(), &fs_file,
+                                     nullptr));
+    std::unique_ptr<WritableFileWriter> file_writer(new WritableFileWriter(
+        std::move(fs_file), manifest_path, FileOptions()));
+    log::Writer log_writer(std::move(file_writer), 0, false, false,
+                           kNoCompression, false, file_size % log::kBlockSize);
+
+    // Write 2 records of a 3-member atomic group
+    VersionEdit edit1;
+    edit1.SetLogNumber(0);
+    edit1.SetNextFile(100);
+    edit1.SetLastSequence(100);
+    edit1.MarkAtomicGroup(2);  // remaining_entries=2, expects 3 total
+    std::string record1;
+    ASSERT_TRUE(edit1.EncodeTo(&record1, 0));
+    ASSERT_OK(log_writer.AddRecord(WriteOptions(), record1));
+
+    VersionEdit edit2;
+    edit2.SetLogNumber(0);
+    edit2.SetNextFile(100);
+    edit2.SetLastSequence(100);
+    edit2.MarkAtomicGroup(1);  // remaining_entries=1
+    std::string record2;
+    ASSERT_TRUE(edit2.EncodeTo(&record2, 0));
+    ASSERT_OK(log_writer.AddRecord(WriteOptions(), record2));
+    // Deliberately NOT writing the third record (remaining_entries=0)
+  }
+
+  // Reopen: recovery should succeed (incomplete trailing group is ignored).
+  // With the fix, ReopenManifestForAppend will NOT reuse the MANIFEST because
+  // manifest_last_valid_record_end_ < physical_size (the incomplete group
+  // records are excluded from the valid end).
+  Reopen(options);
+  ASSERT_EQ("v", Get("k"));
+
+  // Write new data to create new MANIFEST records.
+  ASSERT_OK(Put("k2", "v2"));
+  ASSERT_OK(Flush());
+  Close();
+
+  // Final reopen: must NOT fail with "corrupted atomic group".
+  Reopen(options);
+  ASSERT_EQ("v", Get("k"));
+  ASSERT_EQ("v2", Get("k2"));
+  Close();
+}
+
 TEST_F(DBBasicTest, EnableDirectIOWithZeroBuf) {
   if (!IsDirectIOSupported()) {
     ROCKSDB_GTEST_BYPASS("Direct IO not supported");
@@ -1320,27 +1469,36 @@ TEST_F(DBBasicTest, LevelLimitReopen) {
   ASSERT_OK(TryReopenWithColumnFamilies({"default", "pikachu"}, options));
 }
 
-TEST_F(DBBasicTest, PutDeleteGet) {
+TEST_P(DBBasicGetWithParam, PutDeleteGet) {
   do {
     CreateAndReopenWithCF({"pikachu"}, CurrentOptions());
     ASSERT_OK(Put(1, "foo", "v1"));
-    ASSERT_EQ("v1", Get(1, "foo"));
+    ASSERT_EQ("v1", Get(1, "foo", nullptr, use_coroutine_));
     ASSERT_OK(Put(1, "foo", "v2"));
-    ASSERT_EQ("v2", Get(1, "foo"));
+    ASSERT_EQ("v2", Get(1, "foo", nullptr, use_coroutine_));
     ASSERT_OK(Delete(1, "foo"));
-    ASSERT_EQ("NOT_FOUND", Get(1, "foo"));
+    ASSERT_EQ("NOT_FOUND", Get(1, "foo", nullptr, use_coroutine_));
+
+    ASSERT_OK(Put("default_foo", "default_v1"));
+    ASSERT_EQ("default_v1",
+              Get("default_foo", static_cast<const Snapshot*>(nullptr),
+                  use_coroutine_));
+
+    PinnableSlice default_pinnable_value;
+    ASSERT_OK(Get("default_foo", &default_pinnable_value, use_coroutine_));
+    ASSERT_EQ("default_v1", default_pinnable_value.ToString());
   } while (ChangeOptions());
 }
 
-TEST_F(DBBasicTest, PutSingleDeleteGet) {
+TEST_P(DBBasicGetWithParam, PutSingleDeleteGet) {
   do {
     CreateAndReopenWithCF({"pikachu"}, CurrentOptions());
     ASSERT_OK(Put(1, "foo", "v1"));
-    ASSERT_EQ("v1", Get(1, "foo"));
+    ASSERT_EQ("v1", Get(1, "foo", nullptr, use_coroutine_));
     ASSERT_OK(Put(1, "foo2", "v2"));
-    ASSERT_EQ("v2", Get(1, "foo2"));
+    ASSERT_EQ("v2", Get(1, "foo2", nullptr, use_coroutine_));
     ASSERT_OK(SingleDelete(1, "foo"));
-    ASSERT_EQ("NOT_FOUND", Get(1, "foo"));
+    ASSERT_EQ("NOT_FOUND", Get(1, "foo", nullptr, use_coroutine_));
     // Ski FIFO and universal compaction because they do not apply to the test
     // case. Skip MergePut because single delete does not get removed when it
     // encounters a merge.
@@ -1348,40 +1506,40 @@ TEST_F(DBBasicTest, PutSingleDeleteGet) {
                          kSkipMergePut));
 }
 
-TEST_F(DBBasicTest, TimedPutBasic) {
+TEST_P(DBBasicGetWithParam, TimedPutBasic) {
   do {
     Options options = CurrentOptions();
     options.merge_operator = MergeOperators::CreateStringAppendOperator();
     CreateAndReopenWithCF({"pikachu"}, options);
     ASSERT_OK(TimedPut(1, "foo", "v1", /*write_unix_time=*/0));
     // Read from memtable
-    ASSERT_EQ("v1", Get(1, "foo"));
+    ASSERT_EQ("v1", Get(1, "foo", nullptr, use_coroutine_));
     ASSERT_OK(TimedPut(1, "foo", "v2.1", /*write_unix_time=*/3));
-    ASSERT_EQ("v2.1", Get(1, "foo"));
+    ASSERT_EQ("v2.1", Get(1, "foo", nullptr, use_coroutine_));
 
     // Read from sst file
     ASSERT_OK(db_->Flush(FlushOptions(), handles_[1]));
     ASSERT_OK(Merge(1, "foo", "v2.2"));
-    ASSERT_EQ("v2.1,v2.2", Get(1, "foo"));
+    ASSERT_EQ("v2.1,v2.2", Get(1, "foo", nullptr, use_coroutine_));
     ASSERT_OK(Delete(1, "foo"));
-    ASSERT_EQ("NOT_FOUND", Get(1, "foo"));
+    ASSERT_EQ("NOT_FOUND", Get(1, "foo", nullptr, use_coroutine_));
 
     ASSERT_OK(TimedPut(1, "bar", "bv1", /*write_unix_time=*/0));
-    ASSERT_EQ("bv1", Get(1, "bar"));
+    ASSERT_EQ("bv1", Get(1, "bar", nullptr, use_coroutine_));
     ASSERT_OK(TimedPut(1, "baz", "bzv1", /*write_unix_time=*/0));
-    ASSERT_EQ("bzv1", Get(1, "baz"));
+    ASSERT_EQ("bzv1", Get(1, "baz", nullptr, use_coroutine_));
     if (option_config_ != kRowCache) {
       std::string range_del_begin = "b";
       std::string range_del_end = "baz";
       Slice begin_rdel = range_del_begin, end_rdel = range_del_end;
       ASSERT_OK(
           db_->DeleteRange(WriteOptions(), handles_[1], begin_rdel, end_rdel));
-      ASSERT_EQ("NOT_FOUND", Get(1, "bar"));
+      ASSERT_EQ("NOT_FOUND", Get(1, "bar", nullptr, use_coroutine_));
     }
 
-    ASSERT_EQ("bzv1", Get(1, "baz"));
+    ASSERT_EQ("bzv1", Get(1, "baz", nullptr, use_coroutine_));
     ASSERT_OK(SingleDelete(1, "baz"));
-    ASSERT_EQ("NOT_FOUND", Get(1, "baz"));
+    ASSERT_EQ("NOT_FOUND", Get(1, "baz", nullptr, use_coroutine_));
   } while (ChangeOptions(kSkipPlainTable));
 }
 
@@ -1407,17 +1565,17 @@ TEST_F(DBBasicTest, EmptyFlush) {
                          kSkipMergePut));
 }
 
-TEST_F(DBBasicTest, GetFromVersions) {
+TEST_P(DBBasicGetWithParam, GetFromVersions) {
   do {
     CreateAndReopenWithCF({"pikachu"}, CurrentOptions());
     ASSERT_OK(Put(1, "foo", "v1"));
     ASSERT_OK(Flush(1));
-    ASSERT_EQ("v1", Get(1, "foo"));
-    ASSERT_EQ("NOT_FOUND", Get(0, "foo"));
+    ASSERT_EQ("v1", Get(1, "foo", nullptr, use_coroutine_));
+    ASSERT_EQ("NOT_FOUND", Get(0, "foo", nullptr, use_coroutine_));
   } while (ChangeOptions());
 }
 
-TEST_F(DBBasicTest, GetSnapshot) {
+TEST_P(DBBasicGetWithParam, GetSnapshot) {
   anon::OptionsOverride options_override;
   options_override.skip_policy = kSkipNoSnapshot;
   do {
@@ -1428,11 +1586,11 @@ TEST_F(DBBasicTest, GetSnapshot) {
       ASSERT_OK(Put(1, key, "v1"));
       const Snapshot* s1 = db_->GetSnapshot();
       ASSERT_OK(Put(1, key, "v2"));
-      ASSERT_EQ("v2", Get(1, key));
-      ASSERT_EQ("v1", Get(1, key, s1));
+      ASSERT_EQ("v2", Get(1, key, nullptr, use_coroutine_));
+      ASSERT_EQ("v1", Get(1, key, s1, use_coroutine_));
       ASSERT_OK(Flush(1));
-      ASSERT_EQ("v2", Get(1, key));
-      ASSERT_EQ("v1", Get(1, key, s1));
+      ASSERT_EQ("v2", Get(1, key, nullptr, use_coroutine_));
+      ASSERT_EQ("v1", Get(1, key, s1, use_coroutine_));
       db_->ReleaseSnapshot(s1);
     }
   } while (ChangeOptions());
@@ -1689,7 +1847,7 @@ TEST_F(DBBasicTest, LockFileRecovery) {
   }
 }
 
-TEST_F(DBBasicTest, Snapshot) {
+TEST_P(DBBasicGetWithParam, Snapshot) {
   env_->SetMockSleep();
   anon::OptionsOverride options_override;
   options_override.skip_policy = kSkipNoSnapshot;
@@ -1723,31 +1881,31 @@ TEST_F(DBBasicTest, Snapshot) {
 
       ASSERT_OK(Put(0, "foo", "0v4"));
       ASSERT_OK(Put(1, "foo", "1v4"));
-      ASSERT_EQ("0v1", Get(0, "foo", s1));
-      ASSERT_EQ("1v1", Get(1, "foo", s1));
-      ASSERT_EQ("0v2", Get(0, "foo", s2));
-      ASSERT_EQ("1v2", Get(1, "foo", s2));
-      ASSERT_EQ("0v3", Get(0, "foo", s3.snapshot()));
-      ASSERT_EQ("1v3", Get(1, "foo", s3.snapshot()));
-      ASSERT_EQ("0v4", Get(0, "foo"));
-      ASSERT_EQ("1v4", Get(1, "foo"));
+      ASSERT_EQ("0v1", Get(0, "foo", s1, use_coroutine_));
+      ASSERT_EQ("1v1", Get(1, "foo", s1, use_coroutine_));
+      ASSERT_EQ("0v2", Get(0, "foo", s2, use_coroutine_));
+      ASSERT_EQ("1v2", Get(1, "foo", s2, use_coroutine_));
+      ASSERT_EQ("0v3", Get(0, "foo", s3.snapshot(), use_coroutine_));
+      ASSERT_EQ("1v3", Get(1, "foo", s3.snapshot(), use_coroutine_));
+      ASSERT_EQ("0v4", Get(0, "foo", nullptr, use_coroutine_));
+      ASSERT_EQ("1v4", Get(1, "foo", nullptr, use_coroutine_));
     }
 
     ASSERT_EQ(2U, GetNumSnapshots());
     ASSERT_EQ(time_snap1, GetTimeOldestSnapshots());
     ASSERT_EQ(GetSequenceOldestSnapshots(), s1->GetSequenceNumber());
-    ASSERT_EQ("0v1", Get(0, "foo", s1));
-    ASSERT_EQ("1v1", Get(1, "foo", s1));
-    ASSERT_EQ("0v2", Get(0, "foo", s2));
-    ASSERT_EQ("1v2", Get(1, "foo", s2));
-    ASSERT_EQ("0v4", Get(0, "foo"));
-    ASSERT_EQ("1v4", Get(1, "foo"));
+    ASSERT_EQ("0v1", Get(0, "foo", s1, use_coroutine_));
+    ASSERT_EQ("1v1", Get(1, "foo", s1, use_coroutine_));
+    ASSERT_EQ("0v2", Get(0, "foo", s2, use_coroutine_));
+    ASSERT_EQ("1v2", Get(1, "foo", s2, use_coroutine_));
+    ASSERT_EQ("0v4", Get(0, "foo", nullptr, use_coroutine_));
+    ASSERT_EQ("1v4", Get(1, "foo", nullptr, use_coroutine_));
 
     db_->ReleaseSnapshot(s1);
-    ASSERT_EQ("0v2", Get(0, "foo", s2));
-    ASSERT_EQ("1v2", Get(1, "foo", s2));
-    ASSERT_EQ("0v4", Get(0, "foo"));
-    ASSERT_EQ("1v4", Get(1, "foo"));
+    ASSERT_EQ("0v2", Get(0, "foo", s2, use_coroutine_));
+    ASSERT_EQ("1v2", Get(1, "foo", s2, use_coroutine_));
+    ASSERT_EQ("0v4", Get(0, "foo", nullptr, use_coroutine_));
+    ASSERT_EQ("1v4", Get(1, "foo", nullptr, use_coroutine_));
     ASSERT_EQ(1U, GetNumSnapshots());
     ASSERT_LT(time_snap1, GetTimeOldestSnapshots());
     ASSERT_EQ(GetSequenceOldestSnapshots(), s2->GetSequenceNumber());
@@ -1755,11 +1913,19 @@ TEST_F(DBBasicTest, Snapshot) {
     db_->ReleaseSnapshot(s2);
     ASSERT_EQ(0U, GetNumSnapshots());
     ASSERT_EQ(GetSequenceOldestSnapshots(), 0);
-    ASSERT_EQ("0v4", Get(0, "foo"));
-    ASSERT_EQ("1v4", Get(1, "foo"));
+    ASSERT_EQ("0v4", Get(0, "foo", nullptr, use_coroutine_));
+    ASSERT_EQ("1v4", Get(1, "foo", nullptr, use_coroutine_));
   } while (ChangeOptions());
 }
 
+#if USE_COROUTINES
+INSTANTIATE_TEST_CASE_P(DBBasicGetWithParam, DBBasicGetWithParam,
+                        testing::Bool());
+#else
+INSTANTIATE_TEST_CASE_P(DBBasicGetWithParam, DBBasicGetWithParam,
+                        testing::Values(false));
+#endif  // USE_COROUTINES
+
 class DBBasicMultiConfigs : public DBBasicTest,
                             public ::testing::WithParamInterface<int> {
  public:
@@ -3708,11 +3874,15 @@ INSTANTIATE_TEST_CASE_P(DBMultiGetTestWithParam, DBMultiGetTestWithParam,
                         testing::Combine(testing::Bool(), testing::Bool()));
 
 #if USE_COROUTINES
-class DBMultiGetAsyncIOTest : public DBBasicTest,
-                              public ::testing::WithParamInterface<bool> {
+class DBMultiGetAsyncIOTest
+    : public DBBasicTest,
+      public ::testing::WithParamInterface<std::tuple<bool, bool>> {
  public:
   DBMultiGetAsyncIOTest()
-      : DBBasicTest(), statistics_(ROCKSDB_NAMESPACE::CreateDBStatistics()) {
+      : DBBasicTest(),
+        optimize_multiget_for_io_(std::get<0>(GetParam())),
+        use_coroutine_(std::get<1>(GetParam())),
+        statistics_(ROCKSDB_NAMESPACE::CreateDBStatistics()) {
     BlockBasedTableOptions bbto;
     bbto.filter_policy.reset(NewBloomFilterPolicy(10));
     options_ = CurrentOptions();
@@ -3786,6 +3956,20 @@ class DBMultiGetAsyncIOTest : public DBBasicTest,
   const std::shared_ptr<Statistics>& statistics() { return statistics_; }
 
  protected:
+  void AssertMultiGetIOBatchSize(uint64_t expected_count,
+                                 uint64_t expected_max) {
+    HistogramData multiget_io_batch_size;
+    statistics()->histogramData(MULTIGET_IO_BATCH_SIZE,
+                                &multiget_io_batch_size);
+    ASSERT_EQ(multiget_io_batch_size.count, expected_count);
+    ASSERT_EQ(multiget_io_batch_size.max, expected_max);
+  }
+
+  bool UseCoroutineRead() const {
+    return use_coroutine_ &&
+           options_.env->GetFileSystem()->GetReadExecutor() != nullptr;
+  }
+
   void PrepareDBForTest() {
 #ifdef ROCKSDB_IOURING_PRESENT
     Reopen(options_);
@@ -3801,6 +3985,9 @@ class DBMultiGetAsyncIOTest : public DBBasicTest,
 
   void ReopenDB() { Reopen(options_); }
 
+  const bool optimize_multiget_for_io_;
+  const bool use_coroutine_;
+
  private:
   std::shared_ptr<Statistics> statistics_;
   Options options_;
@@ -3809,39 +3996,28 @@ class DBMultiGetAsyncIOTest : public DBBasicTest,
 TEST_P(DBMultiGetAsyncIOTest, GetFromL0) {
   // All 3 keys in L0. The L0 files should be read serially.
   std::vector<std::string> key_strs{Key(0), Key(40), Key(80)};
-  std::vector<Slice> keys{key_strs[0], key_strs[1], key_strs[2]};
-  std::vector<PinnableSlice> values(key_strs.size());
-  std::vector<Status> statuses(key_strs.size());
 
   PrepareDBForTest();
 
-  ReadOptions ro;
-  ro.async_io = true;
-  ro.optimize_multiget_for_io = GetParam();
-  dbfull()->MultiGet(ro, dbfull()->DefaultColumnFamily(), keys.size(),
-                     keys.data(), values.data(), statuses.data());
+  std::vector<std::string> values =
+      MultiGet(key_strs, /*snapshot=*/nullptr, /*async=*/true,
+               optimize_multiget_for_io_, use_coroutine_);
   ASSERT_EQ(values.size(), 3);
-  ASSERT_OK(statuses[0]);
-  ASSERT_OK(statuses[1]);
-  ASSERT_OK(statuses[2]);
   ASSERT_EQ(values[0], "val_l0_" + std::to_string(0));
   ASSERT_EQ(values[1], "val_l0_" + std::to_string(40));
   ASSERT_EQ(values[2], "val_l0_" + std::to_string(80));
 
-  HistogramData multiget_io_batch_size;
-
-  statistics()->histogramData(MULTIGET_IO_BATCH_SIZE, &multiget_io_batch_size);
-
-  // With async IO, lookups will happen in parallel for each key
 #ifdef ROCKSDB_IOURING_PRESENT
-  if (GetParam()) {
-    ASSERT_EQ(multiget_io_batch_size.count, 1);
-    ASSERT_EQ(multiget_io_batch_size.max, 3);
+  if (UseCoroutineRead()) {
+    AssertMultiGetIOBatchSize(3, 1);
+    ASSERT_EQ(statistics()->getTickerCount(MULTIGET_COROUTINE_COUNT), 3);
+  } else if (optimize_multiget_for_io_) {
+    AssertMultiGetIOBatchSize(1, 3);
     ASSERT_EQ(statistics()->getTickerCount(MULTIGET_COROUTINE_COUNT), 3);
   } else {
     // Without Async IO, MultiGet will call MultiRead 3 times, once for each
     // L0 file
-    ASSERT_EQ(multiget_io_batch_size.count, 3);
+    AssertMultiGetIOBatchSize(3, 1);
   }
 #else   // ROCKSDB_IOURING_PRESENT
   ASSERT_EQ(statistics()->getTickerCount(MULTIGET_COROUTINE_COUNT), 0);
@@ -3849,64 +4025,58 @@ TEST_P(DBMultiGetAsyncIOTest, GetFromL0) {
 }
 
 TEST_P(DBMultiGetAsyncIOTest, GetFromL1) {
-  std::vector<std::string> key_strs;
-  std::vector<Slice> keys;
-  std::vector<PinnableSlice> values;
-  std::vector<Status> statuses;
-
-  key_strs.push_back(Key(33));
-  key_strs.push_back(Key(54));
-  key_strs.push_back(Key(102));
-  keys.emplace_back(key_strs[0]);
-  keys.emplace_back(key_strs[1]);
-  keys.emplace_back(key_strs[2]);
-  values.resize(keys.size());
-  statuses.resize(keys.size());
+  std::vector<std::string> key_strs{Key(33), Key(54), Key(102)};
 
   PrepareDBForTest();
 
-  ReadOptions ro;
-  ro.async_io = true;
-  ro.optimize_multiget_for_io = GetParam();
-  dbfull()->MultiGet(ro, dbfull()->DefaultColumnFamily(), keys.size(),
-                     keys.data(), values.data(), statuses.data());
+  std::vector<std::string> values =
+      MultiGet(key_strs, /*snapshot=*/nullptr, /*async=*/true,
+               optimize_multiget_for_io_, use_coroutine_);
   ASSERT_EQ(values.size(), 3);
-  ASSERT_EQ(statuses[0], Status::OK());
-  ASSERT_EQ(statuses[1], Status::OK());
-  ASSERT_EQ(statuses[2], Status::OK());
   ASSERT_EQ(values[0], "val_l1_" + std::to_string(33));
   ASSERT_EQ(values[1], "val_l1_" + std::to_string(54));
   ASSERT_EQ(values[2], "val_l1_" + std::to_string(102));
 
-  HistogramData multiget_io_batch_size;
-
-  statistics()->histogramData(MULTIGET_IO_BATCH_SIZE, &multiget_io_batch_size);
-
 #ifdef ROCKSDB_IOURING_PRESENT
-  // A batch of 3 async IOs is expected, one for each overlapping file in L1
-  ASSERT_EQ(multiget_io_batch_size.count, 1);
-  ASSERT_EQ(multiget_io_batch_size.max, 3);
-  ASSERT_EQ(statistics()->getTickerCount(MULTIGET_COROUTINE_COUNT), 3);
+  if (UseCoroutineRead()) {
+    AssertMultiGetIOBatchSize(3, 1);
+  } else {
+    // A batch of 3 async IOs is expected, one for each overlapping file in L1.
+    AssertMultiGetIOBatchSize(1, 3);
+  }
+  ASSERT_EQ(statistics()->getTickerCount(MULTIGET_COROUTINE_COUNT),
+            UseCoroutineRead() ? 6 : 3);
 #else   // ROCKSDB_IOURING_PRESENT
   ASSERT_EQ(statistics()->getTickerCount(MULTIGET_COROUTINE_COUNT), 0);
 #endif  // ROCKSDB_IOURING_PRESENT
 }
 
+TEST_P(DBMultiGetAsyncIOTest, SingleGetFromL1UsesCoroutineRead) {
+  int coroutine_read_count = 0;
+  SyncPoint::GetInstance()->SetCallBack(
+      "RandomAccessFileReader::ReadCoroutine:SubmitReadAsync",
+      [&](void*) { ++coroutine_read_count; });
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  PrepareDBForTest();
+
+  ASSERT_EQ(
+      "val_l1_" + std::to_string(33),
+      Get(Key(33), static_cast<const Snapshot*>(nullptr), use_coroutine_));
+
+  SyncPoint::GetInstance()->DisableProcessing();
+  SyncPoint::GetInstance()->ClearAllCallBacks();
+
+  if (UseCoroutineRead()) {
+    ASSERT_GT(coroutine_read_count, 0);
+  } else {
+    ASSERT_EQ(coroutine_read_count, 0);
+  }
+}
+
 #ifdef ROCKSDB_IOURING_PRESENT
 TEST_P(DBMultiGetAsyncIOTest, GetFromL1Error) {
-  std::vector<std::string> key_strs;
-  std::vector<Slice> keys;
-  std::vector<PinnableSlice> values;
-  std::vector<Status> statuses;
-
-  key_strs.push_back(Key(33));
-  key_strs.push_back(Key(54));
-  key_strs.push_back(Key(102));
-  keys.emplace_back(key_strs[0]);
-  keys.emplace_back(key_strs[1]);
-  keys.emplace_back(key_strs[2]);
-  values.resize(keys.size());
-  statuses.resize(keys.size());
+  std::vector<std::string> key_strs{Key(33), Key(54), Key(102)};
 
   int count = 0;
   SyncPoint::GetInstance()->SetCallBack(
@@ -3935,147 +4105,100 @@ TEST_P(DBMultiGetAsyncIOTest, GetFromL1Error) {
 
   PrepareDBForTest();
 
-  ReadOptions ro;
-  ro.async_io = true;
-  ro.optimize_multiget_for_io = GetParam();
-  dbfull()->MultiGet(ro, dbfull()->DefaultColumnFamily(), keys.size(),
-                     keys.data(), values.data(), statuses.data());
+  std::vector<std::string> values =
+      MultiGet(key_strs, /*snapshot=*/nullptr, /*async=*/true,
+               optimize_multiget_for_io_, use_coroutine_);
   SyncPoint::GetInstance()->DisableProcessing();
   ASSERT_EQ(values.size(), 3);
-  ASSERT_EQ(statuses[0], Status::OK());
-  ASSERT_EQ(statuses[1], Status::OK());
-  ASSERT_EQ(statuses[2], Status::IOError());
+  ASSERT_EQ(values[0], "val_l1_" + std::to_string(33));
+  ASSERT_EQ(values[1], "val_l1_" + std::to_string(54));
+  ASSERT_EQ(values[2], Status::IOError().ToString());
 
-  HistogramData multiget_io_batch_size;
-
-  statistics()->histogramData(MULTIGET_IO_BATCH_SIZE, &multiget_io_batch_size);
-
-  // A batch of 3 async IOs is expected, one for each overlapping file in L1
-  ASSERT_EQ(multiget_io_batch_size.count, 1);
-  ASSERT_EQ(multiget_io_batch_size.max, 2);
-  ASSERT_EQ(statistics()->getTickerCount(MULTIGET_COROUTINE_COUNT), 2);
+  if (UseCoroutineRead()) {
+    AssertMultiGetIOBatchSize(2, 1);
+  } else {
+    // A batch of 3 async IOs is expected, one for each overlapping file in L1.
+    AssertMultiGetIOBatchSize(1, 2);
+  }
+  ASSERT_EQ(statistics()->getTickerCount(MULTIGET_COROUTINE_COUNT),
+            UseCoroutineRead() ? 5 : 2);
 }
 #endif  // ROCKSDB_IOURING_PRESENT
 
 TEST_P(DBMultiGetAsyncIOTest, LastKeyInFile) {
-  std::vector<std::string> key_strs;
-  std::vector<Slice> keys;
-  std::vector<PinnableSlice> values;
-  std::vector<Status> statuses;
-
   // 21 is the last key in the first L1 file
-  key_strs.push_back(Key(21));
-  key_strs.push_back(Key(54));
-  key_strs.push_back(Key(102));
-  keys.emplace_back(key_strs[0]);
-  keys.emplace_back(key_strs[1]);
-  keys.emplace_back(key_strs[2]);
-  values.resize(keys.size());
-  statuses.resize(keys.size());
+  std::vector<std::string> key_strs{Key(21), Key(54), Key(102)};
 
   PrepareDBForTest();
 
-  ReadOptions ro;
-  ro.async_io = true;
-  ro.optimize_multiget_for_io = GetParam();
-  dbfull()->MultiGet(ro, dbfull()->DefaultColumnFamily(), keys.size(),
-                     keys.data(), values.data(), statuses.data());
+  std::vector<std::string> values =
+      MultiGet(key_strs, /*snapshot=*/nullptr, /*async=*/true,
+               optimize_multiget_for_io_, use_coroutine_);
   ASSERT_EQ(values.size(), 3);
-  ASSERT_EQ(statuses[0], Status::OK());
-  ASSERT_EQ(statuses[1], Status::OK());
-  ASSERT_EQ(statuses[2], Status::OK());
   ASSERT_EQ(values[0], "val_l1_" + std::to_string(21));
   ASSERT_EQ(values[1], "val_l1_" + std::to_string(54));
   ASSERT_EQ(values[2], "val_l1_" + std::to_string(102));
 
 #ifdef ROCKSDB_IOURING_PRESENT
-  HistogramData multiget_io_batch_size;
-
-  statistics()->histogramData(MULTIGET_IO_BATCH_SIZE, &multiget_io_batch_size);
-
   // Since the first MultiGet key is the last key in a file, the MultiGet is
   // expected to lookup in that file first, before moving on to other files.
   // So the first file lookup will issue one async read, and the next lookup
   // will lookup 2 files in parallel and issue 2 async reads
-  ASSERT_EQ(multiget_io_batch_size.count, 2);
-  ASSERT_EQ(multiget_io_batch_size.max, 2);
+  if (UseCoroutineRead()) {
+    AssertMultiGetIOBatchSize(3, 1);
+  } else {
+    AssertMultiGetIOBatchSize(2, 2);
+  }
 #endif  // ROCKSDB_IOURING_PRESENT
 }
 
 TEST_P(DBMultiGetAsyncIOTest, GetFromL1AndL2) {
-  std::vector<std::string> key_strs;
-  std::vector<Slice> keys;
-  std::vector<PinnableSlice> values;
-  std::vector<Status> statuses;
-
   // 33 and 102 are in L1, and 56 is in L2
-  key_strs.push_back(Key(33));
-  key_strs.push_back(Key(56));
-  key_strs.push_back(Key(102));
-  keys.emplace_back(key_strs[0]);
-  keys.emplace_back(key_strs[1]);
-  keys.emplace_back(key_strs[2]);
-  values.resize(keys.size());
-  statuses.resize(keys.size());
+  std::vector<std::string> key_strs{Key(33), Key(56), Key(102)};
 
   PrepareDBForTest();
 
-  ReadOptions ro;
-  ro.async_io = true;
-  ro.optimize_multiget_for_io = GetParam();
-  dbfull()->MultiGet(ro, dbfull()->DefaultColumnFamily(), keys.size(),
-                     keys.data(), values.data(), statuses.data());
+  std::vector<std::string> values =
+      MultiGet(key_strs, /*snapshot=*/nullptr, /*async=*/true,
+               optimize_multiget_for_io_, use_coroutine_);
   ASSERT_EQ(values.size(), 3);
-  ASSERT_EQ(statuses[0], Status::OK());
-  ASSERT_EQ(statuses[1], Status::OK());
-  ASSERT_EQ(statuses[2], Status::OK());
   ASSERT_EQ(values[0], "val_l1_" + std::to_string(33));
   ASSERT_EQ(values[1], "val_l2_" + std::to_string(56));
   ASSERT_EQ(values[2], "val_l1_" + std::to_string(102));
 
 #ifdef ROCKSDB_IOURING_PRESENT
-  HistogramData multiget_io_batch_size;
-
-  statistics()->histogramData(MULTIGET_IO_BATCH_SIZE, &multiget_io_batch_size);
-
   // There are 2 keys in L1 in twp separate files, and 1 in L2. With
   // optimize_multiget_for_io, all three lookups will happen in parallel.
   // Otherwise, the L2 lookup will happen after L1.
-  ASSERT_EQ(multiget_io_batch_size.count, GetParam() ? 1 : 2);
-  ASSERT_EQ(multiget_io_batch_size.max, GetParam() ? 3 : 2);
+  if (UseCoroutineRead()) {
+    AssertMultiGetIOBatchSize(3, 1);
+  } else {
+    AssertMultiGetIOBatchSize(optimize_multiget_for_io_ ? 1 : 2,
+                              optimize_multiget_for_io_ ? 3 : 2);
+  }
 #endif  // ROCKSDB_IOURING_PRESENT
 }
 
 TEST_P(DBMultiGetAsyncIOTest, GetFromL2WithRangeOverlapL0L1) {
-  std::vector<std::string> key_strs;
-  std::vector<Slice> keys;
-  std::vector<PinnableSlice> values;
-  std::vector<Status> statuses;
-
   // 19 and 26 are in L2, but overlap with L0 and L1 file ranges
-  key_strs.push_back(Key(19));
-  key_strs.push_back(Key(26));
-  keys.emplace_back(key_strs[0]);
-  keys.emplace_back(key_strs[1]);
-  values.resize(keys.size());
-  statuses.resize(keys.size());
+  std::vector<std::string> key_strs{Key(19), Key(26)};
 
   PrepareDBForTest();
 
-  ReadOptions ro;
-  ro.async_io = true;
-  ro.optimize_multiget_for_io = GetParam();
-  dbfull()->MultiGet(ro, dbfull()->DefaultColumnFamily(), keys.size(),
-                     keys.data(), values.data(), statuses.data());
+  std::vector<std::string> values =
+      MultiGet(key_strs, /*snapshot=*/nullptr, /*async=*/true,
+               optimize_multiget_for_io_, use_coroutine_);
   ASSERT_EQ(values.size(), 2);
-  ASSERT_EQ(statuses[0], Status::OK());
-  ASSERT_EQ(statuses[1], Status::OK());
   ASSERT_EQ(values[0], "val_l2_" + std::to_string(19));
   ASSERT_EQ(values[1], "val_l2_" + std::to_string(26));
 
 #ifdef ROCKSDB_IOURING_PRESENT
-  // Bloom filters in L0/L1 will avoid the coroutine calls in those levels
-  ASSERT_EQ(statistics()->getTickerCount(MULTIGET_COROUTINE_COUNT), 2);
+  // Bloom filters in L0/L1 avoid the coroutine calls in those levels. The
+  // coroutine path counts every per-file MultiGetFromSST coroutine (including
+  // serial reads), whereas the async_io path counts only the parallel overlap
+  // batch, so the coroutine path observes one more.
+  ASSERT_EQ(statistics()->getTickerCount(MULTIGET_COROUTINE_COUNT),
+            UseCoroutineRead() ? 3 : 2);
 #else   // ROCKSDB_IOURING_PRESENT
   ASSERT_EQ(statistics()->getTickerCount(MULTIGET_COROUTINE_COUNT), 0);
 #endif  // ROCKSDB_IOURING_PRESENT
@@ -4083,62 +4206,35 @@ TEST_P(DBMultiGetAsyncIOTest, GetFromL2WithRangeOverlapL0L1) {
 
 #ifdef ROCKSDB_IOURING_PRESENT
 TEST_P(DBMultiGetAsyncIOTest, GetFromL2WithRangeDelInL1) {
-  std::vector<std::string> key_strs;
-  std::vector<Slice> keys;
-  std::vector<PinnableSlice> values;
-  std::vector<Status> statuses;
-
   // 139 and 163 are in L2, but overlap with a range deletes in L1
-  key_strs.push_back(Key(139));
-  key_strs.push_back(Key(163));
-  keys.emplace_back(key_strs[0]);
-  keys.emplace_back(key_strs[1]);
-  values.resize(keys.size());
-  statuses.resize(keys.size());
+  std::vector<std::string> key_strs{Key(139), Key(163)};
 
   PrepareDBForTest();
 
-  ReadOptions ro;
-  ro.async_io = true;
-  ro.optimize_multiget_for_io = GetParam();
-  dbfull()->MultiGet(ro, dbfull()->DefaultColumnFamily(), keys.size(),
-                     keys.data(), values.data(), statuses.data());
+  std::vector<std::string> values =
+      MultiGet(key_strs, /*snapshot=*/nullptr, /*async=*/true,
+               optimize_multiget_for_io_, use_coroutine_);
   ASSERT_EQ(values.size(), 2);
-  ASSERT_EQ(statuses[0], Status::NotFound());
-  ASSERT_EQ(statuses[1], Status::NotFound());
+  ASSERT_EQ(values[0], "NOT_FOUND");
+  ASSERT_EQ(values[1], "NOT_FOUND");
 
   // Bloom filters in L0/L1 will avoid the coroutine calls in those levels
   ASSERT_EQ(statistics()->getTickerCount(MULTIGET_COROUTINE_COUNT), 2);
 }
 
 TEST_P(DBMultiGetAsyncIOTest, GetFromL1AndL2WithRangeDelInL1) {
-  std::vector<std::string> key_strs;
-  std::vector<Slice> keys;
-  std::vector<PinnableSlice> values;
-  std::vector<Status> statuses;
-
   // 139 and 163 are in L2, but overlap with a range deletes in L1
-  key_strs.push_back(Key(139));
-  key_strs.push_back(Key(144));
-  key_strs.push_back(Key(163));
-  keys.emplace_back(key_strs[0]);
-  keys.emplace_back(key_strs[1]);
-  keys.emplace_back(key_strs[2]);
-  values.resize(keys.size());
-  statuses.resize(keys.size());
+  std::vector<std::string> key_strs{Key(139), Key(144), Key(163)};
 
   PrepareDBForTest();
 
-  ReadOptions ro;
-  ro.async_io = true;
-  ro.optimize_multiget_for_io = GetParam();
-  dbfull()->MultiGet(ro, dbfull()->DefaultColumnFamily(), keys.size(),
-                     keys.data(), values.data(), statuses.data());
-  ASSERT_EQ(values.size(), keys.size());
-  ASSERT_EQ(statuses[0], Status::NotFound());
-  ASSERT_EQ(statuses[1], Status::OK());
+  std::vector<std::string> values =
+      MultiGet(key_strs, /*snapshot=*/nullptr, /*async=*/true,
+               optimize_multiget_for_io_, use_coroutine_);
+  ASSERT_EQ(values.size(), 3);
+  ASSERT_EQ(values[0], "NOT_FOUND");
   ASSERT_EQ(values[1], "val_l1_" + std::to_string(144));
-  ASSERT_EQ(statuses[2], Status::NotFound());
+  ASSERT_EQ(values[2], "NOT_FOUND");
 
   // Bloom filters in L0/L1 will avoid the coroutine calls in those levels
   ASSERT_EQ(statistics()->getTickerCount(MULTIGET_COROUTINE_COUNT), 3);
@@ -4146,32 +4242,18 @@ TEST_P(DBMultiGetAsyncIOTest, GetFromL1AndL2WithRangeDelInL1) {
 #endif  // ROCKSDB_IOURING_PRESENT
 
 TEST_P(DBMultiGetAsyncIOTest, GetNoIOUring) {
-  std::vector<std::string> key_strs;
-  std::vector<Slice> keys;
-  std::vector<PinnableSlice> values;
-  std::vector<Status> statuses;
-
-  key_strs.push_back(Key(33));
-  key_strs.push_back(Key(54));
-  key_strs.push_back(Key(102));
-  keys.emplace_back(key_strs[0]);
-  keys.emplace_back(key_strs[1]);
-  keys.emplace_back(key_strs[2]);
-  values.resize(keys.size());
-  statuses.resize(keys.size());
+  std::vector<std::string> key_strs{Key(33), Key(54), Key(102)};
 
   enable_io_uring = false;
   ReopenDB();
 
-  ReadOptions ro;
-  ro.async_io = true;
-  ro.optimize_multiget_for_io = GetParam();
-  dbfull()->MultiGet(ro, dbfull()->DefaultColumnFamily(), keys.size(),
-                     keys.data(), values.data(), statuses.data());
+  std::vector<std::string> values =
+      MultiGet(key_strs, /*snapshot=*/nullptr, /*async=*/true,
+               optimize_multiget_for_io_, use_coroutine_);
   ASSERT_EQ(values.size(), 3);
-  ASSERT_EQ(statuses[0], Status::OK());
-  ASSERT_EQ(statuses[1], Status::OK());
-  ASSERT_EQ(statuses[2], Status::OK());
+  ASSERT_EQ(values[0], "val_l1_" + std::to_string(33));
+  ASSERT_EQ(values[1], "val_l1_" + std::to_string(54));
+  ASSERT_EQ(values[2], "val_l1_" + std::to_string(102));
 
   HistogramData async_read_bytes;
 
@@ -4179,11 +4261,17 @@ TEST_P(DBMultiGetAsyncIOTest, GetNoIOUring) {
 
   // A batch of 3 async IOs is expected, one for each overlapping file in L1
   ASSERT_EQ(async_read_bytes.count, 0);
-  ASSERT_EQ(statistics()->getTickerCount(MULTIGET_COROUTINE_COUNT), 0);
+  const uint64_t coroutine_count =
+      statistics()->getTickerCount(MULTIGET_COROUTINE_COUNT);
+  if (UseCoroutineRead()) {
+    ASSERT_GT(coroutine_count, 0);
+  } else {
+    ASSERT_EQ(coroutine_count, 0);
+  }
 }
 
 INSTANTIATE_TEST_CASE_P(DBMultiGetAsyncIOTest, DBMultiGetAsyncIOTest,
-                        testing::Bool());
+                        testing::Combine(testing::Bool(), testing::Bool()));
 #endif  // USE_COROUTINES
 
 TEST_F(DBBasicTest, MultiGetStats) {
@@ -5459,9 +5547,9 @@ class DBBasicTestMultiGet : public DBTestBase {
   std::vector<std::string> cf_names_;
 };
 
-class DBBasicTestWithParallelIO : public DBBasicTestMultiGet,
-                                  public testing::WithParamInterface<
-                                      std::tuple<bool, bool, bool, uint32_t>> {
+class DBBasicTestWithParallelIO : public testing::WithParamInterface<
+                                      std::tuple<bool, bool, bool, uint32_t>>,
+                                  public DBBasicTestMultiGet {
  public:
   DBBasicTestWithParallelIO()
       : DBBasicTestMultiGet("/db_basic_test_with_parallel_io", 1,
@@ -6331,7 +6419,7 @@ TEST_F(DBBasicTest, DisallowMemtableWrite) {
   Options options_disallow = options_allow;
   options_disallow.disallow_memtable_writes = true;
   options_disallow.paranoid_memory_checks = true;
-  options_disallow.memtable_veirfy_per_key_checksum_on_seek = true;
+  options_disallow.memtable_verify_per_key_checksum_on_seek = true;
 
   DestroyAndReopen(options_allow);
   // CFs allowing and disallowing memtable write

db/db_block_cache_test.cc

@@ -1482,10 +1482,13 @@ TEST_P(DBBlockCacheTypeTest, AddRedundantStats) {
   // Access just data, forcing redundant load+insert
   ReadOptions read_options;
   std::unique_ptr<Iterator> iter{db_->NewIterator(read_options)};
+  ASSERT_OK(iter->Refresh());
+
   cache->SetNthLookupNotFound(1);
   iter->SeekToFirst();
   ASSERT_TRUE(iter->Valid());
   ASSERT_EQ(iter->key(), "bar");
+  ASSERT_EQ(iter->value(), "value");
 
   EXPECT_EQ(2, TestGetTickerCount(options, BLOCK_CACHE_INDEX_ADD));
   EXPECT_EQ(2, TestGetTickerCount(options, BLOCK_CACHE_FILTER_ADD));

db/db_compaction_test.cc

@@ -8,11 +8,13 @@
 // found in the LICENSE file. See the AUTHORS file for names of contributors.
 
 #include <tuple>
+#include <utility>
 
 #include "compaction/compaction_picker_universal.h"
 #include "db/blob/blob_index.h"
 #include "db/db_test_util.h"
 #include "db/dbformat.h"
+#include "db/table_cache.h"
 #include "env/mock_env.h"
 #include "file/filename.h"
 #include "port/port.h"
@@ -6651,6 +6653,572 @@ TEST_P(DBCompactionDirectIOTest, DirectIO) {
 INSTANTIATE_TEST_CASE_P(DBCompactionDirectIOTest, DBCompactionDirectIOTest,
                         testing::Bool());
 
+// With use_direct_io_for_compaction_reads OFF, compaction reads must stay
+// buffered: neither the compaction-input FileOptions nor a kernel O_DIRECT
+// open should fire. Runs on every platform (the sync points just don't fire
+// where O_DIRECT isn't reachable). Pairs with
+// UseDirectIoForCompactionReadsEndToEnd for the on case.
+TEST_F(DBCompactionTest, UseDirectIoForCompactionReadsOffStaysBuffered) {
+  Options options = CurrentOptions();
+  Destroy(options);
+  options.create_if_missing = true;
+  options.disable_auto_compactions = true;
+  options.use_direct_reads = false;
+  options.use_direct_io_for_compaction_reads = false;
+  options.use_direct_io_for_flush_and_compaction = false;
+
+  std::atomic<bool> observed_direct_compaction_read{false};
+  std::atomic<int> observed_callbacks{0};
+  std::atomic<int> observed_odirect_opens{0};
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "CompactionJob::CreateInputIterator:InputFileOptions", [&](void* arg) {
+        const auto* fo = static_cast<const FileOptions*>(arg);
+        observed_callbacks.fetch_add(1, std::memory_order_relaxed);
+        if (fo->use_direct_reads) {
+          observed_direct_compaction_read.store(true,
+                                                std::memory_order_relaxed);
+        }
+      });
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "NewRandomAccessFile:O_DIRECT", [&](void* /*arg*/) {
+        observed_odirect_opens.fetch_add(1, std::memory_order_relaxed);
+      });
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
+
+  ASSERT_OK(TryReopen(options));
+
+  const std::string value(4096, 'v');
+  for (int i = 0; i < 64; ++i) {
+    ASSERT_OK(Put(Key(i), value));
+  }
+  ASSERT_OK(Flush());
+  for (int i = 0; i < 64; ++i) {
+    ASSERT_OK(Put(Key(i), value));
+  }
+  ASSERT_OK(Flush());
+
+  ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
+
+  ASSERT_GT(observed_callbacks.load(), 0);
+  ASSERT_FALSE(observed_direct_compaction_read.load());
+  ASSERT_EQ(0, observed_odirect_opens.load());
+
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->ClearAllCallBacks();
+  Destroy(options);
+}
+
+TEST_F(DBCompactionTest,
+       UseDirectIoForCompactionReadsUsesBoundedEphemeralReaders) {
+  auto fs = std::make_shared<MockFileSystem>(Env::Default()->GetSystemClock(),
+                                             /*supports_direct_io=*/true);
+  std::unique_ptr<Env> mock_env = NewCompositeEnv(fs);
+
+  Options options = CurrentOptions();
+  options.env = mock_env.get();
+  Destroy(options);
+  options.create_if_missing = true;
+  options.disable_auto_compactions = true;
+  options.use_direct_reads = false;
+  options.use_direct_io_for_compaction_reads = true;
+  options.use_direct_io_for_flush_and_compaction = false;
+  options.max_subcompactions = 3;
+  options.statistics = CreateDBStatistics();
+
+  BlockBasedTableOptions table_options;
+  table_options.block_cache = NewLRUCache(64 << 20);
+  table_options.cache_index_and_filter_blocks = true;
+  table_options.pin_l0_filter_and_index_blocks_in_cache = true;
+  table_options.filter_policy.reset(NewBloomFilterPolicy(10));
+  options.table_factory.reset(NewBlockBasedTableFactory(table_options));
+
+  std::atomic<int> fresh_reader_opens{0};
+  std::atomic<int> malformed_fresh_reader_options{0};
+  std::atomic<int> avoid_shared_metadata_cache_opens{0};
+  std::atomic<int> shared_metadata_cache_uses{0};
+  std::atomic<size_t> input_iterator_file_bound{0};
+
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "TableCache::FindTable:FreshTableReader", [&](void* arg) {
+        const auto* open_options = static_cast<TableCacheOpenOptions*>(arg);
+        fresh_reader_opens.fetch_add(1, std::memory_order_relaxed);
+        if (!open_options->open_ephemeral_table_reader ||
+            !open_options->avoid_shared_metadata_cache ||
+            !open_options->skip_filters) {
+          malformed_fresh_reader_options.fetch_add(1,
+                                                   std::memory_order_relaxed);
+        }
+      });
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "BlockBasedTable::PrefetchIndexAndFilterBlocks:SharedMetadataCache",
+      [&](void* arg) {
+        const auto* context = static_cast<std::pair<bool, bool>*>(arg);
+        const bool avoid_shared_metadata_cache = context->first;
+        const bool use_cache = context->second;
+        if (avoid_shared_metadata_cache) {
+          avoid_shared_metadata_cache_opens.fetch_add(
+              1, std::memory_order_relaxed);
+          if (use_cache) {
+            shared_metadata_cache_uses.fetch_add(1, std::memory_order_relaxed);
+          }
+        }
+      });
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "VersionSet::MakeInputIterator:NewCompactionMergingIterator",
+      [&](void* arg) {
+        input_iterator_file_bound.fetch_add(*static_cast<size_t*>(arg),
+                                            std::memory_order_relaxed);
+      });
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
+
+  ASSERT_OK(TryReopen(options));
+
+  const std::string value(4096, 'v');
+  for (int i = 0; i < 64; ++i) {
+    ASSERT_OK(Put(Key(i), value));
+  }
+  ASSERT_OK(Flush());
+  for (int i = 0; i < 64; ++i) {
+    ASSERT_OK(Put(Key(i), value));
+  }
+  ASSERT_OK(Flush());
+
+  ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
+
+  ASSERT_GT(fresh_reader_opens.load(), 0);
+  EXPECT_EQ(0, malformed_fresh_reader_options.load());
+  ASSERT_GT(avoid_shared_metadata_cache_opens.load(), 0);
+  EXPECT_EQ(0, shared_metadata_cache_uses.load());
+  ASSERT_GT(input_iterator_file_bound.load(), 0);
+  EXPECT_LE(static_cast<size_t>(fresh_reader_opens.load()),
+            input_iterator_file_bound.load());
+
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->ClearAllCallBacks();
+  Destroy(options);
+}
+
+// End-to-end check that use_direct_io_for_compaction_reads opens compaction
+// inputs with O_DIRECT while use_direct_reads stays off (user reads buffered).
+// The NewRandomAccessFile:O_DIRECT sync point in env/fs_posix.cc fires once
+// per fresh open with the O_DIRECT flag, so this proves the kernel path, not
+// just the FileOptions. Only runs on platforms that take the O_DIRECT path.
+#if !defined(OS_MACOSX) && !defined(OS_OPENBSD) && !defined(OS_SOLARIS) && \
+    !defined(OS_WIN)
+TEST_F(DBCompactionTest, UseDirectIoForCompactionReadsEndToEnd) {
+  if (!IsDirectIOSupported()) {
+    ROCKSDB_GTEST_BYPASS("Direct IO not supported");
+    return;
+  }
+
+  Options options = CurrentOptions();
+  Destroy(options);
+  options.create_if_missing = true;
+  options.disable_auto_compactions = true;
+  // User reads stay buffered, compaction reads should switch to O_DIRECT.
+  options.use_direct_reads = false;
+  options.use_direct_io_for_compaction_reads = true;
+  // Isolate the read-side change; leave the compaction write path buffered.
+  options.use_direct_io_for_flush_and_compaction = false;
+
+  // Sync-point callbacks fire on compaction threads while the test thread
+  // reads these counters, so use atomics to avoid a data race.
+  std::atomic<int> observed_run_starts{0};
+  std::atomic<int> observed_odirect_opens{0};
+  std::atomic<bool> observed_direct_compaction_read{false};
+  std::atomic<int> observed_callbacks{0};
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency({});
+  // Plumbing-level probe: the compaction-input FileOptions should carry
+  // use_direct_reads = true when the new flag is enabled.
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "CompactionJob::CreateInputIterator:InputFileOptions", [&](void* arg) {
+        const auto* fo = static_cast<const FileOptions*>(arg);
+        observed_callbacks.fetch_add(1, std::memory_order_relaxed);
+        if (fo != nullptr && fo->use_direct_reads) {
+          observed_direct_compaction_read.store(true,
+                                                std::memory_order_relaxed);
+        }
+      });
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "CompactionJob::Run():Start", [&](void* /*arg*/) {
+        observed_run_starts.fetch_add(1, std::memory_order_relaxed);
+      });
+  // Kernel-level probe: this fires only when open() is issued with O_DIRECT,
+  // proving we change the actual cache mode, not just the FileOptions struct.
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "NewRandomAccessFile:O_DIRECT", [&](void* /*arg*/) {
+        observed_odirect_opens.fetch_add(1, std::memory_order_relaxed);
+      });
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
+
+  Status s = TryReopen(options);
+  if (s.IsNotSupported() || s.IsInvalidArgument()) {
+    ROCKSDB_GTEST_BYPASS(
+        "Direct IO reads not supported in this test environment");
+    ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
+    ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->ClearAllCallBacks();
+    return;
+  }
+  ASSERT_OK(s);
+
+  // Produce two L0 files with OVERLAPPING key ranges so that CompactRange has
+  // actual merge work to do (otherwise RocksDB performs a trivial file move
+  // and never constructs a CompactionJob).
+  const std::string value(4096, 'v');
+  for (int i = 0; i < 64; ++i) {
+    ASSERT_OK(Put(Key(i), value));
+  }
+  ASSERT_OK(Flush());
+  for (int i = 0; i < 64; ++i) {
+    ASSERT_OK(Put(Key(i), value));
+  }
+  ASSERT_OK(Flush());
+
+  // User reads should still go through the buffered path. Confirm that the
+  // option does not silently flip use_direct_reads for user reads.
+  for (int i = 0; i < 8; ++i) {
+    std::string actual;
+    ASSERT_OK(db_->Get(ReadOptions(), Key(i), &actual));
+    ASSERT_EQ(value, actual);
+  }
+
+  ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+  // Wait for compaction to complete and CompactionJob to be constructed.
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
+
+  // Confirm the compaction actually ran; otherwise the missing sync-point hits
+  // would be a test-setup problem, not an option regression.
+  ASSERT_GT(observed_run_starts.load(), 0)
+      << "CompactionJob::Run():Start never fired; CompactRange did not "
+         "schedule a compaction.";
+  ASSERT_GT(observed_callbacks.load(), 0);
+  ASSERT_TRUE(observed_direct_compaction_read.load());
+  // At least one compaction-input open went through O_DIRECT. Without the
+  // TableCache bypass this would be zero, since compaction would reuse the
+  // buffered handles cached for user reads.
+  EXPECT_GT(observed_odirect_opens.load(), 0)
+      << "no compaction-input opens went through O_DIRECT; "
+         "observed_odirect_opens="
+      << observed_odirect_opens.load();
+
+  // Quick sanity sweep after compaction to confirm data is intact.
+  for (int i = 0; i < 64; ++i) {
+    std::string actual;
+    ASSERT_OK(db_->Get(ReadOptions(), Key(i), &actual));
+    ASSERT_EQ(value, actual);
+  }
+
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->ClearAllCallBacks();
+  Destroy(options);
+}
+
+// Exercise the LevelIterator bypass path (L1+ compactions) with range
+// tombstones, where the ephemeral TableReader's lifetime is coupled to the
+// range_tombstone_iter the file iterator returns. The end-to-end test above
+// only makes two L0 files, which take the direct NewIterator path and never
+// hit LevelIterator. Here we build L1/L2 with tombstones and compact L1->L2 so
+// LevelIterator::NewFileIterator drives the bypass; a wrong reader/iterator
+// lifetime would crash or trip sanitizers as LevelIterator switches files.
+// Correctness is checked by computing the expected state of every key and
+// asserting Get() matches: wave-2 puts beat wave-1, and each key's most-recent
+// covering tombstone shadows older puts.
+TEST_F(DBCompactionTest,
+       UseDirectIoForCompactionReadsLevelIteratorWithTombstones) {
+  if (!IsDirectIOSupported()) {
+    ROCKSDB_GTEST_BYPASS("Direct IO not supported");
+    return;
+  }
+
+  Options options = CurrentOptions();
+  Destroy(options);
+  options.create_if_missing = true;
+  options.disable_auto_compactions = true;
+  options.use_direct_reads = false;
+  options.use_direct_io_for_compaction_reads = true;
+  options.use_direct_io_for_flush_and_compaction = false;
+  // Small files / small level base so we can pack data into L1 and L2 with
+  // a few flushes and CompactRange calls instead of needing millions of keys.
+  options.write_buffer_size = 64 * 1024;
+  options.target_file_size_base = 64 * 1024;
+  options.max_bytes_for_level_base = 256 * 1024;
+  options.level0_file_num_compaction_trigger = 100;  // never auto-trigger
+
+  std::atomic<int> observed_odirect_opens{0};
+  std::atomic<int> observed_run_starts{0};
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "NewRandomAccessFile:O_DIRECT", [&](void* /*arg*/) {
+        observed_odirect_opens.fetch_add(1, std::memory_order_relaxed);
+      });
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "CompactionJob::Run():Start", [&](void* /*arg*/) {
+        observed_run_starts.fetch_add(1, std::memory_order_relaxed);
+      });
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
+
+  Status s = TryReopen(options);
+  if (s.IsNotSupported() || s.IsInvalidArgument()) {
+    ROCKSDB_GTEST_BYPASS(
+        "Direct IO reads not supported in this test environment");
+    ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
+    ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->ClearAllCallBacks();
+    return;
+  }
+  ASSERT_OK(s);
+
+  // Use distinguishable values per wave so we can verify which wave's put
+  // won the merge for each key, not just "some put won".
+  const std::string wave1_value(1024, '1');
+  const std::string wave2_value(1024, '2');
+
+  auto write_batch = [&](int begin, int end, const std::string& value,
+                         bool with_range_tombstone) {
+    for (int i = begin; i < end; ++i) {
+      ASSERT_OK(Put(Key(i), value));
+    }
+    if (with_range_tombstone) {
+      // Drop a slice in the middle of the just-written range. The
+      // DeleteRange follows the puts in this batch, so its sequence number
+      // is higher and it shadows the puts on [del_lo, del_hi) within this
+      // SST.
+      ASSERT_OK(db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(),
+                                 Key(begin + (end - begin) / 4),
+                                 Key(begin + 3 * (end - begin) / 4)));
+    }
+    ASSERT_OK(Flush());
+  };
+
+  // Wave 1: four flushed SSTs covering [0, 800), each with a tombstone
+  // covering the middle half of its own range. Then compact down so the
+  // next phase exercises LevelIterator over L1+ files.
+  constexpr int kWave1Begin = 0;
+  constexpr int kWave1End = 800;
+  constexpr int kWave1BatchSize = 200;
+  for (int batch_begin = kWave1Begin; batch_begin < kWave1End;
+       batch_begin += kWave1BatchSize) {
+    write_batch(batch_begin, batch_begin + kWave1BatchSize, wave1_value,
+                /*with_range_tombstone=*/true);
+  }
+  ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
+
+  // Wave 2: two more flushed SSTs at L0 that overlap wave 1, each with their
+  // own range tombstone. The puts in wave 2 have higher seqno than wave 1,
+  // so they win when not tombstoned.
+  struct Wave2Range {
+    int begin;
+    int end;
+  };
+  const Wave2Range wave2_ranges[] = {{50, 250}, {350, 550}};
+  for (const auto& r : wave2_ranges) {
+    write_batch(r.begin, r.end, wave2_value, /*with_range_tombstone=*/true);
+  }
+
+  const int run_starts_before = observed_run_starts.load();
+  const int odirect_before = observed_odirect_opens.load();
+
+  // Compact everything together, forcing a LevelIterator over the lower-level
+  // files on the bypass path. Wrong ephemeral reader / tombstone-iter
+  // lifetimes should trip sanitizers here.
+  ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+  ASSERT_OK(dbfull()->TEST_WaitForCompact());
+
+  ASSERT_GT(observed_run_starts.load(), run_starts_before)
+      << "expected at least one compaction to run during the L1+ phase";
+  EXPECT_GT(observed_odirect_opens.load(), odirect_before)
+      << "no compaction-input opens went through O_DIRECT during L1+ "
+         "compaction; LevelIterator bypass path may be broken";
+
+  // Compute the precise expected state per key from the test parameters.
+  // For each k in the wave-1 range:
+  //   - If k is inside a wave-2 batch: wave-2 wins. NotFound iff k is in
+  //     that batch's tombstone range; otherwise present with wave2_value.
+  //   - Else: wave 1 wins. NotFound iff k is in its batch's tombstone
+  //     range; otherwise present with wave1_value.
+  enum class Expectation { kAbsent, kWave1Value, kWave2Value };
+  auto classify = [&](int k) -> Expectation {
+    // Wave 2 first (it shadows wave 1 wherever it covers).
+    for (const auto& r : wave2_ranges) {
+      if (k >= r.begin && k < r.end) {
+        const int width = r.end - r.begin;
+        const int del_lo = r.begin + width / 4;
+        const int del_hi = r.begin + 3 * width / 4;
+        if (k >= del_lo && k < del_hi) {
+          return Expectation::kAbsent;
+        }
+        return Expectation::kWave2Value;
+      }
+    }
+    // Fall back to wave 1.
+    const int batch = (k - kWave1Begin) / kWave1BatchSize;
+    const int batch_begin = kWave1Begin + batch * kWave1BatchSize;
+    const int del_lo = batch_begin + kWave1BatchSize / 4;
+    const int del_hi = batch_begin + 3 * kWave1BatchSize / 4;
+    if (k >= del_lo && k < del_hi) {
+      return Expectation::kAbsent;
+    }
+    return Expectation::kWave1Value;
+  };
+
+  int present_w1 = 0;
+  int present_w2 = 0;
+  int absent = 0;
+  std::string actual;
+  for (int k = kWave1Begin; k < kWave1End; ++k) {
+    const Status get_s = db_->Get(ReadOptions(), Key(k), &actual);
+    const Expectation exp = classify(k);
+    switch (exp) {
+      case Expectation::kAbsent:
+        EXPECT_TRUE(get_s.IsNotFound())
+            << "key " << k << " expected NotFound (covered by tombstone); "
+            << "got status=" << get_s.ToString()
+            << " value_len=" << (get_s.ok() ? actual.size() : 0);
+        ++absent;
+        break;
+      case Expectation::kWave1Value:
+        ASSERT_OK(get_s) << "key " << k << " expected wave-1 value";
+        EXPECT_EQ(wave1_value, actual)
+            << "key " << k << " expected wave-1 value but got a different one";
+        ++present_w1;
+        break;
+      case Expectation::kWave2Value:
+        ASSERT_OK(get_s) << "key " << k << " expected wave-2 value";
+        EXPECT_EQ(wave2_value, actual)
+            << "key " << k << " expected wave-2 value but got a different one";
+        ++present_w2;
+        break;
+    }
+  }
+  // All three buckets should be non-empty, so the test really exercises both
+  // tombstone paths and the wave-2-wins path. A zero here means the setup
+  // drifted and the setup (not these checks) should be fixed.
+  EXPECT_GT(present_w1, 0);
+  EXPECT_GT(present_w2, 0);
+  EXPECT_GT(absent, 0);
+
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->ClearAllCallBacks();
+  Destroy(options);
+}
+
+// With the option enabled, run reader threads against the live DB while manual
+// compactions run in the background. Each compaction-input file is open through
+// an ephemeral O_DIRECT handle while the shared TableCache still serves user
+// reads through a buffered handle, so the same SST is open in two cache modes
+// at once. This stresses that coexistence under TSAN/ASAN/UBSAN; it asserts
+// reads stay consistent and final values match the last writes, not anything
+// about timing.
+TEST_F(DBCompactionTest, UseDirectIoForCompactionReadsConcurrentReadStress) {
+  if (!IsDirectIOSupported()) {
+    ROCKSDB_GTEST_BYPASS("Direct IO not supported");
+    return;
+  }
+
+  Options options = CurrentOptions();
+  Destroy(options);
+  options.create_if_missing = true;
+  options.disable_auto_compactions = true;
+  options.use_direct_reads = false;
+  options.use_direct_io_for_compaction_reads = true;
+  options.use_direct_io_for_flush_and_compaction = false;
+  options.write_buffer_size = 64 * 1024;
+  options.target_file_size_base = 64 * 1024;
+  options.max_bytes_for_level_base = 256 * 1024;
+  options.level0_file_num_compaction_trigger = 100;  // never auto-trigger
+
+  Status s = TryReopen(options);
+  if (s.IsNotSupported() || s.IsInvalidArgument()) {
+    ROCKSDB_GTEST_BYPASS(
+        "Direct IO reads not supported in this test environment");
+    return;
+  }
+  ASSERT_OK(s);
+
+  constexpr int kNumKeys = 512;
+  constexpr int kValueSize = 256;
+  const std::string base_value(kValueSize, 'b');
+
+  // Initial population, flushed into a few L0 SSTs.
+  for (int i = 0; i < kNumKeys; ++i) {
+    ASSERT_OK(Put(Key(i), base_value));
+    if (i > 0 && i % 128 == 0) {
+      ASSERT_OK(Flush());
+    }
+  }
+  ASSERT_OK(Flush());
+
+  std::atomic<bool> stop{false};
+  std::atomic<int64_t> reads_observed{0};
+  std::atomic<int64_t> read_errors{0};
+
+  // Reader threads loop until stop is set, recording any status that isn't OK
+  // or NotFound. The point is to surface Corruption/IOError/use-after-free, so
+  // the main thread can fail the test.
+  constexpr int kNumReaders = 4;
+  std::vector<port::Thread> readers;
+  readers.reserve(kNumReaders);
+  for (int t = 0; t < kNumReaders; ++t) {
+    readers.emplace_back([&, t]() {
+      std::string value;
+      while (!stop.load(std::memory_order_acquire)) {
+        const int k =
+            (t * 7919 +
+             static_cast<int>(reads_observed.load(std::memory_order_relaxed))) %
+            kNumKeys;
+        Status read_s = db_->Get(ReadOptions(), Key(k), &value);
+        if (!read_s.ok() && !read_s.IsNotFound()) {
+          read_errors.fetch_add(1, std::memory_order_relaxed);
+        }
+        reads_observed.fetch_add(1, std::memory_order_relaxed);
+      }
+    });
+  }
+
+  // Drive a few rounds of writes and compactions on the main thread while
+  // the readers hammer Get(). Each round overwrites every key with a
+  // round-tagged value and then compacts.
+  constexpr int kRounds = 4;
+  std::string last_value = base_value;
+  for (int round = 0; round < kRounds; ++round) {
+    const std::string round_value(kValueSize,
+                                  static_cast<char>('A' + (round % 26)));
+    for (int i = 0; i < kNumKeys; ++i) {
+      ASSERT_OK(Put(Key(i), round_value));
+      if (i > 0 && i % 64 == 0) {
+        ASSERT_OK(Flush());
+      }
+    }
+    ASSERT_OK(Flush());
+    ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+    ASSERT_OK(dbfull()->TEST_WaitForCompact());
+    last_value = round_value;
+  }
+
+  stop.store(true, std::memory_order_release);
+  for (auto& reader : readers) {
+    reader.join();
+  }
+
+  // Every key must now hold the final round's value. Catches wholesale
+  // corruption the sampling readers might miss, and confirms compaction
+  // finished under the bypass path.
+  std::string value;
+  for (int i = 0; i < kNumKeys; ++i) {
+    ASSERT_OK(db_->Get(ReadOptions(), Key(i), &value));
+    EXPECT_EQ(last_value, value);
+  }
+  EXPECT_GT(reads_observed.load(), 0)
+      << "reader threads never observed a single read; test was a no-op";
+  EXPECT_EQ(0, read_errors.load())
+      << "concurrent reads against compaction with bypass-path saw "
+      << read_errors.load() << " non-OK/non-NotFound status returns";
+
+  Destroy(options);
+}
+#endif  // !defined(OS_MACOSX) && !defined(OS_OPENBSD) && ...
+
 class CompactionPriTest : public DBTestBase,
                           public testing::WithParamInterface<uint32_t> {
  public:

db/db_etc2_test.cc

@@ -28,6 +28,7 @@
 #include "rocksdb/wal_filter.h"
 #include "test_util/testutil.h"
 #include "util/defer.h"
+#include "util/mutexlock.h"
 #include "util/random.h"
 #include "utilities/fault_injection_env.h"
 
@@ -330,6 +331,8 @@ class DBTestSharedWriteBufferAcrossCFs
 
 TEST_P(DBTestSharedWriteBufferAcrossCFs, SharedWriteBufferAcrossCFs) {
   Options options = CurrentOptions();
+  options.statistics = CreateDBStatistics();
+  options.statistics->set_stats_level(StatsLevel::kAll);
   options.arena_block_size = 4096;
   auto flush_listener = std::make_shared<FlushCounterListener>();
   options.listeners.push_back(flush_listener);
@@ -497,6 +500,29 @@ TEST_P(DBTestSharedWriteBufferAcrossCFs, SharedWriteBufferAcrossCFs) {
     ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_.get(), "nikitich"),
               static_cast<uint64_t>(2));
   }
+  const uint64_t wbm_flushes =
+      TestGetTickerCount(options, FLUSH_REASON_WRITE_BUFFER_MANAGER);
+  ASSERT_GT(wbm_flushes, 0);
+  EXPECT_EQ(0, TestGetTickerCount(options, FLUSH_REASON_WRITE_BUFFER_FULL));
+
+  HistogramData all_memtable_memory;
+  options.statistics->histogramData(FLUSH_MEMTABLE_MEMORY_BYTES,
+                                    &all_memtable_memory);
+  EXPECT_GE(all_memtable_memory.count, wbm_flushes);
+  EXPECT_GT(all_memtable_memory.sum, 0);
+
+  HistogramData all_memtable_data_size;
+  options.statistics->histogramData(FLUSH_MEMTABLE_TOTAL_DATA_SIZE,
+                                    &all_memtable_data_size);
+  EXPECT_GE(all_memtable_data_size.count, wbm_flushes);
+  EXPECT_GT(all_memtable_data_size.sum, 0);
+
+  HistogramData wbm_memtable_memory;
+  options.statistics->histogramData(
+      FLUSH_WRITE_BUFFER_MANAGER_MEMTABLE_MEMORY_BYTES, &wbm_memtable_memory);
+  EXPECT_EQ(wbm_flushes, wbm_memtable_memory.count);
+  EXPECT_GT(wbm_memtable_memory.sum, 0);
+
   if (cost_cache_) {
     ASSERT_GE(cache->GetUsage(), 256 * 1024);
     Close();
@@ -7877,7 +7903,10 @@ class FastOpenTestFS : public FileSystemWrapper {
                                IODebugContext* dbg) override {
     if (file_opts.file_metadata != nullptr) {
       metadata_passed_on_open_.fetch_add(1, std::memory_order_relaxed);
-      last_metadata_received_ = *file_opts.file_metadata;
+      {
+        MutexLock l(&last_metadata_mu_);
+        last_metadata_received_ = *file_opts.file_metadata;
+      }
     }
     IOStatus s = target()->NewRandomAccessFile(fname, file_opts, result, dbg);
     if (s.ok()) {
@@ -7896,18 +7925,21 @@ class FastOpenTestFS : public FileSystemWrapper {
   }
 
   std::string GetLastMetadataReceived() const {
+    MutexLock l(&last_metadata_mu_);
     return last_metadata_received_;
   }
 
   void ResetCounters() {
     metadata_retrieved_count_.store(0, std::memory_order_relaxed);
     metadata_passed_on_open_.store(0, std::memory_order_relaxed);
+    MutexLock l(&last_metadata_mu_);
     last_metadata_received_.clear();
   }
 
  private:
   std::atomic<int> metadata_retrieved_count_{0};
   std::atomic<int> metadata_passed_on_open_{0};
+  mutable port::Mutex last_metadata_mu_;
   std::string last_metadata_received_;
 };
 
@@ -8249,6 +8281,122 @@ TEST_F(DBTest2, FastSstOpenDisableAfterMetadataPersisted) {
   ASSERT_OK(DestroyDB(test_dbname, options));
 }
 
+TEST_F(DBTest2, FastSstOpenMetadataPreservedAcrossManifestRotation) {
+  // Regression test: FileMetaData::file_open_metadata for unchanged live
+  // SSTs must survive a MANIFEST rotation. WriteCurrentStateToManifest
+  // snapshots live SSTs into a new MANIFEST; if it drops file_open_metadata
+  // for those unchanged SSTs, the fast_sst_open optimization is silently
+  // disabled for them until they are rewritten by flush/compaction/ingest.
+  auto test_fs = std::make_shared<FastOpenTestFS>(env_->GetFileSystem());
+  std::unique_ptr<Env> test_env(new CompositeEnvWrapper(env_, test_fs));
+
+  Options options;
+  options.env = test_env.get();
+  options.fast_sst_open = true;
+  options.create_if_missing = true;
+  // Keep the pre-rotation SST from being rewritten by compaction.
+  options.disable_auto_compactions = true;
+
+  std::string test_dbname = test::PerThreadDBPath("fast_open_manifest_rotate");
+  ASSERT_OK(DestroyDB(test_dbname, options));
+
+  std::unique_ptr<DB> db;
+  ASSERT_OK(DB::Open(options, test_dbname, &db));
+
+  // Flush an SST; metadata is retrieved and persisted in the MANIFEST.
+  ASSERT_OK(db->Put(WriteOptions(), "key1", "value1"));
+  ASSERT_OK(db->Flush(FlushOptions()));
+  ASSERT_EQ(1, test_fs->GetMetadataRetrievedCount());
+
+  // Shrink max_manifest_file_size so the next MANIFEST write rotates and
+  // exercises WriteCurrentStateToManifest.
+  ASSERT_OK(db->SetDBOptions(
+      {{"max_manifest_file_size", "1"}, {"max_manifest_space_amp_pct", "0"}}));
+
+  uint64_t manifest_before =
+      static_cast<DBImpl*>(db.get())->TEST_Current_Manifest_FileNo();
+
+  // Flush a second SST to trigger the rotation. SST #1 is unchanged, so
+  // the rotated MANIFEST snapshot must carry SST #1's file_open_metadata.
+  ASSERT_OK(db->Put(WriteOptions(), "key2", "value2"));
+  ASSERT_OK(db->Flush(FlushOptions()));
+
+  uint64_t manifest_after =
+      static_cast<DBImpl*>(db.get())->TEST_Current_Manifest_FileNo();
+  ASSERT_GT(manifest_after, manifest_before);
+
+  db.reset();
+  test_fs->ResetCounters();
+
+  // Reopen. Every live SST should be opened with file_open_metadata.
+  // Pre-fix: only SST #2 (added by the rotating edit) has metadata; SST #1
+  // was stripped by the rotation snapshot, so the count would be 1.
+  ASSERT_OK(DB::Open(options, test_dbname, &db));
+
+  std::string value;
+  ASSERT_OK(db->Get(ReadOptions(), "key1", &value));
+  ASSERT_EQ("value1", value);
+  ASSERT_OK(db->Get(ReadOptions(), "key2", &value));
+  ASSERT_EQ("value2", value);
+
+  ASSERT_EQ(2, test_fs->GetMetadataPassedOnOpenCount());
+
+  db.reset();
+  ASSERT_OK(DestroyDB(test_dbname, options));
+}
+
+TEST_F(DBTest2, FastSstOpenMetadataPreservedAcrossTrivialMove) {
+  // Regression test: FileMetaData::file_open_metadata must survive a trivial
+  // move. Trivial move re-records an SST at a new level via DeleteFile +
+  // AddFile without physically rewriting the file. If AddFile drops
+  // file_open_metadata, the file loses its fast_sst_open advantage until
+  // it is next rewritten by real compaction/flush/ingest.
+  auto test_fs = std::make_shared<FastOpenTestFS>(env_->GetFileSystem());
+  std::unique_ptr<Env> test_env(new CompositeEnvWrapper(env_, test_fs));
+
+  Options options;
+  options.env = test_env.get();
+  options.fast_sst_open = true;
+  options.create_if_missing = true;
+  options.disable_auto_compactions = true;
+
+  std::string test_dbname = test::PerThreadDBPath("fast_open_trivial_move");
+  ASSERT_OK(DestroyDB(test_dbname, options));
+
+  std::unique_ptr<DB> db;
+  ASSERT_OK(DB::Open(options, test_dbname, &db));
+
+  // Flush a single SST at L0; metadata is retrieved once and persisted.
+  ASSERT_OK(db->Put(WriteOptions(), "key1", "value1"));
+  ASSERT_OK(db->Flush(FlushOptions()));
+  ASSERT_EQ(1, test_fs->GetMetadataRetrievedCount());
+
+  // Manual compaction. With a single L0 file and empty lower levels, the
+  // compaction picker converts this into a trivial move (no physical
+  // rewrite), exercising PrepareTrivialMoveEdit.
+  ASSERT_OK(db->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+  // If the SST had been rewritten, metadata retrieval would have re-run on
+  // the new file. Unchanged count confirms this was a trivial move.
+  ASSERT_EQ(1, test_fs->GetMetadataRetrievedCount());
+
+  db.reset();
+  test_fs->ResetCounters();
+
+  // Reopen. The trivially-moved SST should still carry its
+  // file_open_metadata and be opened via the fast_sst_open fast path.
+  // Pre-fix, the trivial move stripped it and this count would be 0.
+  ASSERT_OK(DB::Open(options, test_dbname, &db));
+
+  std::string value;
+  ASSERT_OK(db->Get(ReadOptions(), "key1", &value));
+  ASSERT_EQ("value1", value);
+
+  ASSERT_EQ(1, test_fs->GetMetadataPassedOnOpenCount());
+
+  db.reset();
+  ASSERT_OK(DestroyDB(test_dbname, options));
+}
+
 }  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {

db/db_filesnapshot.cc

@@ -252,7 +252,7 @@ Status DBImpl::GetLiveFilesStorageInfo(
         "capturing live files. Retry with flush enabled.");
   }
   if (flush_memtable) {
-    wal_locked = lock_wal_count_ > 0;
+    wal_locked = !lock_wal_owner_thread_id_counts_.empty();
     if (wal_locked) {
       if (needs_blob_direct_write_flush) {
         mutex_.Unlock();
@@ -260,8 +260,12 @@ Status DBImpl::GetLiveFilesStorageInfo(
             "Blob direct write requires flushing active blob files before "
             "capturing live files. Retry with WAL unlocked.");
       }
-      ROCKS_LOG_INFO(immutable_db_options_.info_log,
-                     "Can't FlushForGetLiveFiles while WAL is locked");
+    }
+    if (wal_locked && !has_unpersisted_data_.load(std::memory_order_relaxed)) {
+      ROCKS_LOG_INFO(
+          immutable_db_options_.info_log,
+          "Skipping FlushForGetLiveFiles while WAL is locked and all data is "
+          "already persisted");
     } else {
       Status status = FlushForGetLiveFiles(opts.atomic_flush);
       if (!status.ok()) {

db/db_flush_test.cc

@@ -8,7 +8,9 @@
 // found in the LICENSE file. See the AUTHORS file for names of contributors.
 
 #include <atomic>
+#include <condition_variable>
 #include <limits>
+#include <mutex>
 
 #include "db/db_impl/db_impl.h"
 #include "db/db_test_util.h"
@@ -473,6 +475,61 @@ TEST_F(DBFlushTest, StatisticsGarbageBasic) {
   Close();
 }
 
+TEST_F(DBFlushTest, FlushReasonStatsWriteBufferFull) {
+  Options options = CurrentOptions();
+  options.statistics = CreateDBStatistics();
+  options.statistics->set_stats_level(StatsLevel::kAll);
+  options.create_if_missing = true;
+  options.compression = kNoCompression;
+  options.disable_auto_compactions = true;
+  options.avoid_flush_during_shutdown = true;
+  options.write_buffer_size = 16 * 1024;
+  options.max_write_buffer_number = 8;
+
+  auto flush_listener = std::make_shared<FlushCounterListener>();
+  flush_listener->expected_flush_reason = FlushReason::kWriteBufferFull;
+  options.listeners.push_back(flush_listener);
+
+  ASSERT_OK(TryReopen(options));
+
+  WriteOptions write_options;
+  write_options.disableWAL = true;
+  for (int i = 0; i < 20; ++i) {
+    ASSERT_OK(Put(Key(i), DummyString(10 * 1024), write_options));
+  }
+  ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable());
+  ASSERT_OK(dbfull()->TEST_WaitForBackgroundWork());
+
+  const uint64_t write_buffer_full_flushes =
+      TestGetTickerCount(options, FLUSH_REASON_WRITE_BUFFER_FULL);
+  ASSERT_GT(write_buffer_full_flushes, 0);
+  EXPECT_EQ(0, TestGetTickerCount(options, FLUSH_REASON_WRITE_BUFFER_MANAGER));
+
+  HistogramData all_memtable_memory;
+  options.statistics->histogramData(FLUSH_MEMTABLE_MEMORY_BYTES,
+                                    &all_memtable_memory);
+  EXPECT_GE(all_memtable_memory.count, write_buffer_full_flushes);
+  EXPECT_GT(all_memtable_memory.sum, 0);
+
+  HistogramData all_memtable_data_size;
+  options.statistics->histogramData(FLUSH_MEMTABLE_TOTAL_DATA_SIZE,
+                                    &all_memtable_data_size);
+  EXPECT_GE(all_memtable_data_size.count, write_buffer_full_flushes);
+  EXPECT_GT(all_memtable_data_size.sum, 0);
+
+  HistogramData write_buffer_full_memtable_memory;
+  options.statistics->histogramData(
+      FLUSH_WRITE_BUFFER_FULL_MEMTABLE_MEMORY_BYTES,
+      &write_buffer_full_memtable_memory);
+  EXPECT_EQ(write_buffer_full_flushes, write_buffer_full_memtable_memory.count);
+  EXPECT_GT(write_buffer_full_memtable_memory.sum, 0);
+
+  HistogramData wbm_memtable_memory;
+  options.statistics->histogramData(
+      FLUSH_WRITE_BUFFER_MANAGER_MEMTABLE_MEMORY_BYTES, &wbm_memtable_memory);
+  EXPECT_EQ(0, wbm_memtable_memory.count);
+}
+
 TEST_F(DBFlushTest, StatisticsGarbageInsertAndDeletes) {
   Options options = CurrentOptions();
   options.statistics = CreateDBStatistics();
@@ -2164,8 +2221,12 @@ TEST_F(DBFlushTest, FireOnFlushCompletedAfterCommittedResult) {
   listener->seq1 = db_->GetLatestSequenceNumber();
   // t1 will wait for the second flush complete before committing flush result.
   auto t1 = port::Thread([&]() {
-    // flush_opts.wait = true
-    ASSERT_OK(db_->Flush(FlushOptions()));
+    // flush_opts.wait = true, and listener_wait = true so that Flush() does not
+    // return until the OnFlushCompleted callbacks (which set completed1 and
+    // completed2) have finished running.
+    FlushOptions flush_opts;
+    flush_opts.listener_wait = true;
+    ASSERT_OK(db_->Flush(flush_opts));
   });
   // Wait for first flush started.
   TEST_SYNC_POINT(
@@ -2178,15 +2239,142 @@ TEST_F(DBFlushTest, FireOnFlushCompletedAfterCommittedResult) {
   flush_opts.wait = false;
   ASSERT_OK(db_->Flush(flush_opts));
   t1.join();
-  // Ensure background work is fully finished including listener callbacks
-  // before accessing listener state.
-  ASSERT_OK(dbfull()->TEST_WaitForBackgroundWork());
+  // listener_wait on t1's flush above guarantees both OnFlushCompleted
+  // callbacks have finished by the time Flush() returned; no need to
+  // additionally wait for background work here.
   ASSERT_TRUE(listener->completed1);
   ASSERT_TRUE(listener->completed2);
   SyncPoint::GetInstance()->DisableProcessing();
   SyncPoint::GetInstance()->ClearAllCallBacks();
 }
 
+namespace {
+// EventListener whose OnFlushCompleted blocks until explicitly released, used
+// to verify FlushOptions::listener_wait. Optionally only blocks on the first
+// callback (useful for atomic flush, which fires one callback per CF).
+class BlockingFlushListener : public EventListener {
+ public:
+  explicit BlockingFlushListener(bool block_only_first)
+      : block_only_first_(block_only_first) {}
+
+  void OnFlushCompleted(DB* /*db*/, const FlushJobInfo& /*info*/) override {
+    if (block_only_first_ && num_completed_.fetch_add(1) != 0) {
+      return;
+    }
+    std::unique_lock<std::mutex> lk(mu_);
+    entered_ = true;
+    cv_.notify_all();
+    cv_.wait(lk, [this] { return release_; });
+    returned_ = true;
+  }
+
+  void WaitUntilEntered() {
+    std::unique_lock<std::mutex> lk(mu_);
+    cv_.wait(lk, [this] { return entered_; });
+  }
+
+  void Release() {
+    {
+      std::lock_guard<std::mutex> lk(mu_);
+      release_ = true;
+    }
+    cv_.notify_all();
+  }
+
+  bool CallbackReturned() {
+    std::lock_guard<std::mutex> lk(mu_);
+    return returned_;
+  }
+
+ private:
+  const bool block_only_first_;
+  std::atomic<int> num_completed_{0};
+  std::mutex mu_;
+  std::condition_variable cv_;
+  bool entered_ = false;
+  bool release_ = false;
+  bool returned_ = false;
+};
+}  // anonymous namespace
+
+TEST_F(DBFlushTest, ListenerWaitWaitsForOnFlushCompleted) {
+  // With FlushOptions::listener_wait, Flush(wait=true) must not return until
+  // the OnFlushCompleted callback has finished, even if the flush result is
+  // already committed and bg_cv_ is signalled by unrelated background work.
+  auto listener = std::make_shared<BlockingFlushListener>(
+      /*block_only_first=*/false);
+  Options options = CurrentOptions();
+  options.listeners.push_back(listener);
+  Reopen(options);
+  ASSERT_OK(Put("k", "v"));
+
+  std::atomic<bool> flush_returned{false};
+  port::Thread flush_thread([&] {
+    // Exercises the TEST_ wrapper, which flushes with wait=true and
+    // listener_wait=true.
+    ASSERT_OK(dbfull()->TEST_FlushMemTableWithListenerWait());
+    flush_returned.store(true);
+  });
+
+  // Wait until we are inside OnFlushCompleted. At this point the memtable has
+  // already been committed/removed, which is what unblocks a plain
+  // Flush(wait=true).
+  listener->WaitUntilEntered();
+
+  // Pump spurious bg_cv_ wakeups. A waiter that ignored listener_wait would
+  // wake, observe the memtable already gone, and return here.
+  for (int i = 0; i < 10; ++i) {
+    db_->DisableManualCompaction();
+    db_->EnableManualCompaction();
+  }
+
+  // The OnFlushCompleted callback has not returned, so neither must Flush().
+  ASSERT_FALSE(listener->CallbackReturned());
+  ASSERT_FALSE(flush_returned.load());
+
+  listener->Release();
+  flush_thread.join();
+  ASSERT_TRUE(flush_returned.load());
+  ASSERT_TRUE(listener->CallbackReturned());
+}
+
+TEST_F(DBFlushTest, ListenerWaitAtomicFlushWaitsForAllOnFlushCompleted) {
+  // Same guarantee for atomic flush across multiple column families. The atomic
+  // flush removes all CFs' memtables together and then fires OnFlushCompleted
+  // per CF, releasing the DB mutex between CFs; listener_wait must still block
+  // Flush() until the (first, still-running) callback has finished.
+  auto listener = std::make_shared<BlockingFlushListener>(
+      /*block_only_first=*/true);
+  Options options = CurrentOptions();
+  options.atomic_flush = true;
+  options.listeners.push_back(listener);
+  CreateAndReopenWithCF({"pikachu"}, options);
+  ASSERT_OK(Put(0, "k0", "v0"));
+  ASSERT_OK(Put(1, "k1", "v1"));
+
+  std::atomic<bool> flush_returned{false};
+  port::Thread flush_thread([&] {
+    FlushOptions fo;
+    fo.wait = true;
+    fo.listener_wait = true;
+    ASSERT_OK(db_->Flush(fo, {handles_[0], handles_[1]}));
+    flush_returned.store(true);
+  });
+
+  listener->WaitUntilEntered();
+  for (int i = 0; i < 10; ++i) {
+    db_->DisableManualCompaction();
+    db_->EnableManualCompaction();
+  }
+  ASSERT_FALSE(listener->CallbackReturned());
+  ASSERT_FALSE(flush_returned.load());
+
+  listener->Release();
+  flush_thread.join();
+  ASSERT_TRUE(flush_returned.load());
+  ASSERT_TRUE(listener->CallbackReturned());
+}
+
 TEST_F(DBFlushTest, FlushWithBlob) {
   constexpr uint64_t min_blob_size = 10;
 
@@ -2747,6 +2935,8 @@ TEST_P(DBAtomicFlushTest, ManualAtomicFlush) {
   Options options = CurrentOptions();
   options.create_if_missing = true;
   options.atomic_flush = GetParam();
+  options.statistics = CreateDBStatistics();
+  options.statistics->set_stats_level(StatsLevel::kAll);
   options.write_buffer_size = (static_cast<size_t>(64) << 20);
   auto flush_listener = std::make_shared<FlushCounterListener>();
   flush_listener->expected_flush_reason = FlushReason::kManualFlush;
@@ -2773,6 +2963,16 @@ TEST_P(DBAtomicFlushTest, ManualAtomicFlush) {
   }
   ASSERT_OK(Flush(cf_ids));
 
+  EXPECT_EQ(options.atomic_flush ? 1 : 0,
+            TestGetTickerCount(options, ATOMIC_FLUSH_REQUEST_REASON_OTHER));
+  EXPECT_EQ(0, TestGetTickerCount(
+                   options, ATOMIC_FLUSH_REQUEST_REASON_WRITE_BUFFER_FULL));
+  EXPECT_EQ(0, TestGetTickerCount(
+                   options, ATOMIC_FLUSH_REQUEST_REASON_WRITE_BUFFER_MANAGER));
+  EXPECT_EQ(0, TestGetTickerCount(
+                   options,
+                   ATOMIC_FLUSH_REQUEST_REASON_MEMTABLE_MAX_RANGE_DELETIONS));
+
   for (size_t i = 0; i != num_cfs; ++i) {
     auto cfh = static_cast<ColumnFamilyHandleImpl*>(handles_[i]);
     ASSERT_EQ(0, cfh->cfd()->imm()->NumNotFlushed());