Merge pull request #3595 from facebook/dev

v1.5.5 last changes
Merge pull request #3593 from facebook/appvbadge
2023-04-04 13:13:52 -07:00 · 2023-04-03 18:25:06 -07:00 · 2023-04-03 17:46:37 -07:00 · 2023-04-03 17:40:39 -07:00 · 2023-04-03 16:52:32 -07:00 · 2023-04-03 14:38:22 -07:00
175 changed files with 6537 additions and 3679 deletions
@@ -1,4 +1,4 @@
-FROM circleci/buildpack-deps:focal
+FROM circleci/buildpack-deps@sha256:f6f10c11b7b8ccfd4f4a5b830c3256803604ce61292b60cb22e26b12f62b0e8c

 RUN sudo dpkg --add-architecture i386
 RUN sudo apt-get -y -qq update
@@ -9,11 +9,13 @@ on:
  pull_request:
    branches: [ dev, release, actionsTest ]

+permissions: read-all
+
 jobs:
  make-all:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: make all
      run: make all

@@ -24,7 +26,7 @@ jobs:
      DEVNULLRIGHTS: 1
      READFROMBLOCKDEVICE: 1
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: make test
      run: make test

@@ -32,28 +34,42 @@ jobs:
  make-test-osx:
    runs-on: macos-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: OS-X test
      run: make test # make -c lib all doesn't work because of the fact that it's not a tty

+  # lasts ~24mn
+  make-test-32bit:
+    runs-on: ubuntu-latest
+    env:
+      DEVNULLRIGHTS: 1
+      READFROMBLOCKDEVICE: 1
+    steps:
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
+    - name: make test
+      run: |
+        sudo apt-get -qqq update
+        make libc6install
+        CFLAGS="-m32" make test
+
  no-intrinsics-fuzztest:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: no intrinsics fuzztest
      run: MOREFLAGS="-DZSTD_NO_INTRINSICS" make -C tests fuzztest

  tsan-zstreamtest:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: thread sanitizer zstreamtest
      run: CC=clang ZSTREAM_TESTTIME=-T3mn make tsan-test-zstream

  ubsan-zstreamtest:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: undefined behavior sanitizer zstreamtest
      run: CC=clang make uasan-test-zstream

@@ -61,15 +77,26 @@ jobs:
  tsan-fuzztest:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: thread sanitizer fuzztest
      run: CC=clang make tsan-fuzztest

+
+  big-tests-zstreamtest32:
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
+    - name: zstream tests in 32bit mode, with big tests
+      run: |
+        sudo apt-get -qqq update
+        make libc6install
+        CC=clang make -C tests test-zstream32 FUZZER_FLAGS="--big-tests"
+
  # lasts ~23mn
  gcc-8-asan-ubsan-testzstd:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: gcc-8 + ASan + UBSan + Test Zstd
      # See https://askubuntu.com/a/1428822
      run: |
@@ -81,14 +108,14 @@ jobs:
  clang-asan-ubsan-testzstd:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: clang + ASan + UBSan + Test Zstd
      run: CC=clang make -j uasan-test-zstd </dev/null V=1

  gcc-asan-ubsan-testzstd-32bit:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: ASan + UBSan + Test Zstd, 32bit mode
      run: |
        sudo apt-get -qqq update
@@ -102,7 +129,7 @@ jobs:
  gcc-8-asan-ubsan-fuzz:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: gcc-8 + ASan + UBSan + Fuzz Test
      # See https://askubuntu.com/a/1428822
      run: |
@@ -114,14 +141,14 @@ jobs:
  clang-asan-ubsan-fuzz:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: clang + ASan + UBSan + Fuzz Test
      run: CC=clang FUZZER_FLAGS="--long-tests" make clean uasan-fuzztest

  gcc-asan-ubsan-fuzz32:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: ASan + UBSan + Fuzz Test 32bit
      run: |
        sudo apt-get -qqq update
@@ -131,7 +158,7 @@ jobs:
  clang-asan-ubsan-fuzz32:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: clang + ASan + UBSan + Fuzz Test 32bit
      run: |
        sudo apt-get -qqq update
@@ -141,28 +168,38 @@ jobs:
  asan-ubsan-regression:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: ASan + UBSan + Regression Test
      run: make -j uasanregressiontest

  clang-ubsan-regression:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: clang + ASan + UBSan + Regression Test
      run: CC=clang make -j uasanregressiontest

  msan-regression:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: MSan + Regression Test
      run: make -j msanregressiontest

+  clang-msan-fuzz-unoptimized:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
+      - name: clang + MSan + Fuzz Test
+        run: |
+          sudo apt-get -qqq update
+          sudo apt-get install clang
+          CC=clang MOREFLAGS="-O0" make clean msan-fuzztest
+
  clang-msan-fuzz:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: clang + MSan + Fuzz Test
      run: |
        sudo apt-get -qqq update
@@ -173,7 +210,7 @@ jobs:
  clang-msan-testzstd:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: clang + MSan + Test Zstd
      run: |
        sudo apt-get update
@@ -183,7 +220,7 @@ jobs:
  armfuzz:
      runs-on: ubuntu-latest
      steps:
-      - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+      - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
      - name: Qemu ARM emulation + Fuzz Test
        run: |
          sudo apt-get -qqq update
@@ -193,7 +230,7 @@ jobs:
  valgrind-fuzz-test:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: valgrind + fuzz test stack mode    # ~ 7mn
      shell: 'script -q -e -c "bash {0}"'
      run: |
@@ -209,8 +246,8 @@ jobs:
      run:
        shell: msys2 {0}
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
-    - uses: msys2/setup-msys2@v2
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
+    - uses: msys2/setup-msys2@5beef6d11f48bba68b9eb503e3adc60b23c0cc36 # tag=v2
      with:
        msystem: MINGW64
        install: make
@@ -253,7 +290,7 @@ jobs:
        dry-run: false
        sanitizer: ${{ matrix.sanitizer }}
    - name: Upload Crash
-      uses: actions/upload-artifact@83fd05a356d7e2593de66fc9913b3002723633cb # tag=v3.1.1
+      uses: actions/upload-artifact@0b7f8abb1508181956e8e162db84b466c27e18ce # tag=v3.1.2
      if: failure() && steps.build.outcome == 'success'
      with:
        name: ${{ matrix.sanitizer }}-artifacts
@@ -10,25 +10,27 @@ on:
  pull_request:
    branches: [ dev, release, actionsTest ]

+permissions: read-all
+
 jobs:
  linux-kernel:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: linux kernel, library + build + test
      run: make -C contrib/linux-kernel test CFLAGS="-Werror -Wunused-const-variable -Wunused-but-set-variable"

  benchmarking:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: make benchmarking
      run: make benchmarking

  check-32bit: # designed to catch https://github.com/facebook/zstd/issues/2428
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: make check on 32-bit
      run: |
        sudo apt update
@@ -38,7 +40,7 @@ jobs:
  check-x32:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: make check on x32 ABI # https://en.wikipedia.org/wiki/X32_ABI
      env:
        CHECK_CONSTRAINED_MEM: true
@@ -47,10 +49,27 @@ jobs:
        APT_PACKAGES="gcc-multilib" make apt-install
        CFLAGS="-mx32 -O1 -fstack-protector" make check V=1

+  build-c89:
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
+    - name: ensure zstd can be build with c89/c90 compilers (+ long long support + variadic macros)
+      run: |
+        make c89build V=1
+
+  build-zstd-dll:
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
+    - name: build zstd bin against a dynamic lib (debuglevel for more dependencies)
+      run: |
+        make -C lib lib-mt-release 
+        DEBUGLEVEL=2 make -C programs zstd-dll
+
  gcc-7-libzstd:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: gcc-7 + libzstdmt compilation
      # See https://askubuntu.com/a/1428822
      run: |
@@ -67,7 +86,7 @@ jobs:
  cmake-build-and-test-check:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: cmake build and test check
      run: |
        FUZZERTEST=-T1mn ZSTREAM_TESTTIME=-T1mn make cmakebuild
@@ -78,7 +97,7 @@ jobs:
  cpp-gnu90-c99-compatibility:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: C++, gnu90 and c99 compatibility
      run: |
        make cxxtest
@@ -92,7 +111,7 @@ jobs:
  mingw-cross-compilation:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: mingw cross-compilation
      run: |
        # sudo update-alternatives --set x86_64-w64-mingw32-g++ /usr/bin/x86_64-w64-mingw32-g++-posix; (doesn't work)
@@ -103,7 +122,7 @@ jobs:
  armbuild:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: ARM Build Test
      run: |
        sudo apt-get -qqq update
@@ -113,7 +132,7 @@ jobs:
  bourne-shell:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: Bourne shell compatibility (shellcheck)
      run: |
        wget https://github.com/koalaman/shellcheck/releases/download/v0.7.1/shellcheck-v0.7.1.linux.x86_64.tar.xz
@@ -123,7 +142,7 @@ jobs:
  zlib-wrapper:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: zlib wrapper test
      run: |
        sudo apt-get -qqq update
@@ -134,7 +153,7 @@ jobs:
  lz4-threadpool-libs:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: LZ4, thread pool, and libs build testslib wrapper test
      run: |
        make lz4install
@@ -148,7 +167,7 @@ jobs:
  gcc-make-tests-32bit:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: Make all, 32bit mode
      run: |
        sudo apt-get -qqq update
@@ -158,7 +177,7 @@ jobs:
  gcc-8-make:
    runs-on: ubuntu-latest
    steps:
-      - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+      - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
      - name: gcc-8 build
        # See https://askubuntu.com/a/1428822
        run: |
@@ -167,10 +186,29 @@ jobs:
          make gcc8install
          CC=gcc-8 CFLAGS="-Werror" make -j all

+  make-external-compressors:
+    strategy:
+      matrix:
+        include:
+          - name: "no external compressors"
+            flags: "HAVE_ZLIB=0 HAVE_LZ4=0 HAVE_LZMA=0"
+          - name: "only zlib"
+            flags: "HAVE_ZLIB=1 HAVE_LZ4=0 HAVE_LZMA=0"
+          - name: "only lz4"
+            flags: "HAVE_ZLIB=0 HAVE_LZ4=1 HAVE_LZMA=0"
+          - name: "only lzma"
+            flags: "HAVE_ZLIB=0 HAVE_LZ4=0 HAVE_LZMA=1"
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
+      - name: Build with ${{matrix.name}}
+        run: ${{matrix.flags}} make zstd
+
+
  implicit-fall-through:
    runs-on: ubuntu-latest
    steps:
-      - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+      - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
      - name: -Wimplicit-fallthrough build
        run: |
          make clean
@@ -181,7 +219,7 @@ jobs:
  meson-linux:
    runs-on: ubuntu-latest
    steps:
-      - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+      - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
      - name: Install packages
        run: |
          sudo apt-get update
@@ -205,11 +243,11 @@ jobs:
  meson-windows:
    runs-on: windows-latest
    steps:
-      - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+      - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
      - name: Install packages
        run: pip install --pre meson
      - name: Initialize the MSVC dev command prompt
-        uses: ilammy/msvc-dev-cmd@7315a94840631165970262a99c72cfb48a65d25d
+        uses: ilammy/msvc-dev-cmd@cec98b9d092141f74527d0afa6feb2af698cfe89
      - name: Configure with Meson
        run: |
          meson setup build/meson/ builddir -Dbin_tests=true -Dbin_programs=true -Dbin_contrib=true
@@ -232,11 +270,13 @@ jobs:
          - generator: "Visual Studio 17 2022"
            flags: "-A Win32"
          - generator: "MinGW Makefiles"
+          - generator: "Visual Studio 17 2022"
+            flags: "-T ClangCL"
    runs-on: windows-2022
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: Add MSBuild to PATH
-      uses: microsoft/setup-msbuild@v1.3
+      uses: microsoft/setup-msbuild@1ff57057b5cfdc39105cd07a01d78e9b0ea0c14c # tag=v1.3
    - name: Build
      working-directory: ${{env.GITHUB_WORKSPACE}}
      run: |
@@ -248,34 +288,45 @@ jobs:

  msbuild-visual-studio:
    strategy:
+      fail-fast: false  # 'false' means Don't stop matrix workflows even if some matrix failed.
      matrix:
        include: [
-          { name: "VS 2022 x64 Debug", platform: x64, configuration: Debug, toolset: v143, runner: "windows-2022"},
-          { name: "VS 2022 Win32 Debug", platform: Win32, configuration: Debug, toolset: v143, runner: "windows-2022"},
-          { name: "VS 2022 x64 Release", platform: x64, configuration: Release, toolset: v143, runner: "windows-2022"},
-          { name: "VS 2022 Win32 Release", platform: Win32, configuration: Release, toolset: v143, runner: "windows-2022"},
-          { name: "VS 2019 x64 Release", platform: Win32, configuration: Release, toolset: v142, runner: "windows-2019"},
-          { name: "VS 2019 Win32 Release", platform: x64, configuration: Release, toolset: v142, runner: "windows-2019"},
+          { name: "VS 2022 x64 Debug", platform: x64, configuration: Debug, toolset: v143, runner: "windows-2022", arch: "" },
+          { name: "VS 2022 Win32 Debug", platform: Win32, configuration: Debug, toolset: v143, runner: "windows-2022", arch: "" },
+          { name: "VS 2022 x64 Release", platform: x64, configuration: Release, toolset: v143, runner: "windows-2022", arch: ""},
+          { name: "VS 2022 Win32 Release", platform: Win32, configuration: Release, toolset: v143, runner: "windows-2022", arch: ""},
+          { name: "VS 2019 x64 Release", platform: Win32, configuration: Release, toolset: v142, runner: "windows-2019", arch: ""},
+          { name: "VS 2019 Win32 Release", platform: x64, configuration: Release, toolset: v142, runner: "windows-2019", arch: ""},
+          { name: "VS 2022 x64 Release AVX2", platform: x64, configuration: Release, toolset: v143, runner: "windows-2022", arch: "AdvancedVectorExtensions2" },
        ]
    runs-on: ${{matrix.runner}}
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: Add MSBuild to PATH
-      uses: microsoft/setup-msbuild@v1.3
-    - name: Build
+      uses: microsoft/setup-msbuild@1ff57057b5cfdc39105cd07a01d78e9b0ea0c14c # tag=v1.3
+    - name: Build ${{matrix.name}}
      working-directory: ${{env.GITHUB_WORKSPACE}}
      # See https://docs.microsoft.com/visualstudio/msbuild/msbuild-command-line-reference
+      if: ${{ matrix.arch == '' }}
      run: >
        msbuild "build\VS2010\zstd.sln" /m /verbosity:minimal /property:PlatformToolset=${{matrix.toolset}}
-        /t:Clean,Build /p:Platform=${{matrix.platform}} /p:Configuration=${{matrix.configuration}}
+        /t:Clean,Build /p:Platform=${{matrix.platform}} /p:Configuration=${{matrix.configuration}} /warnaserror
+    - name: Build ${{matrix.name}}
+      working-directory: ${{env.GITHUB_WORKSPACE}}
+      # See https://docs.microsoft.com/visualstudio/msbuild/msbuild-command-line-reference
+      if: ${{ matrix.arch != '' }}
+      run: >
+        msbuild "build\VS2010\zstd.sln" /m /verbosity:minimal /property:PlatformToolset=${{matrix.toolset}}
+        /t:Clean,Build /p:Platform=${{matrix.platform}} /p:Configuration=${{matrix.configuration}} /warnaserror
+        /p:InstructionSet=${{matrix.arch}}

-  # This tests that we don't accidently grow the size too much.
+  # This tests that we don't accidentally grow the size too much.
  # If the size grows intentionally, you can raise these numbers.
  # But we do need to think about binary size, since it is a concern.
  libzstd-size:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: libzstd size test
      run: |
        make clean && make -j -C lib libzstd && ./tests/check_size.py lib/libzstd.so 1100000
@@ -286,7 +337,7 @@ jobs:
  minimal-decompressor-macros:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: minimal decompressor macros
      run: |
        make clean && make -j all ZSTD_LIB_MINIFY=1 MOREFLAGS="-Werror"
@@ -301,7 +352,7 @@ jobs:
  dynamic-bmi2:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: dynamic bmi2 tests
      run: |
        make clean && make -j check MOREFLAGS="-O0 -Werror -mbmi2"
@@ -313,13 +364,12 @@ jobs:
  test-variants:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: make all variants & validate
      run: |
        make -j -C programs allVariants MOREFLAGS=-O0
        ./tests/test-variants.sh

-
  qemu-consistency:
    name: QEMU ${{ matrix.name }}
    runs-on: ubuntu-20.04
@@ -339,7 +389,7 @@ jobs:
      XCC: ${{ matrix.xcc }}
      XEMU: ${{ matrix.xemu }}
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: apt update & install
      run: |
        sudo apt-get update
@@ -384,6 +434,7 @@ jobs:
  mingw-short-test:
    runs-on: windows-latest
    strategy:
+      fail-fast: false  # 'false' means Don't stop matrix workflows even if some matrix failed.
      matrix:
        include: [
          { compiler: gcc, msystem: MINGW32, cflags: "-Werror"},
@@ -394,54 +445,34 @@ jobs:
      run:
        shell: msys2 {0}
    steps:
-<<<<<<< HEAD
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
-    - name: Mingw short test
-      run: |
-        ECHO "Building ${{matrix.compiler}} ${{matrix.platform}}"
-        $env:PATH_ORIGINAL = $env:PATH
-        $env:PATH_MINGW32 = "C:\msys64\mingw32\bin"
-        $env:PATH_MINGW64 = "C:\msys64\mingw64\bin"
-        COPY C:\msys64\usr\bin\make.exe C:\msys64\mingw32\bin\make.exe
-        COPY C:\msys64\usr\bin\make.exe C:\msys64\mingw64\bin\make.exe
-        IF ("${{matrix.platform}}" -eq "x64")
-        {
-            $env:PATH = $env:PATH_MINGW64 + ";" + $env:PATH_ORIGINAL
-        }
-        ELSEIF ("${{matrix.platform}}" -eq "x86")
-        {
-            $env:PATH = $env:PATH_MINGW32 + ";" + $env:PATH_ORIGINAL
-        }
-=======
-    - uses: actions/checkout@v2
-    - uses: msys2/setup-msys2@v2
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
+    - uses: msys2/setup-msys2@5beef6d11f48bba68b9eb503e3adc60b23c0cc36 # tag=v2
      with:
        msystem: ${{ matrix.msystem }}
        install: make diffutils
        update: true
    # Based on https://ariya.io/2020/07/on-github-actions-with-msys2
-    - name: install mingw gcc
+    - name: install mingw gcc i686
      if: ${{ (matrix.msystem == 'MINGW32') && (matrix.compiler == 'gcc') }}
      run: pacman --noconfirm -S mingw-w64-i686-gcc
-    - name: install mingw gcc
+    - name: install mingw gcc x86_64
      if: ${{ (matrix.msystem == 'MINGW64') && (matrix.compiler == 'gcc') }}
      run: pacman --noconfirm -S mingw-w64-x86_64-gcc
-    - name: install mingw clang x86_64
-      if: ${{ (matrix.msystem == 'MINGW64') && (matrix.compiler == 'clang') }}
-      run: pacman --noconfirm -S mingw-w64-x86_64-clang
    - name: install mingw clang i686
      if: ${{ (matrix.msystem == 'MINGW32') && (matrix.compiler == 'clang') }}
      run: pacman --noconfirm -S mingw-w64-i686-clang
+    - name: install mingw clang x86_64
+      if: ${{ (matrix.msystem == 'MINGW64') && (matrix.compiler == 'clang') }}
+      run: pacman --noconfirm -S mingw-w64-x86_64-clang
    - name: run mingw tests
      run: |
->>>>>>> e16c485f (Overhaul windows CI)
        make -v
        export CC=${{ matrix.compiler }}
        $CC --version
        CFLAGS="${{ matrix.cflags }}" make -j allzstd
        echo "Testing $CC ${{ matrix.msystem }}"
        make clean
-        make check
+        MSYS="" make check

  visual-runtime-tests:
    runs-on: windows-latest
@@ -450,9 +481,9 @@ jobs:
        platform: [x64, Win32]
        configuration: [Release]
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: Add MSBuild to PATH
-      uses: microsoft/setup-msbuild@v1.3
+      uses: microsoft/setup-msbuild@1ff57057b5cfdc39105cd07a01d78e9b0ea0c14c # tag=v1.3
    - name: Build and run tests
      working-directory: ${{env.GITHUB_WORKSPACE}}
      env:
@@ -464,13 +495,39 @@ jobs:
        COPY build\VS2010\bin\${{matrix.platform}}_${{matrix.configuration}}\*.exe tests\
        CD tests
        sh -e playTests.sh
-        DIR
        .\fuzzer.exe -T2m

+  # Following instructions at: https://github.com/marketplace/actions/install-cygwin-action
+  cygwin-tests:
+    runs-on: windows-latest
+    steps:
+    - run: git config --global core.autocrlf input
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
+    - uses: cygwin/cygwin-install-action@f5e0f048310c425e84bc789f493a828c6dc80a25 # tag=master
+      with:
+        platform: x86_64
+        packages: >-
+          autoconf,
+          automake,
+          gcc-g++,
+          make,
+          mingw64-x86_64-gcc-g++,
+          patch,
+          perl
+    - name: cygwin tests
+      shell: C:\cygwin\bin\bash.exe --noprofile --norc -eo pipefail '{0}'
+      run: >-
+        export PATH=/usr/bin:$(cygpath ${SYSTEMROOT})/system32 &&
+        export CFLAGS="-Werror -O1" &&
+        ls &&
+        make -j allzstd &&
+        make -C tests fuzzer &&
+        ./tests/fuzzer.exe -v -T1m
+
  intel-cet-compatibility:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: Build Zstd
      run: |
        make -j zstd V=1
@@ -491,7 +548,7 @@ jobs:
    container:
      image: debian:testing
    steps:
-      - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+      - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
      - name: Install dependencies
        run: |
          apt -y update
@@ -506,7 +563,7 @@ jobs:
  versions-compatibility:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: Versions Compatibility Test
      run: |
        make -C tests versionsTest
@@ -514,11 +571,34 @@ jobs:
  clangbuild:
    runs-on: ubuntu-latest
    steps:
-    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
    - name: make clangbuild
      run: |
        make clangbuild

+  clang-pgo:
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
+    - name: Build PGO Zstd with Clang
+      env:
+        CC: clang-14
+        LLVM_PROFDATA: llvm-profdata-14
+      run: |
+        make -C programs zstd-pgo
+        ./programs/zstd -b
+
+  gcc-pgo:
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
+    - name: Build PGO Zstd with GCC
+      env:
+        CC: gcc
+      run: |
+        make -C programs zstd-pgo
+        ./programs/zstd -b
+

 # For reference : icc tests
 # icc tests are currently failing on Github Actions, likely to issues during installation stage
@@ -538,7 +618,7 @@ jobs:
 #        sudo add-apt-repository "deb https://apt.repos.intel.com/oneapi all main"
 #        sudo apt-get update
 #        sudo apt-get install -y intel-basekit intel-hpckit
-#    - uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+#    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
 #    - name: make check
 #      run: |
 #        make CC=/opt/intel/oneapi/compiler/latest/linux/bin/intel64/icc check
@@ -5,21 +5,19 @@ on:
    types:
      - published

-permissions:
-  contents: read
+permissions: read-all

 jobs:
  publish-release-artifacts:
    permissions:
-      contents: read # to fetch code (actions/checkout)
-      actions: write # to attach binaries to release artifacts (skx/github-action-publish-binaries)
+      contents: write # to fetch code and upload artifacts

    runs-on: ubuntu-latest
    if: startsWith(github.ref, 'refs/tags/')

    steps:
      - name: Checkout
-        uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+        uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3

      - name: Archive
        env:
@@ -68,7 +66,7 @@ jobs:
          fi

      - name: Publish
-        uses: skx/github-action-publish-binaries@release-2.0
+        uses: skx/github-action-publish-binaries@b9ca5643b2f1d7371a6cba7f35333f1461bbc703 # tag=release-2.0
        env:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
        with:
@@ -27,12 +27,12 @@ jobs:

    steps:
      - name: "Checkout code"
-        uses: actions/checkout@ac593985615ec2ede58e132d2e21d2b1cbd6127c # tag=v3
+        uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
        with:
          persist-credentials: false

      - name: "Run analysis"
-        uses: ossf/scorecard-action@937ffa90d79c7d720498178154ad4c7ba1e4ad8c # tag=v2.1.0
+        uses: ossf/scorecard-action@e38b1902ae4f44df626f11ba0734b14fb91f8f86 # tag=v2.1.2
        with:
          results_file: results.sarif
          results_format: sarif
@@ -51,7 +51,7 @@ jobs:
      # Upload the results as artifacts (optional). Commenting out will disable uploads of run results in SARIF
      # format to the repository Actions tab.
      - name: "Upload artifact"
-        uses: actions/upload-artifact@83fd05a356d7e2593de66fc9913b3002723633cb # tag=v3.1.1
+        uses: actions/upload-artifact@0b7f8abb1508181956e8e162db84b466c27e18ce # tag=v3.1.2
        with:
          name: SARIF file
          path: results.sarif
@@ -59,6 +59,6 @@ jobs:

      # Upload the results to GitHub's code scanning dashboard.
      - name: "Upload to code-scanning"
-        uses: github/codeql-action/upload-sarif@959cbb7472c4d4ad70cdfe6f4976053fe48ab394 # tag=v2.1.37
+        uses: github/codeql-action/upload-sarif@67a35a08586135a9573f4327e904ecbf517a882d # tag=v2.2.8
        with:
          sarif_file: results.sarif
@@ -0,0 +1,51 @@
+name: windows-artifacts
+
+on:
+  push:
+    branches: [ test_artifacts, win_artifacts ]
+  release:
+    types:
+      - published
+
+permissions: read-all
+
+jobs:
+  windows-64-artifacts:
+    # see https://ariya.io/2020/07/on-github-actions-with-msys2
+    runs-on: windows-latest
+    defaults:
+      run:
+        shell: msys2 {0}
+    steps:
+    - uses: actions/checkout@8f4b7f84864484a7bf31766abe9204da3cbe65b3 # tag=v3
+    - uses: msys2/setup-msys2@5beef6d11f48bba68b9eb503e3adc60b23c0cc36 # tag=v2
+      with:
+        msystem: MINGW64
+        install: make zlib git p7zip mingw-w64-x86_64-gcc
+        update: true
+    - name: display versions
+      run: |
+        make -v
+        cc -v
+
+    - name: Building zlib to static link
+      run: |
+        git clone --depth 1 --branch v1.2.11 https://github.com/madler/zlib
+        make -C zlib -f win32/Makefile.gcc libz.a
+
+    - name: Building zstd programs in 64-bit mode
+      run: |
+        CPPFLAGS=-I../zlib LDFLAGS=../zlib/libz.a make -j allzstd MOREFLAGS=-static V=1
+
+    - name: Create artifacts
+      run: |
+        ./lib/dll/example/build_package.bat
+        mv bin/ zstd-${{ github.ref_name }}-win64/
+        7z a -tzip -mx9 zstd-${{ github.ref_name }}-win64.zip zstd-${{ github.ref_name }}-win64/
+        cd ..
+
+    - name: Publish zstd-$VERSION-win64.zip
+      uses: actions/upload-artifact@0b7f8abb1508181956e8e162db84b466c27e18ce # tag=v3
+      with:
+        path: ${{ github.workspace }}/zstd-${{ github.ref_name }}-win64.zip
+        name: zstd-${{ github.ref_name }}-win64.zip
@@ -1,37 +1,81 @@
-v1.5.3 (June, 2022)
-perf: 5-30% faster dictionary compression at levels 1-4 (#3086, #3114, #3152, @embg)
-perf: 5-10% faster streaming compression at levels 1-2 (#3114, @embg)
-perf: Remove branch in ZSTD_fast_noDict (#3129, @felixhandte)
-perf: Add option to prefetch CDict tables (#3177, @embg)
-perf: Minor compression ratio improvement (#2983, @Cyan4973)
-perf: Minor speed improvement for Huffman decoding (#3013, @WojciechMula)
-perf: Enable STATIC_BMI2 for gcc/clang (#3080, @TocarIP)
-perf: Optimize ZSTD_row_getMatchMask for ARM levels 8-10 (#3139, #3160, @danlark1)
-perf: aarch64 performance improvements (#3145, #3141, @JunHe77)
-perf: Lazy parameters adaptation (#2974, @Cyan4973)
-cli: Async write for decompression (#2975, @yoniko)
-cli: Use buffered output (#2985, @yoniko)
-cli: Change zstdless behavior to align with zless (#2909, @binhdvo)
-cli: AsyncIO compression (#3021, #3022, @yoniko)
-cli: Print zlib/lz4/lzma library versions in verbose version output (#3030, @terrelln)
-cli: Fix for -r on empty directory (#3027, @brailovich)
-cli: Fix required decompression memory usage reported by -vv + --long (#3042, @u1f35c)
+v1.5.5 (Apr 2023)
+fix: fix rare corruption bug affecting the high compression mode, reported by @danlark1 (#3517, @terrelln)
+perf: improve mid-level compression speed (#3529, #3533, #3543, @yoniko and #3552, @terrelln)
+lib: deprecated bufferless block-level API (#3534) by @terrelln
+cli: mmap large dictionaries to save memory, by @daniellerozenblit
+cli: improve speed of --patch-from mode (~+50%) (#3545) by @daniellerozenblit
+cli: improve i/o speed (~+10%) when processing lots of small files (#3479) by @felixhandte
+cli: zstd no longer crashes when requested to write into write-protected directory (#3541) by @felixhandte
+cli: fix decompression into block device using -o, reported by @georgmu (#3583)
+build: fix zstd CLI compiled with lzma support but not zlib support (#3494) by @Hello71
+build: fix cmake does no longer require 3.18 as minimum version (#3510) by @kou
+build: fix MSVC+ClangCL linking issue (#3569) by @tru
+build: fix zstd-dll, version of zstd CLI that links to the dynamic library (#3496) by @yoniko
+build: fix MSVC warnings (#3495) by @embg
+doc: updated zstd specification to clarify corner cases, by @Cyan4973
+doc: document how to create fat binaries for macos (#3568) by @rickmark
+misc: improve seekable format ingestion speed (~+100%) for very small chunk sizes (#3544) by @Cyan4973
+misc: tests/fullbench can benchmark multiple files (#3516) by @dloidolt
+
+v1.5.4 (Feb 2023)
+perf: +20% faster huffman decompression for targets that can't compile x64 assembly (#3449, @terrelln)
+perf: up to +10% faster streaming compression at levels 1-2 (#3114, @embg)
+perf: +4-13% for levels 5-12 by optimizing function generation (#3295, @terrelln)
+pref: +3-11% compression speed for `arm` target (#3199, #3164, #3145, #3141, #3138, @JunHe77 and #3139, #3160, @danlark1)
+perf: +5-30% faster dictionary compression at levels 1-4 (#3086, #3114, #3152, @embg)
+perf: +10-20% cold dict compression speed by prefetching CDict tables (#3177, @embg)
+perf: +1% faster compression by removing a branch in ZSTD_fast_noDict (#3129, @felixhandte)
+perf: Small compression ratio improvements in high compression mode (#2983, #3391, @Cyan4973 and #3285, #3302, @daniellerozenblit)
+perf: small speed improvement by better detecting `STATIC_BMI2` for `clang` (#3080, @TocarIP)
+perf: Improved streaming performance when `ZSTD_c_stableInBuffer` is set (#2974, @Cyan4973)
+cli: Asynchronous I/O for improved cli speed (#2975, #2985, #3021, #3022, @yoniko)
+cli: Change `zstdless` behavior to align with `zless` (#2909, @binhdvo)
+cli: Keep original file if `-c` or `--stdout` is given (#3052, @dirkmueller)
+cli: Keep original files when result is concatenated into a single output with `-o` (#3450, @Cyan4973)
+cli: Preserve Permissions and Ownership of regular files (#3432, @felixhandte)
+cli: Print zlib/lz4/lzma library versions with `-vv` (#3030, @terrelln)
+cli: Print checksum value for single frame files with `-lv`  (#3332, @Cyan4973)
+cli: Print `dictID` when present with `-lv` (#3184, @htnhan)
+cli: when `stderr` is *not* the console, disable status updates, but preserve final summary (#3458, @Cyan4973)
+cli: support `--best` and `--no-name` in `gzip` compatibility mode (#3059, @dirkmueller)
+cli: support for `posix` high resolution timer `clock_gettime()`, for improved benchmark accuracy (#3423, @Cyan4973)
+cli: improved help/usage (`-h`,  `-H`) formatting (#3094, @dirkmueller and #3385, @jonpalmisc)
+cli: Fix better handling of bogus numeric values (#3268, @ctkhanhly)
+cli: Fix input consists of multiple files _and_ `stdin` (#3222, @yoniko)
+cli: Fix tiny files passthrough (#3215, @cgbur)
+cli: Fix for `-r` on empty directory (#3027, @brailovich)
+cli: Fix empty string as argument for `--output-dir-*` (#3220, @embg)
+cli: Fix decompression memory usage reported by `-vv --long` (#3042, @u1f35c, and #3232, @zengyijing)
 cli: Fix infinite loop when empty input is passed to trainer (#3081, @terrelln)
-cli: Implement more gzip compatibility (#3059, @dirkmueller)
-cli: Keep original file if -c or --stdout is given (#3052, @dirkmueller)
+cli: Fix `--adapt` doesn't work when `--no-progress` is also set (#3354, @terrelln)
+api: Support for Block-Level Sequence Producer (#3333, @embg)
+api: Support for in-place decompression (#3432, @terrelln)
+api: New  `ZSTD_CCtx_setCParams()`  function, set all parameters defined in a  `ZSTD_compressionParameters`  structure (#3403, @Cyan4973)
+api: Streaming decompression detects incorrect header ID sooner (#3175, @Cyan4973)
+api: Window size resizing optimization for edge case (#3345, @daniellerozenblit)
+api: More accurate error codes for busy-loop scenarios (#3413, #3455, @Cyan4973)
+api: Fix limit overflow in `compressBound` and `decompressBound` (#3362, #3373, Cyan4973) reported by @nigeltao
+api: Deprecate several advanced experimental functions: streaming (#3408, @embg), copy (#3196, @mileshu)
+bug: Fix corruption that rarely occurs in 32-bit mode with wlog=25 (#3361, @terrelln)
 bug: Fix for block-splitter (#3033, @Cyan4973)
 bug: Fixes for Sequence Compression API (#3023, #3040, @Cyan4973)
 bug: Fix leaking thread handles on Windows (#3147, @animalize)
 bug: Fix timing issues with cmake/meson builds (#3166, #3167, #3170, @Cyan4973)
 build: Allow user to select legacy level for cmake (#3050, @shadchin)
 build: Enable legacy support by default in cmake (#3079, @niamster)
-build: Meson improvements (#3039, #3122, @eli-schwartz)
+build: Meson build script improvements (#3039, #3120, #3122, #3327, #3357, @eli-schwartz and #3276, @neheb)
 build: Add aarch64 to supported architectures for zstd_trace (#3054, @ooosssososos)
-doc: Split help in long and short version, cleanup formatting (#3094, @dirkmueller)
-doc: Updated man page, providing more details for --train mode (#3112, @Cyan4973)
+build: support AIX architecture (#3219, @qiongsiwu)
+build: Fix `ZSTD_LIB_MINIFY` build macro, which now reduces static library size by half (#3366, @terrelln)
+build: Fix Windows issues with Multithreading translation layer (#3364, #3380, @yoniko) and ARM64 target (#3320, @cwoffenden)
+build: Fix `cmake` script (#3382, #3392, @terrelln and #3252 @Tachi107 and #3167 @Cyan4973)
+doc: Updated man page, providing more details for `--train` mode (#3112, @Cyan4973)
 doc: Add decompressor errata document (#3092, @terrelln)
 misc: Enable Intel CET (#2992, #2994, @hjl-tools)
-misc: Streaming decompression can detect incorrect header ID sooner (#3175, @Cyan4973)
+misc: Fix `contrib/` seekable format (#3058, @yhoogstrate and #3346, @daniellerozenblit)
+misc: Improve speed of the one-file library generator (#3241, @wahern and #3005, @cwoffenden)
+
+v1.5.3 (dev version, unpublished)

 v1.5.2 (Jan, 2022)
 perf: Regain Minimal memset()-ing During Reuse of Compression Contexts (@Cyan4973, #2969)
@@ -123,7 +123,7 @@ contrib: lib
 	$(MAKE) -C contrib/seekable_format/examples all
 	$(MAKE) -C contrib/seekable_format/tests test
 	$(MAKE) -C contrib/largeNbDicts all
-	$(MAKE) -C contrib/externalMatchfinder all
+	$(MAKE) -C contrib/externalSequenceProducer all
 	cd build/single_file_libs/ ; ./build_decoder_test.sh
 	cd build/single_file_libs/ ; ./build_library_test.sh

@@ -143,7 +143,7 @@ clean:
 	$(Q)$(MAKE) -C contrib/seekable_format/examples $@ > $(VOID)
 	$(Q)$(MAKE) -C contrib/seekable_format/tests $@ > $(VOID)
 	$(Q)$(MAKE) -C contrib/largeNbDicts $@ > $(VOID)
-	$(Q)$(MAKE) -C contrib/externalMatchfinder $@ > $(VOID)
+	$(Q)$(MAKE) -C contrib/externalSequenceProducer $@ > $(VOID)
 	$(Q)$(RM) zstd$(EXT) zstdmt$(EXT) tmp*
 	$(Q)$(RM) -r lz4
 	@echo Cleaning completed
@@ -200,15 +200,15 @@ travis-install:
 .PHONY: gcc5build gcc6build gcc7build clangbuild m32build armbuild aarch64build ppcbuild ppc64build
 gcc5build: clean
 	gcc-5 -v
-	CC=gcc-5 $(MAKE) all MOREFLAGS="-Werror"
+	CC=gcc-5 $(MAKE) all MOREFLAGS="-Werror $(MOREFLAGS)"

 gcc6build: clean
 	gcc-6 -v
-	CC=gcc-6 $(MAKE) all MOREFLAGS="-Werror"
+	CC=gcc-6 $(MAKE) all MOREFLAGS="-Werror $(MOREFLAGS)"

 gcc7build: clean
 	gcc-7 -v
-	CC=gcc-7 $(MAKE) all MOREFLAGS="-Werror"
+	CC=gcc-7 $(MAKE) all MOREFLAGS="-Werror $(MOREFLAGS)"

 clangbuild: clean
 	clang -v
@@ -232,17 +232,17 @@ ppc64build: clean

 .PHONY: armfuzz aarch64fuzz ppcfuzz ppc64fuzz
 armfuzz: clean
-	CC=arm-linux-gnueabi-gcc QEMU_SYS=qemu-arm-static MOREFLAGS="-static" FUZZER_FLAGS=--no-big-tests $(MAKE) -C $(TESTDIR) fuzztest
+	CC=arm-linux-gnueabi-gcc QEMU_SYS=qemu-arm-static MOREFLAGS="-static $(MOREFLAGS)" FUZZER_FLAGS="--no-big-tests $(FUZZER_FLAGS)" $(MAKE) -C $(TESTDIR) fuzztest

 aarch64fuzz: clean
 	ld -v
-	CC=aarch64-linux-gnu-gcc QEMU_SYS=qemu-aarch64-static MOREFLAGS="-static" FUZZER_FLAGS=--no-big-tests $(MAKE) -C $(TESTDIR) fuzztest
+	CC=aarch64-linux-gnu-gcc QEMU_SYS=qemu-aarch64-static MOREFLAGS="-static $(MOREFLAGS)" FUZZER_FLAGS="--no-big-tests $(FUZZER_FLAGS)" $(MAKE) -C $(TESTDIR) fuzztest

 ppcfuzz: clean
-	CC=powerpc-linux-gnu-gcc QEMU_SYS=qemu-ppc-static MOREFLAGS="-static" FUZZER_FLAGS=--no-big-tests $(MAKE) -C $(TESTDIR) fuzztest
+	CC=powerpc-linux-gnu-gcc QEMU_SYS=qemu-ppc-static MOREFLAGS="-static $(MOREFLAGS)" FUZZER_FLAGS="--no-big-tests $(FUZZER_FLAGS)" $(MAKE) -C $(TESTDIR) fuzztest

 ppc64fuzz: clean
-	CC=powerpc-linux-gnu-gcc QEMU_SYS=qemu-ppc64-static MOREFLAGS="-m64 -static" FUZZER_FLAGS=--no-big-tests $(MAKE) -C $(TESTDIR) fuzztest
+	CC=powerpc-linux-gnu-gcc QEMU_SYS=qemu-ppc64-static MOREFLAGS="-m64 -static $(MOREFLAGS)" FUZZER_FLAGS="--no-big-tests $(FUZZER_FLAGS)" $(MAKE) -C $(TESTDIR) fuzztest

 .PHONY: cxxtest gcc5test gcc6test armtest aarch64test ppctest ppc64test
 cxxtest: CXXFLAGS += -Wall -Wextra -Wundef -Wshadow -Wcast-align -Werror
@@ -251,34 +251,34 @@ cxxtest: clean

 gcc5test: clean
 	gcc-5 -v
-	$(MAKE) all CC=gcc-5 MOREFLAGS="-Werror"
+	$(MAKE) all CC=gcc-5 MOREFLAGS="-Werror $(MOREFLAGS)"

 gcc6test: clean
 	gcc-6 -v
-	$(MAKE) all CC=gcc-6 MOREFLAGS="-Werror"
+	$(MAKE) all CC=gcc-6 MOREFLAGS="-Werror $(MOREFLAGS)"

 armtest: clean
 	$(MAKE) -C $(TESTDIR) datagen   # use native, faster
-	$(MAKE) -C $(TESTDIR) test CC=arm-linux-gnueabi-gcc QEMU_SYS=qemu-arm-static ZSTDRTTEST= MOREFLAGS="-Werror -static" FUZZER_FLAGS=--no-big-tests
+	$(MAKE) -C $(TESTDIR) test CC=arm-linux-gnueabi-gcc QEMU_SYS=qemu-arm-static ZSTDRTTEST= MOREFLAGS="-Werror -static $(MOREFLAGS)" FUZZER_FLAGS="--no-big-tests $(FUZZER_FLAGS)"

 aarch64test:
 	$(MAKE) -C $(TESTDIR) datagen   # use native, faster
-	$(MAKE) -C $(TESTDIR) test CC=aarch64-linux-gnu-gcc QEMU_SYS=qemu-aarch64-static ZSTDRTTEST= MOREFLAGS="-Werror -static" FUZZER_FLAGS=--no-big-tests
+	$(MAKE) -C $(TESTDIR) test CC=aarch64-linux-gnu-gcc QEMU_SYS=qemu-aarch64-static ZSTDRTTEST= MOREFLAGS="-Werror -static $(MOREFLAGS)" FUZZER_FLAGS="--no-big-tests $(FUZZER_FLAGS)"

 ppctest: clean
 	$(MAKE) -C $(TESTDIR) datagen   # use native, faster
-	$(MAKE) -C $(TESTDIR) test CC=powerpc-linux-gnu-gcc QEMU_SYS=qemu-ppc-static ZSTDRTTEST= MOREFLAGS="-Werror -Wno-attributes -static" FUZZER_FLAGS=--no-big-tests
+	$(MAKE) -C $(TESTDIR) test CC=powerpc-linux-gnu-gcc QEMU_SYS=qemu-ppc-static ZSTDRTTEST= MOREFLAGS="-Werror -Wno-attributes -static $(MOREFLAGS)" FUZZER_FLAGS="--no-big-tests $(FUZZER_FLAGS)"

 ppc64test: clean
 	$(MAKE) -C $(TESTDIR) datagen   # use native, faster
-	$(MAKE) -C $(TESTDIR) test CC=powerpc-linux-gnu-gcc QEMU_SYS=qemu-ppc64-static ZSTDRTTEST= MOREFLAGS="-m64 -static" FUZZER_FLAGS=--no-big-tests
+	$(MAKE) -C $(TESTDIR) test CC=powerpc-linux-gnu-gcc QEMU_SYS=qemu-ppc64-static ZSTDRTTEST= MOREFLAGS="-m64 -static $(MOREFLAGS)" FUZZER_FLAGS="--no-big-tests $(FUZZER_FLAGS)"

 .PHONY: arm-ppc-compilation
 arm-ppc-compilation:
-	$(MAKE) -C $(PRGDIR) clean zstd CC=arm-linux-gnueabi-gcc QEMU_SYS=qemu-arm-static ZSTDRTTEST= MOREFLAGS="-Werror -static"
-	$(MAKE) -C $(PRGDIR) clean zstd CC=aarch64-linux-gnu-gcc QEMU_SYS=qemu-aarch64-static ZSTDRTTEST= MOREFLAGS="-Werror -static"
-	$(MAKE) -C $(PRGDIR) clean zstd CC=powerpc-linux-gnu-gcc QEMU_SYS=qemu-ppc-static ZSTDRTTEST= MOREFLAGS="-Werror -Wno-attributes -static"
-	$(MAKE) -C $(PRGDIR) clean zstd CC=powerpc-linux-gnu-gcc QEMU_SYS=qemu-ppc64-static ZSTDRTTEST= MOREFLAGS="-m64 -static"
+	$(MAKE) -C $(PRGDIR) clean zstd CC=arm-linux-gnueabi-gcc QEMU_SYS=qemu-arm-static ZSTDRTTEST= MOREFLAGS="-Werror -static $(MOREFLAGS)"
+	$(MAKE) -C $(PRGDIR) clean zstd CC=aarch64-linux-gnu-gcc QEMU_SYS=qemu-aarch64-static ZSTDRTTEST= MOREFLAGS="-Werror -static $(MOREFLAGS)"
+	$(MAKE) -C $(PRGDIR) clean zstd CC=powerpc-linux-gnu-gcc QEMU_SYS=qemu-ppc-static ZSTDRTTEST= MOREFLAGS="-Werror -Wno-attributes -static $(MOREFLAGS)"
+	$(MAKE) -C $(PRGDIR) clean zstd CC=powerpc-linux-gnu-gcc QEMU_SYS=qemu-ppc64-static ZSTDRTTEST= MOREFLAGS="-m64 -static $(MOREFLAGS)"

 regressiontest:
 	$(MAKE) -C $(FUZZDIR) regressiontest
@@ -308,31 +308,31 @@ update_regressionResults:
 # run UBsan with -fsanitize-recover=pointer-overflow
 # this only works with recent compilers such as gcc 8+
 usan: clean
-	$(MAKE) test CC=clang MOREFLAGS="-g -fno-sanitize-recover=all -fsanitize-recover=pointer-overflow -fsanitize=undefined -Werror"
+	$(MAKE) test CC=clang MOREFLAGS="-g -fno-sanitize-recover=all -fsanitize-recover=pointer-overflow -fsanitize=undefined -Werror $(MOREFLAGS)"

 asan: clean
-	$(MAKE) test CC=clang MOREFLAGS="-g -fsanitize=address -Werror"
+	$(MAKE) test CC=clang MOREFLAGS="-g -fsanitize=address -Werror $(MOREFLAGS)"

 asan-%: clean
-	LDFLAGS=-fuse-ld=gold MOREFLAGS="-g -fno-sanitize-recover=all -fsanitize=address -Werror" $(MAKE) -C $(TESTDIR) $*
+	LDFLAGS=-fuse-ld=gold MOREFLAGS="-g -fno-sanitize-recover=all -fsanitize=address -Werror $(MOREFLAGS)" $(MAKE) -C $(TESTDIR) $*

 msan: clean
-	$(MAKE) test CC=clang MOREFLAGS="-g -fsanitize=memory -fno-omit-frame-pointer -Werror" HAVE_LZMA=0   # datagen.c fails this test for no obvious reason
+	$(MAKE) test CC=clang MOREFLAGS="-g -fsanitize=memory -fno-omit-frame-pointer -Werror $(MOREFLAGS)" HAVE_LZMA=0   # datagen.c fails this test for no obvious reason

 msan-%: clean
-	LDFLAGS=-fuse-ld=gold MOREFLAGS="-g -fno-sanitize-recover=all -fsanitize=memory -fno-omit-frame-pointer -Werror" FUZZER_FLAGS=--no-big-tests $(MAKE) -C $(TESTDIR) HAVE_LZMA=0 $*
+	LDFLAGS=-fuse-ld=gold MOREFLAGS="-g -fno-sanitize-recover=all -fsanitize=memory -fno-omit-frame-pointer -Werror $(MOREFLAGS)" FUZZER_FLAGS="--no-big-tests $(FUZZER_FLAGS)" $(MAKE) -C $(TESTDIR) HAVE_LZMA=0 $*

 asan32: clean
-	$(MAKE) -C $(TESTDIR) test32 CC=clang MOREFLAGS="-g -fsanitize=address"
+	$(MAKE) -C $(TESTDIR) test32 CC=clang MOREFLAGS="-g -fsanitize=address $(MOREFLAGS)"

 uasan: clean
-	$(MAKE) test CC=clang MOREFLAGS="-g -fno-sanitize-recover=all -fsanitize-recover=pointer-overflow -fsanitize=address,undefined -Werror"
+	$(MAKE) test CC=clang MOREFLAGS="-g -fno-sanitize-recover=all -fsanitize-recover=pointer-overflow -fsanitize=address,undefined -Werror $(MOREFLAGS)"

 uasan-%: clean
-	LDFLAGS=-fuse-ld=gold MOREFLAGS="-g -fno-sanitize-recover=all -fsanitize-recover=pointer-overflow -fsanitize=address,undefined -Werror" $(MAKE) -C $(TESTDIR) $*
+	LDFLAGS=-fuse-ld=gold MOREFLAGS="-g -fno-sanitize-recover=all -fsanitize-recover=pointer-overflow -fsanitize=address,undefined -Werror $(MOREFLAGS)" $(MAKE) -C $(TESTDIR) $*

 tsan-%: clean
-	LDFLAGS=-fuse-ld=gold MOREFLAGS="-g -fno-sanitize-recover=all -fsanitize=thread -Werror" $(MAKE) -C $(TESTDIR) $* FUZZER_FLAGS=--no-big-tests
+	LDFLAGS=-fuse-ld=gold MOREFLAGS="-g -fno-sanitize-recover=all -fsanitize=thread -Werror $(MOREFLAGS)" $(MAKE) -C $(TESTDIR) $* FUZZER_FLAGS="--no-big-tests $(FUZZER_FLAGS)"

 .PHONY: apt-install
 apt-install:
@@ -404,7 +404,7 @@ cmakebuild:

 c89build: clean
 	$(CC) -v
-	CFLAGS="-std=c89 -Werror -O0" $(MAKE) allmost  # will fail, due to missing support for `long long`
+	CFLAGS="-std=c89 -Werror -Wno-attributes -Wpedantic -Wno-long-long -Wno-variadic-macros -O0" $(MAKE) lib zstd

 gnu90build: clean
 	$(CC) -v
@@ -13,15 +13,12 @@ a list of known ports and bindings is provided on [Zstandard homepage](https://f
 **Development branch status:**

 [![Build Status][travisDevBadge]][travisLink]
-[![Build status][AppveyorDevBadge]][AppveyorLink]
 [![Build status][CircleDevBadge]][CircleLink]
 [![Build status][CirrusDevBadge]][CirrusLink]
 [![Fuzzing Status][OSSFuzzBadge]][OSSFuzzLink]

 [travisDevBadge]: https://api.travis-ci.com/facebook/zstd.svg?branch=dev "Continuous Integration test suite"
 [travisLink]: https://travis-ci.com/facebook/zstd
-[AppveyorDevBadge]: https://ci.appveyor.com/api/projects/status/xt38wbdxjk5mrbem/branch/dev?svg=true "Windows test suite"
-[AppveyorLink]: https://ci.appveyor.com/project/YannCollet/zstd-p0yf0
 [CircleDevBadge]: https://circleci.com/gh/facebook/zstd/tree/dev.svg?style=shield "Short test suite"
 [CircleLink]: https://circleci.com/gh/facebook/zstd
 [CirrusDevBadge]: https://api.cirrus-ci.com/github/facebook/zstd.svg?branch=dev
@@ -154,6 +151,18 @@ to create `zstd` binary, and `libzstd` dynamic and static libraries.

 By default, `CMAKE_BUILD_TYPE` is set to `Release`.

+#### Support for Fat (Universal2) Output
+
+`zstd` can be built and installed with support for both Apple Silicon (M1/M2) as well as Intel by using CMake's Universal2 support.
+To perform a Fat/Universal2 build and install use the following commands:
+
+```bash
+cmake -B build-cmake-debug -S build/cmake -G Ninja -DCMAKE_OSX_ARCHITECTURES="x86_64;x86_64h;arm64"
+cd build-cmake-debug
+ninja
+sudo ninja install
+```
+
 ### Meson

 A Meson project is provided within [`build/meson`](build/meson). Follow
@@ -191,13 +200,15 @@ The output binary will be in `buck-out/gen/programs/`.

 ## Testing

-You can run quick local smoke tests by executing the `playTest.sh` script from the `src/tests` directory.
-Two env variables `$ZSTD_BIN` and `$DATAGEN_BIN` are needed for the test script to locate the zstd and datagen binary.
-For information on CI testing, please refer to TESTING.md
+You can run quick local smoke tests by running `make check`.
+If you can't use `make`, execute the `playTest.sh` script from the `src/tests` directory.
+Two env variables `$ZSTD_BIN` and `$DATAGEN_BIN` are needed for the test script to locate the `zstd` and `datagen` binary.
+For information on CI testing, please refer to `TESTING.md`.

 ## Status

-Zstandard is currently deployed within Facebook. It is used continuously to compress large amounts of data in multiple formats and use cases.
+Zstandard is currently deployed within Facebook and many other large cloud infrastructures.
+It is run continuously to compress large amounts of data in multiple formats and use cases.
 Zstandard is considered safe for production environments.

 ## License
@@ -169,7 +169,6 @@
      <WarningLevel>Level4</WarningLevel>
      <Optimization>Disabled</Optimization>
      <PreprocessorDefinitions>ZSTD_DLL_EXPORT=1;ZSTD_MULTITHREAD=1;ZSTD_LEGACY_SUPPORT=5;WIN32;_DEBUG;_CONSOLE;_CRT_SECURE_NO_WARNINGS;%(PreprocessorDefinitions)</PreprocessorDefinitions>
-      <MinimalRebuild>true</MinimalRebuild>
      <BasicRuntimeChecks>EnableFastChecks</BasicRuntimeChecks>
      <RuntimeLibrary>MultiThreadedDebugDLL</RuntimeLibrary>
      <DebugInformationFormat>EditAndContinue</DebugInformationFormat>
@@ -162,7 +162,6 @@
      <WarningLevel>Level4</WarningLevel>
      <Optimization>Disabled</Optimization>
      <PreprocessorDefinitions>ZSTD_MULTITHREAD=1;ZSTD_LEGACY_SUPPORT=5;WIN32;_DEBUG;_CONSOLE;_CRT_SECURE_NO_WARNINGS;%(PreprocessorDefinitions)</PreprocessorDefinitions>
-      <MinimalRebuild>true</MinimalRebuild>
      <BasicRuntimeChecks>EnableFastChecks</BasicRuntimeChecks>
      <RuntimeLibrary>MultiThreadedDebugDLL</RuntimeLibrary>
      <DebugInformationFormat>EditAndContinue</DebugInformationFormat>
@@ -114,6 +114,7 @@
    <RootNamespace>zstd</RootNamespace>
    <OutDir>$(SolutionDir)bin\$(Platform)_$(Configuration)\</OutDir>
    <IntDir>$(SolutionDir)bin\obj\$(RootNamespace)_$(Platform)_$(Configuration)\</IntDir>
+    <InstructionSet>NotSet</InstructionSet>
  </PropertyGroup>
  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'" Label="Configuration">
@@ -187,6 +188,7 @@
      <PreprocessorDefinitions>ZSTD_MULTITHREAD=1;ZSTD_LEGACY_SUPPORT=5;WIN32;_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
      <TreatWarningAsError>true</TreatWarningAsError>
      <EnablePREfast>false</EnablePREfast>
+      <EnableEnhancedInstructionSet>$(InstructionSet)</EnableEnhancedInstructionSet>
    </ClCompile>
    <Link>
      <SubSystem>Console</SubSystem>
@@ -203,6 +205,7 @@
      <PreprocessorDefinitions>ZSTD_MULTITHREAD=1;ZSTD_LEGACY_SUPPORT=5;WIN32;_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
      <TreatWarningAsError>true</TreatWarningAsError>
      <EnablePREfast>false</EnablePREfast>
+      <EnableEnhancedInstructionSet>$(InstructionSet)</EnableEnhancedInstructionSet>
    </ClCompile>
    <Link>
      <SubSystem>Console</SubSystem>
@@ -222,6 +225,7 @@
      <EnablePREfast>false</EnablePREfast>
      <TreatWarningAsError>false</TreatWarningAsError>
      <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
+      <EnableEnhancedInstructionSet>$(InstructionSet)</EnableEnhancedInstructionSet>
    </ClCompile>
    <Link>
      <SubSystem>Console</SubSystem>
@@ -244,6 +248,7 @@
      <EnablePREfast>false</EnablePREfast>
      <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
      <AdditionalOptions>/DZSTD_MULTITHREAD %(AdditionalOptions)</AdditionalOptions>
+      <EnableEnhancedInstructionSet>$(InstructionSet)</EnableEnhancedInstructionSet>
    </ClCompile>
    <Link>
      <SubSystem>Console</SubSystem>
@@ -1,6 +1,15 @@
 include(CheckCXXCompilerFlag)
 include(CheckCCompilerFlag)
-include(CheckLinkerFlag)
+# VERSION_GREATER_EQUAL requires CMake 3.7 or later.
+# https://cmake.org/cmake/help/latest/command/if.html#version-greater-equal
+if (CMAKE_VERSION VERSION_LESS 3.18)
+    set(ZSTD_HAVE_CHECK_LINKER_FLAG false)
+else ()
+    set(ZSTD_HAVE_CHECK_LINKER_FLAG true)
+endif ()
+if (ZSTD_HAVE_CHECK_LINKER_FLAG)
+    include(CheckLinkerFlag)
+endif()

 function(EnableCompilerFlag _flag _C _CXX _LD)
    string(REGEX REPLACE "\\+" "PLUS" varname "${_flag}")
@@ -20,7 +29,20 @@ function(EnableCompilerFlag _flag _C _CXX _LD)
        endif ()
    endif ()
    if (_LD)
-        CHECK_LINKER_FLAG(C ${_flag} LD_FLAG_${varname})
+        # We never add a linker flag with CMake < 3.18. We will
+        # implement CHECK_LINKER_FLAG() like feature for CMake < 3.18
+        # or require CMake >= 3.18 when we need to add a required
+        # linker flag in future.
+        #
+        # We also skip linker flags check for MSVC compilers (which includes
+        # clang-cl) since currently check_linker_flag() doesn't give correct
+        # results for this configuration,
+        # see: https://gitlab.kitware.com/cmake/cmake/-/issues/22023
+        if (ZSTD_HAVE_CHECK_LINKER_FLAG AND NOT MSVC)
+            CHECK_LINKER_FLAG(C ${_flag} LD_FLAG_${varname})
+        else ()
+            set(LD_FLAG_${varname} false)
+        endif ()
        if (LD_FLAG_${varname})
            set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} ${_flag}" PARENT_SCOPE)
            set(CMAKE_SHARED_LINKER_FLAGS "${CMAKE_SHARED_LINKER_FLAGS} ${_flag}" PARENT_SCOPE)
@@ -132,7 +132,7 @@ endif

 subdir('lib')

-if bin_programs
+if bin_programs or bin_tests
  subdir('programs')
 endif

@@ -72,7 +72,14 @@ zstd = executable('zstd',
  c_args: zstd_c_args,
  dependencies: zstd_deps,
  export_dynamic: export_dynamic_on_windows, # Since Meson 0.45.0
-  install: true)
+  build_by_default: bin_programs,
+  install: bin_programs)
+
+if not bin_programs
+  # we generate rules to build the programs, but don't install anything
+  # so do not continue to installing scripts and manpages
+  subdir_done()
+endif

 zstd_frugal_sources = [join_paths(zstd_rootdir, 'programs/zstdcli.c'),
  join_paths(zstd_rootdir, 'programs/timefn.c'),
@@ -162,7 +162,7 @@ if host_machine_os != os_windows
      playTests_sh,
      args: opt,
      env: ['ZSTD_BIN=' + zstd.full_path(), 'DATAGEN_BIN=./datagen'],
-      depends: [datagen],
+      depends: [datagen, zstd],
      suite: suite,
      workdir: meson.current_build_dir(),
      timeout: 2800) # Timeout should work on HDD drive
@@ -1,13 +1,13 @@
 # Dockerfile
 # First image to build the binary
-FROM alpine as builder
+FROM alpine@sha256:69665d02cb32192e52e07644d76bc6f25abeb5410edc1c7a81a10ba3f0efb90a as builder

 RUN apk --no-cache add make gcc libc-dev
 COPY . /src
 RUN mkdir /pkg && cd /src && make && make DESTDIR=/pkg install

 # Second minimal image to only keep the built binary
-FROM alpine
+FROM alpine@sha256:69665d02cb32192e52e07644d76bc6f25abeb5410edc1c7a81a10ba3f0efb90a

 # Copy the built files
 COPY --from=builder /pkg /
@@ -1,2 +0,0 @@
-# build artifacts
-externalMatchfinder
@@ -1,14 +0,0 @@
-externalMatchfinder
-=====================
-
-`externalMatchfinder` is a test tool for the external matchfinder API.
-It demonstrates how to use the API to perform a simple round-trip test.
-
-A sample matchfinder is provided in matchfinder.c, but the user can swap
-this out with a different one if desired. The sample matchfinder implements
-LZ compression with a 1KB hashtable. Dictionary compression is not currently supported.
-
-Command line :
-```
-externalMatchfinder filename
-```
@@ -0,0 +1,2 @@
+# build artifacts
+externalSequenceProducer
@@ -23,11 +23,11 @@ DEBUGFLAGS= -Wall -Wextra -Wcast-qual -Wcast-align -Wshadow \
            -Wredundant-decls
 CFLAGS  += $(DEBUGFLAGS) $(MOREFLAGS)

-default: externalMatchfinder
+default: externalSequenceProducer

-all: externalMatchfinder
+all: externalSequenceProducer

-externalMatchfinder: matchfinder.c main.c $(LIBZSTD)
+externalSequenceProducer: sequence_producer.c main.c $(LIBZSTD)
 	$(CC) $(CPPFLAGS) $(CFLAGS) $^ $(LDFLAGS) -o $@

 .PHONY: $(LIBZSTD)
@@ -37,4 +37,4 @@ $(LIBZSTD):
 clean:
 	$(RM) *.o
 	$(MAKE) -C $(LIBDIR) clean > /dev/null
-	$(RM) externalMatchfinder
+	$(RM) externalSequenceProducer
@@ -0,0 +1,14 @@
+externalSequenceProducer
+=====================
+
+`externalSequenceProducer` is a test tool for the Block-Level Sequence Producer API.
+It demonstrates how to use the API to perform a simple round-trip test.
+
+A sample sequence producer is provided in sequence_producer.c, but the user can swap
+this out with a different one if desired. The sample sequence producer implements
+LZ parsing with a 1KB hashtable. Dictionary-based parsing is not currently supported.
+
+Command line :
+```
+externalSequenceProducer filename
+```
@@ -16,7 +16,7 @@
 #define ZSTD_STATIC_LINKING_ONLY
 #include "zstd.h"
 #include "zstd_errors.h"
-#include "matchfinder.h" // simpleExternalMatchFinder
+#include "sequence_producer.h" // simpleSequenceProducer

 #define CHECK(res)                                      \
 do {                                                    \
@@ -28,23 +28,23 @@ do {                                                    \

 int main(int argc, char *argv[]) {
    if (argc != 2) {
-        printf("Usage: exampleMatchfinder <file>\n");
+        printf("Usage: externalSequenceProducer <file>\n");
        return 1;
    }

    ZSTD_CCtx* const zc = ZSTD_createCCtx();

-    int simpleExternalMatchState = 0xdeadbeef;
+    int simpleSequenceProducerState = 0xdeadbeef;

    // Here is the crucial bit of code!
-    ZSTD_registerExternalMatchFinder(
+    ZSTD_registerSequenceProducer(
        zc,
-        &simpleExternalMatchState,
-        simpleExternalMatchFinder
+        &simpleSequenceProducerState,
+        simpleSequenceProducer
    );

    {
-        size_t const res = ZSTD_CCtx_setParameter(zc, ZSTD_c_enableMatchFinderFallback, 1);
+        size_t const res = ZSTD_CCtx_setParameter(zc, ZSTD_c_enableSeqProducerFallback, 1);
        CHECK(res);
    }

@@ -9,15 +9,15 @@
 */

 #include "zstd_compress_internal.h"
-#include "matchfinder.h"
+#include "sequence_producer.h"

 #define HSIZE 1024
 static U32 const HLOG = 10;
 static U32 const MLS = 4;
 static U32 const BADIDX = 0xffffffff;

-size_t simpleExternalMatchFinder(
-  void* externalMatchState,
+size_t simpleSequenceProducer(
+  void* sequenceProducerState,
  ZSTD_Sequence* outSeqs, size_t outSeqsCapacity,
  const void* src, size_t srcSize,
  const void* dict, size_t dictSize,
@@ -31,7 +31,7 @@ size_t simpleExternalMatchFinder(
    size_t seqCount = 0;
    U32 hashTable[HSIZE];

-    (void)externalMatchState;
+    (void)sequenceProducerState;
    (void)dict;
    (void)dictSize;
    (void)outSeqsCapacity;
@@ -14,8 +14,8 @@
 #define ZSTD_STATIC_LINKING_ONLY
 #include "zstd.h"

-size_t simpleExternalMatchFinder(
-  void* externalMatchState,
+size_t simpleSequenceProducer(
+  void* sequenceProducerState,
  ZSTD_Sequence* outSeqs, size_t outSeqsCapacity,
  const void* src, size_t srcSize,
  const void* dict, size_t dictSize,
@@ -985,7 +985,7 @@ int usage(const char* exeName)
    DISPLAY ("-#          : use compression level # (default: %u) \n", CLEVEL_DEFAULT);
    DISPLAY ("-D #        : use # as a dictionary (default: create one) \n");
    DISPLAY ("-i#         : nb benchmark rounds (default: %u) \n", BENCH_TIME_DEFAULT_S);
-    DISPLAY ("-p#         : print speed for all rounds 0=fastest 1=median (default: 0)");
+    DISPLAY ("-p#         : print speed for all rounds 0=fastest 1=median (default: 0) \n");
    DISPLAY ("--nbBlocks=#: use # blocks for bench (default: one per file) \n");
    DISPLAY ("--nbDicts=# : create # dictionaries for bench (default: one per block) \n");
    DISPLAY ("-h          : help (this text) \n");
@@ -32,7 +32,6 @@ libzstd:
 		-DZSTD_LEGACY_SUPPORT=0 \
 		-DZSTD_STATIC_LINKING_ONLY \
 		-DFSE_STATIC_LINKING_ONLY \
-		-DHUF_STATIC_LINKING_ONLY \
 		-DXXH_STATIC_LINKING_ONLY \
 		-D__GNUC__ \
 		-D__linux__=1 \
@@ -24,9 +24,6 @@ EXPORT_SYMBOL_GPL(HUF_readStats_wksp);
 EXPORT_SYMBOL_GPL(ZSTD_isError);
 EXPORT_SYMBOL_GPL(ZSTD_getErrorName);
 EXPORT_SYMBOL_GPL(ZSTD_getErrorCode);
-EXPORT_SYMBOL_GPL(ZSTD_customMalloc);
-EXPORT_SYMBOL_GPL(ZSTD_customCalloc);
-EXPORT_SYMBOL_GPL(ZSTD_customFree);

 MODULE_LICENSE("Dual BSD/GPL");
 MODULE_DESCRIPTION("Zstd Common");
@@ -84,7 +84,7 @@ static uint64_t ZSTD_div64(uint64_t dividend, uint32_t divisor) {

 #include <linux/kernel.h>

-#define assert(x) WARN_ON((x))
+#define assert(x) WARN_ON(!(x))

 #endif /* ZSTD_DEPS_ASSERT */
 #endif /* ZSTD_DEPS_NEED_ASSERT */
@@ -37,10 +37,16 @@ CFLAGS   += -Wno-deprecated-declarations
 PZSTD_INC  = -I$(ZSTDDIR) -I$(ZSTDDIR)/common -I$(PROGDIR) -I.
 GTEST_INC  = -isystem googletest/googletest/include

+# If default C++ version is older than C++11, explicitly set C++11, which is the
+# minimum required by the code.
+ifeq ($(shell echo "\043if __cplusplus < 201103L\n\043error\n\043endif" | $(CXX) -x c++ -Werror -c - -o /dev/null 2>/dev/null && echo 1 || echo 0),0)
+PZSTD_CXX_STD := -std=c++11
+endif
+
 PZSTD_CPPFLAGS  = $(PZSTD_INC)
 PZSTD_CCXXFLAGS =
 PZSTD_CFLAGS    = $(PZSTD_CCXXFLAGS)
-PZSTD_CXXFLAGS  = $(PZSTD_CCXXFLAGS) -std=c++11
+PZSTD_CXXFLAGS  = $(PZSTD_CCXXFLAGS) $(PZSTD_CXX_STD)
 PZSTD_LDFLAGS   =
 EXTRA_FLAGS     =
 ALL_CFLAGS      = $(EXTRA_FLAGS) $(CPPFLAGS) $(PZSTD_CPPFLAGS) $(CFLAGS)   $(PZSTD_CFLAGS)
@@ -0,0 +1,42 @@
+# Zstandard Seekable Format
+
+The seekable format splits compressed data into a series of independent "frames",
+each compressed individually,
+so that decompression of a section in the middle of an archive
+only requires zstd to decompress at most a frame's worth of extra data,
+instead of the entire archive.
+
+The frames are appended, so that the decompression of the entire payload
+still regenerates the original content, using any compliant zstd decoder.
+
+On top of that, the seekable format generates a jump table,
+which makes it possible to jump directly to the position of the relevant frame
+when requesting only a segment of the data.
+The jump table is simply ignored by zstd decoders unaware of the seekable format.
+
+The format is delivered with an API to create seekable archives
+and to retrieve arbitrary segments inside the archive.
+
+### Maximum Frame Size parameter
+
+When creating a seekable archive, the main parameter is the maximum frame size.
+
+At compression time, user can manually select the boundaries between segments,
+but they don't have to: long segments will be automatically split
+when larger than selected maximum frame size.
+
+Small frame sizes reduce decompression cost when requesting small segments,
+because the decoder will nonetheless have to decompress an entire frame
+to recover just a single byte from it.
+
+A good rule of thumb is to select a maximum frame size roughly equivalent
+to the access pattern when it's known.
+For example, if the application tends to request 4KB blocks,
+then it's a good idea to set a maximum frame size in the vicinity of 4 KB.
+
+But small frame sizes also reduce compression ratio,
+and increase the cost for the jump table,
+so there is a balance to find.
+
+In general, try to avoid really tiny frame sizes (<1 KB),
+which would have a large negative impact on compression ratio.
@@ -25,7 +25,7 @@

 #include "pool.h"      // use zstd thread pool for demo

-#include "zstd_seekable.h"
+#include "../zstd_seekable.h"

 static void* malloc_orDie(size_t size)
 {
@@ -29,7 +29,7 @@

 #include "pool.h"      // use zstd thread pool for demo

-#include "zstd_seekable.h"
+#include "../zstd_seekable.h"

 #define MIN(a, b) ((a) < (b) ? (a) : (b))

@@ -13,7 +13,7 @@
 #define ZSTD_STATIC_LINKING_ONLY
 #include <zstd.h>      // presumes zstd library is installed

-#include "zstd_seekable.h"
+#include "../zstd_seekable.h"

 static void* malloc_orDie(size_t size)
 {
@@ -112,20 +112,23 @@ static char* createOutFilename_orDie(const char* filename)
    return (char*)outSpace;
 }

-int main(int argc, const char** argv) {
+#define CLEVEL_DEFAULT 5
+int main(int argc, const char** argv)
+{
    const char* const exeName = argv[0];
-    if (argc!=3) {
-        printf("wrong arguments\n");
-        printf("usage:\n");
-        printf("%s FILE FRAME_SIZE\n", exeName);
+    if (argc<3 || argc>4) {
+        printf("wrong arguments \n");
+        printf("usage: \n");
+        printf("%s FILE FRAME_SIZE [LEVEL] \n", exeName);
        return 1;
    }

    {   const char* const inFileName = argv[1];
        unsigned const frameSize = (unsigned)atoi(argv[2]);
+        int const cLevel = (argc==4) ? atoi(argv[3]) : CLEVEL_DEFAULT;

        char* const outFileName = createOutFilename_orDie(inFileName);
-        compressFile_orDie(inFileName, outFileName, 5, frameSize);
+        compressFile_orDie(inFileName, outFileName, cLevel, frameSize);
        free(outFileName);
    }

@@ -16,7 +16,7 @@
 #include <zstd.h>      // presumes zstd library is installed
 #include <zstd_errors.h>

-#include "zstd_seekable.h"
+#include "../zstd_seekable.h"

 #define MIN(a, b) ((a) < (b) ? (a) : (b))

@@ -3,8 +3,55 @@
 #include <stdlib.h>  // malloc
 #include <stdio.h>
 #include <assert.h>
+#include <string.h>

-#include "zstd_seekable.h"
+#include "../zstd_seekable.h"
+
+
+/* ZSTD_seekable_customFile implementation that reads/seeks a buffer while keeping track of total bytes read */
+typedef struct {
+    const void *ptr;
+    size_t size;
+    size_t pos;
+    size_t totalRead;
+} buffWrapperWithTotal_t;
+
+static int readBuffWithTotal(void* opaque, void* buffer, size_t n)
+{
+    buffWrapperWithTotal_t* const buff = (buffWrapperWithTotal_t*)opaque;
+    assert(buff != NULL);
+    if (buff->pos + n > buff->size) return -1;
+    memcpy(buffer, (const char*)buff->ptr + buff->pos, n);
+    buff->pos += n;
+    buff->totalRead += n;
+    return 0;
+}
+
+static int seekBuffWithTotal(void* opaque, long long offset, int origin)
+{
+    buffWrapperWithTotal_t* const buff = (buffWrapperWithTotal_t*) opaque;
+    unsigned long long newOffset;
+    assert(buff != NULL);
+    switch (origin) {
+    case SEEK_SET:
+        assert(offset >= 0);
+        newOffset = (unsigned long long)offset;
+        break;
+    case SEEK_CUR:
+        newOffset = (unsigned long long)((long long)buff->pos + offset);
+        break;
+    case SEEK_END:
+        newOffset = (unsigned long long)((long long)buff->size + offset);
+        break;
+    default:
+        assert(0);  /* not possible */
+    }
+    if (newOffset > buff->size) {
+        return -1;
+    }
+    buff->pos = newOffset;
+    return 0;
+}

 /* Basic unit tests for zstd seekable format */
 int main(int argc, const char** argv)
@@ -220,6 +267,92 @@ int main(int argc, const char** argv)
    }
    printf("Success!\n");

+
+    printf("Test %u - multiple decompress calls: ", testNb++);
+    {   char const inBuffer[] = "Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt";
+        size_t const inSize = sizeof(inBuffer);
+
+        size_t const seekCapacity = 5000;
+        void* const seekBuffer = malloc(seekCapacity);
+        assert(seekBuffer != NULL);
+        size_t seekSize;
+
+        size_t const outCapacity = inSize;
+        char* const outBuffer = malloc(outCapacity);
+        assert(outBuffer != NULL);
+
+        ZSTD_seekable_CStream* const zscs = ZSTD_seekable_createCStream();
+        assert(zscs != NULL);
+
+        /* compress test data with a small frame size to ensure multiple frames in the output */
+        unsigned const maxFrameSize = 40;
+        { size_t const initStatus = ZSTD_seekable_initCStream(zscs, 9, 0 /* checksumFlag */, maxFrameSize);
+          assert(!ZSTD_isError(initStatus));
+        }
+
+        {   ZSTD_outBuffer outb = { .dst=seekBuffer, .pos=0, .size=seekCapacity };
+            ZSTD_inBuffer inb = { .src=inBuffer, .pos=0, .size=inSize };
+
+            while (inb.pos < inb.size) {
+                size_t const cStatus = ZSTD_seekable_compressStream(zscs, &outb, &inb);
+                assert(!ZSTD_isError(cStatus));
+            }
+
+            size_t const endStatus = ZSTD_seekable_endStream(zscs, &outb);
+            assert(!ZSTD_isError(endStatus));
+            seekSize = outb.pos;
+        }
+
+        ZSTD_seekable* const stream = ZSTD_seekable_create();
+        assert(stream != NULL);
+        buffWrapperWithTotal_t buffWrapper = {seekBuffer, seekSize, 0, 0};
+        { ZSTD_seekable_customFile srcFile = {&buffWrapper, &readBuffWithTotal, &seekBuffWithTotal};
+          size_t const initStatus = ZSTD_seekable_initAdvanced(stream, srcFile);
+          assert(!ZSTD_isError(initStatus)); }
+
+        /* Perform a series of small reads and seeks (repeatedly read 1 byte and skip 1 byte)
+           and check that we didn't reread input data unnecessarily */
+        size_t pos;
+        for (pos = 0; pos < inSize; pos += 2) {
+            size_t const decStatus = ZSTD_seekable_decompress(stream, outBuffer, 1, pos);
+            if (decStatus != 1 || outBuffer[0] != inBuffer[pos]) {
+                goto _test_error;
+            }
+        }
+        if (buffWrapper.totalRead > seekSize) {
+            /* We read more than the compressed size, meaning there were some rereads.
+               This is unneeded because we only seeked forward. */
+            printf("Too much data read: %zu read, with compressed size %zu\n", buffWrapper.totalRead, seekSize);
+            goto _test_error;
+        }
+
+        /* Perform some reads and seeks to ensure correctness */
+        struct {
+            size_t offset;
+            size_t size;
+        } const tests[] = {  /* Assume the frame size is 40 */
+            {20, 40}, /* read partial data from two frames */
+            {60, 10}, /* continue reading from the same offset */
+            {50, 20}, /* seek backward within the same frame */
+            {10, 10}, /* seek backward to a different frame */
+            {25, 10}, /* seek forward within the same frame */
+            {60, 10}, /* seek forward to a different frame */
+        };
+        size_t idx;
+        for (idx = 0; idx < sizeof(tests) / sizeof(tests[0]); idx++) {
+            size_t const decStatus = ZSTD_seekable_decompress(stream, outBuffer, tests[idx].size, tests[idx].offset);
+            if (decStatus != tests[idx].size || memcmp(outBuffer, inBuffer + tests[idx].offset, tests[idx].size) != 0) {
+                goto _test_error;
+            }
+        }
+
+        free(seekBuffer);
+        free(outBuffer);
+        ZSTD_seekable_freeCStream(zscs);
+        ZSTD_seekable_free(stream);
+    }
+    printf("Success!\n");
+
    /* TODO: Add more tests */
    printf("Finished tests\n");
    return 0;
@@ -15,8 +15,8 @@ extern "C" {

 #define ZSTD_SEEKABLE_MAXFRAMES 0x8000000U

-/* Limit the maximum size to avoid any potential issues storing the compressed size */
-#define ZSTD_SEEKABLE_MAX_FRAME_DECOMPRESSED_SIZE 0x80000000U
+/* Limit maximum size to avoid potential issues storing the compressed size */
+#define ZSTD_SEEKABLE_MAX_FRAME_DECOMPRESSED_SIZE 0x40000000U

 /*-****************************************************************************
 *  Seekable Format
@@ -48,10 +48,19 @@ typedef struct ZSTD_seekTable_s ZSTD_seekTable;
 *
 *  Use ZSTD_seekable_initCStream() to initialize a ZSTD_seekable_CStream object
 *  for a new compression operation.
-*  `maxFrameSize` indicates the size at which to automatically start a new
-*  seekable frame.  `maxFrameSize == 0` implies the default maximum size.
-*  `checksumFlag` indicates whether or not the seek table should include frame
-*  checksums on the uncompressed data for verification.
+*  - `maxFrameSize` indicates the size at which to automatically start a new
+*            seekable frame.
+*            `maxFrameSize == 0` implies the default maximum size.
+*            Smaller frame sizes allow faster decompression of small segments,
+*            since retrieving a single byte requires decompression of
+*            the full frame where the byte belongs.
+*            In general, size the frames to roughly correspond to
+*            the access granularity (when it's known).
+*            But small sizes also reduce compression ratio.
+*            Avoid really tiny frame sizes (< 1 KB),
+*            that would hurt compression ratio considerably.
+*  - `checksumFlag` indicates whether or not the seek table should include frame
+*            checksums on the uncompressed data for verification.
 *  @return : a size hint for input to provide for compression, or an error code
 *            checkable with ZSTD_isError()
 *
@@ -230,6 +230,8 @@ size_t ZSTD_seekable_compressStream(ZSTD_seekable_CStream* zcs, ZSTD_outBuffer*
    const BYTE* const inBase = (const BYTE*) input->src + input->pos;
    size_t inLen = input->size - input->pos;

+    assert(zcs->maxFrameSize < INT_MAX);
+    ZSTD_CCtx_setParameter(zcs->cstream, ZSTD_c_srcSizeHint, (int)zcs->maxFrameSize);
    inLen = MIN(inLen, (size_t)(zcs->maxFrameSize - zcs->frameDSize));

    /* if we haven't finished flushing the last frame, don't start writing a new one */
@@ -493,7 +493,7 @@ size_t ZSTD_seekable_decompress(ZSTD_seekable* zs, void* dst, size_t len, unsign
    size_t srcBytesRead = 0;
    do {
        /* check if we can continue from a previous decompress job */
-        if (targetFrame != zs->curFrame || offset != zs->decompressedOffset) {
+        if (targetFrame != zs->curFrame || offset < zs->decompressedOffset) {
            zs->decompressedOffset = zs->seekTable.entries[targetFrame].dOffset;
            zs->curFrame = targetFrame;

@@ -517,9 +517,9 @@ size_t ZSTD_seekable_decompress(ZSTD_seekable* zs, void* dst, size_t len, unsign
            size_t forwardProgress;
            if (zs->decompressedOffset < offset) {
                /* dummy decompressions until we get to the target offset */
-                outTmp = (ZSTD_outBuffer){zs->outBuff, MIN(SEEKABLE_BUFF_SIZE, offset - zs->decompressedOffset), 0};
+                outTmp = (ZSTD_outBuffer){zs->outBuff, (size_t) (MIN(SEEKABLE_BUFF_SIZE, offset - zs->decompressedOffset)), 0};
            } else {
-                outTmp = (ZSTD_outBuffer){dst, len, zs->decompressedOffset - offset};
+                outTmp = (ZSTD_outBuffer){dst, len, (size_t) (zs->decompressedOffset - offset)};
            }

            prevOutPos = outTmp.pos;
@@ -16,7 +16,7 @@ Distribution of this document is unlimited.

 ### Version

-0.3.7 (2020-12-09)
+0.3.9 (2023-03-08)


 Introduction
@@ -470,6 +470,7 @@ This field uses 2 lowest bits of first byte, describing 4 different block types
        repeated `Regenerated_Size` times.
 - `Compressed_Literals_Block` - This is a standard Huffman-compressed block,
        starting with a Huffman tree description.
+        In this mode, there are at least 2 different literals represented in the Huffman tree description.
        See details below.
 - `Treeless_Literals_Block` - This is a Huffman-compressed block,
        using Huffman tree _from previous Huffman-compressed literals block_.
@@ -533,15 +534,20 @@ __`Size_Format` for `Compressed_Literals_Block` and `Treeless_Literals_Block`__
 Both `Compressed_Size` and `Regenerated_Size` fields follow __little-endian__ convention.
 Note: `Compressed_Size` __includes__ the size of the Huffman Tree description
 _when_ it is present.
+Note 2: `Compressed_Size` can never be `==0`.
+Even in single-stream scenario, assuming an empty content, it must be `>=1`,
+since it contains at least the final end bit flag.
+In 4-streams scenario, a valid `Compressed_Size` is necessarily `>= 10`
+(6 bytes for the jump table, + 4x1 bytes for the 4 streams).

-4 streams is superior to 1 stream in decompression speed,
+4 streams is faster than 1 stream in decompression speed,
 by exploiting instruction level parallelism.
 But it's also more expensive,
 costing on average ~7.3 bytes more than the 1 stream mode, mostly from the jump table.

 In general, use the 4 streams mode when there are more literals to decode,
 to favor higher decompression speeds.
-Beyond 1KB, the 4 streams mode is compulsory anyway.
+Note that beyond >1KB of literals, the 4 streams mode is compulsory.

 Note that a minimum of 6 bytes is required for the 4 streams mode.
 That's a technical minimum, but it's not recommended to employ the 4 streams mode
@@ -566,6 +572,7 @@ or from a dictionary.

 ### `Huffman_Tree_Description`
 This section is only present when `Literals_Block_Type` type is `Compressed_Literals_Block` (`2`).
+The tree describes the weights of all literals symbols that can be present in the literals block, at least 2 and up to 256.
 The format of the Huffman tree description can be found at [Huffman Tree description](#huffman-tree-description).
 The size of `Huffman_Tree_Description` is determined during decoding process,
 it must be used to determine where streams begin.
@@ -575,10 +582,10 @@ it must be used to determine where streams begin.
 ### Jump Table
 The Jump Table is only present when there are 4 Huffman-coded streams.

-Reminder : Huffman compressed data consists of either 1 or 4 Huffman-coded streams.
+Reminder : Huffman compressed data consists of either 1 or 4 streams.

 If only one stream is present, it is a single bitstream occupying the entire
-remaining portion of the literals block, encoded as described within
+remaining portion of the literals block, encoded as described in
 [Huffman-Coded Streams](#huffman-coded-streams).

 If there are four streams, `Literals_Section_Header` only provided
@@ -589,17 +596,18 @@ except for the last stream which may be up to 3 bytes smaller,
 to reach a total decompressed size as specified in `Regenerated_Size`.

 The compressed size of each stream is provided explicitly in the Jump Table.
-Jump Table is 6 bytes long, and consist of three 2-byte __little-endian__ fields,
+Jump Table is 6 bytes long, and consists of three 2-byte __little-endian__ fields,
 describing the compressed sizes of the first three streams.
-`Stream4_Size` is computed from total `Total_Streams_Size` minus sizes of other streams.
+`Stream4_Size` is computed from `Total_Streams_Size` minus sizes of other streams:

 `Stream4_Size = Total_Streams_Size - 6 - Stream1_Size - Stream2_Size - Stream3_Size`.

-Note: if `Stream1_Size + Stream2_Size + Stream3_Size > Total_Streams_Size`,
+`Stream4_Size` is necessarily `>= 1`. Therefore,
+if `Total_Streams_Size < Stream1_Size + Stream2_Size + Stream3_Size + 6 + 1`,
 data is considered corrupted.

 Each of these 4 bitstreams is then decoded independently as a Huffman-Coded stream,
-as described at [Huffman-Coded Streams](#huffman-coded-streams)
+as described in [Huffman-Coded Streams](#huffman-coded-streams)


 Sequences Section
@@ -1197,7 +1205,7 @@ Huffman Coding
 --------------
 Zstandard Huffman-coded streams are read backwards,
 similar to the FSE bitstreams.
-Therefore, to find the start of the bitstream, it is therefore to
+Therefore, to find the start of the bitstream, it is required to
 know the offset of the last byte of the Huffman-coded stream.

 After writing the last bit containing information, the compressor
@@ -1239,9 +1247,15 @@ Transformation from `Weight` to `Number_of_Bits` follows this formula :
 ```
 Number_of_Bits = Weight ? (Max_Number_of_Bits + 1 - Weight) : 0
 ```
-The last symbol's `Weight` is deduced from previously decoded ones,
-by completing to the nearest power of 2.
-This power of 2 gives `Max_Number_of_Bits`, the depth of the current tree.
+When a literal value is not present, it receives a `Weight` of 0.
+The least frequent symbol receives a `Weight` of 1.
+Consequently, the `Weight` 1 is necessarily present.
+The most frequent symbol receives a `Weight` anywhere between 1 and 11 (max).
+The last symbol's `Weight` is deduced from previously retrieved Weights,
+by completing to the nearest power of 2. It's necessarily non 0.
+If it's not possible to reach a clean power of 2 with a single `Weight` value,
+the Huffman Tree Description is considered invalid.
+This final power of 2 gives `Max_Number_of_Bits`, the depth of the current tree.
 `Max_Number_of_Bits` must be <= 11,
 otherwise the representation is considered corrupted.

@@ -1254,7 +1268,7 @@ Let's presume the following Huffman tree must be described :

 The tree depth is 4, since its longest elements uses 4 bits
 (longest elements are the one with smallest frequency).
-Value `5` will not be listed, as it can be determined from values for 0-4,
+Literal value `5` will not be listed, as it can be determined from previous values 0-4,
 nor will values above `5` as they are all 0.
 Values from `0` to `4` will be listed using `Weight` instead of `Number_of_Bits`.
 Weight formula is :
@@ -1274,7 +1288,7 @@ The `Weight` of `5` can be determined by advancing to the next power of 2.
 The sum of `2^(Weight-1)` (excluding 0's) is :
 `8 + 4 + 2 + 0 + 1 = 15`.
 Nearest larger power of 2 value is 16.
-Therefore, `Max_Number_of_Bits = 4` and `Weight[5] = 16-15 = 1`.
+Therefore, `Max_Number_of_Bits = 4` and `Weight[5] = log_2(16 - 15) + 1 = 1`.

 #### Huffman Tree header

@@ -1683,6 +1697,8 @@ or at least provide a meaningful error code explaining for which reason it canno

 Version changes
 ---------------
+- 0.3.9 : clarifications for Huffman-compressed literal sizes.
+- 0.3.8 : clarifications for Huffman Blocks and Huffman Tree descriptions.
 - 0.3.7 : clarifications for Repeat_Offsets, matching RFC8878
 - 0.3.6 : clarifications for Dictionary_ID
 - 0.3.5 : clarifications for Block_Maximum_Size
@@ -1,10 +1,10 @@
 <html>
 <head>
 <meta http-equiv="Content-Type" content="text/html; charset=ISO-8859-1">
-<title>zstd 1.5.3 Manual</title>
+<title>zstd 1.5.5 Manual</title>
 </head>
 <body>
-<h1>zstd 1.5.3 Manual</h1>
+<h1>zstd 1.5.5 Manual</h1>
 <hr>
 <a name="Contents"></a><h2>Contents</h2>
 <ol>
@@ -22,15 +22,15 @@
 <li><a href="#Chapter12">Dictionary helper functions</a></li>
 <li><a href="#Chapter13">Advanced dictionary and prefix API (Requires v1.4.0+)</a></li>
 <li><a href="#Chapter14">experimental API (static linking only)</a></li>
-<li><a href="#Chapter15">Frame size functions</a></li>
+<li><a href="#Chapter15">Frame header and size functions</a></li>
 <li><a href="#Chapter16">Memory management</a></li>
 <li><a href="#Chapter17">Advanced compression functions</a></li>
 <li><a href="#Chapter18">Advanced decompression functions</a></li>
 <li><a href="#Chapter19">Advanced streaming functions</a></li>
-<li><a href="#Chapter20">Buffer-less and synchronous inner streaming functions</a></li>
+<li><a href="#Chapter20">Buffer-less and synchronous inner streaming functions (DEPRECATED)</a></li>
 <li><a href="#Chapter21">Buffer-less streaming compression (synchronous mode)</a></li>
 <li><a href="#Chapter22">Buffer-less streaming decompression (synchronous mode)</a></li>
-<li><a href="#Chapter23">Block level API</a></li>
+<li><a href="#Chapter23">Block level API (DEPRECATED)</a></li>
 </ol>
 <hr>
 <a name="Chapter1"></a><h2>Introduction</h2><pre>
@@ -80,7 +80,8 @@
                const void* src, size_t srcSize,
                      int compressionLevel);
 </b><p>  Compresses `src` content as a single zstd compressed frame into already allocated `dst`.
-  Hint : compression runs faster if `dstCapacity` >=  `ZSTD_compressBound(srcSize)`.
+  NOTE: Providing `dstCapacity >= ZSTD_compressBound(srcSize)` guarantees that zstd will have
+        enough space to successfully compress the data.
  @return : compressed size written into `dst` (<= `dstCapacity),
            or an error code if it fails (which can be tested using ZSTD_isError()). 
 </p></pre><BR>
@@ -121,8 +122,8 @@ unsigned long long ZSTD_getFrameContentSize(const void *src, size_t srcSize);
   note 6 : This function replaces ZSTD_getDecompressedSize() 
 </p></pre><BR>

-<pre><b>ZSTDLIB_API
-ZSTD_DEPRECATED("Replaced by ZSTD_getFrameContentSize")
+<pre><b>ZSTD_DEPRECATED("Replaced by ZSTD_getFrameContentSize")
+ZSTDLIB_API
 unsigned long long ZSTD_getDecompressedSize(const void* src, size_t srcSize);
 </b><p>  NOTE: This function is now obsolete, in favor of ZSTD_getFrameContentSize().
  Both functions work the same way, but ZSTD_getDecompressedSize() blends
@@ -139,8 +140,30 @@ unsigned long long ZSTD_getDecompressedSize(const void* src, size_t srcSize);
        or an error code if input is invalid 
 </p></pre><BR>

-<h3>Helper functions</h3><pre></pre><b><pre>#define ZSTD_COMPRESSBOUND(srcSize)   ((srcSize) + ((srcSize)>>8) + (((srcSize) < (128<<10)) ? (((128<<10) - (srcSize)) >> 11) </b>/* margin, from 64 to 0 */ : 0))  /* this formula ensures that bound(A) + bound(B) <= bound(A+B) as long as A and B >= 128 KB */<b>
-size_t      ZSTD_compressBound(size_t srcSize); </b>/*!< maximum compressed size in worst case single-pass scenario */<b>
+<h3>Helper functions</h3><pre></pre><b><pre></b>/* ZSTD_compressBound() :<b>
+ * maximum compressed size in worst case single-pass scenario.
+ * When invoking `ZSTD_compress()` or any other one-pass compression function,
+ * it's recommended to provide @dstCapacity >= ZSTD_compressBound(srcSize)
+ * as it eliminates one potential failure scenario,
+ * aka not enough room in dst buffer to write the compressed frame.
+ * Note : ZSTD_compressBound() itself can fail, if @srcSize > ZSTD_MAX_INPUT_SIZE .
+ *        In which case, ZSTD_compressBound() will return an error code
+ *        which can be tested using ZSTD_isError().
+ *
+ * ZSTD_COMPRESSBOUND() :
+ * same as ZSTD_compressBound(), but as a macro.
+ * It can be used to produce constants, which can be useful for static allocation,
+ * for example to size a static array on stack.
+ * Will produce constant value 0 if srcSize too large.
+ */
+#define ZSTD_MAX_INPUT_SIZE ((sizeof(size_t)==8) ? 0xFF00FF00FF00FF00LLU : 0xFF00FF00U)
+#define ZSTD_COMPRESSBOUND(srcSize)   (((size_t)(srcSize) >= ZSTD_MAX_INPUT_SIZE) ? 0 : (srcSize) + ((srcSize)>>8) + (((srcSize) < (128<<10)) ? (((128<<10) - (srcSize)) >> 11) </b>/* margin, from 64 to 0 */ : 0))  /* this formula ensures that bound(A) + bound(B) <= bound(A+B) as long as A and B >= 128 KB */<b>
+size_t ZSTD_compressBound(size_t srcSize); </b>/*!< maximum compressed size in worst case single-pass scenario */<b>
+</b>/* ZSTD_isError() :<b>
+ * Most ZSTD_* functions returning a size_t value can be tested for error,
+ * using ZSTD_isError().
+ * @return 1 if error, 0 otherwise
+ */
 unsigned    ZSTD_isError(size_t code);          </b>/*!< tells if a `size_t` function result is an error code */<b>
 const char* ZSTD_getErrorName(size_t code);     </b>/*!< provides readable string from an error code */<b>
 int         ZSTD_minCLevel(void);               </b>/*!< minimum negative compression level allowed, requires v1.4.0+ */<b>
@@ -362,6 +385,8 @@ size_t     ZSTD_freeDCtx(ZSTD_DCtx* dctx);  </b>/* accept NULL pointer */<b>
     * ZSTD_c_useBlockSplitter
     * ZSTD_c_useRowMatchFinder
     * ZSTD_c_prefetchCDictTables
+     * ZSTD_c_enableSeqProducerFallback
+     * ZSTD_c_maxBlockSize
     * Because they are not stable, it's necessary to define ZSTD_STATIC_LINKING_ONLY to access them.
     * note : never ever use experimentalParam? names directly;
     *        also, the enums values themselves are unstable and can still change.
@@ -381,7 +406,10 @@ size_t     ZSTD_freeDCtx(ZSTD_DCtx* dctx);  </b>/* accept NULL pointer */<b>
     ZSTD_c_experimentalParam13=1010,
     ZSTD_c_experimentalParam14=1011,
     ZSTD_c_experimentalParam15=1012,
-     ZSTD_c_experimentalParam16=1013
+     ZSTD_c_experimentalParam16=1013,
+     ZSTD_c_experimentalParam17=1014,
+     ZSTD_c_experimentalParam18=1015,
+     ZSTD_c_experimentalParam19=1016
 } ZSTD_cParameter;
 </b></pre><BR>
 <pre><b>typedef struct {
@@ -444,7 +472,7 @@ size_t     ZSTD_freeDCtx(ZSTD_DCtx* dctx);  </b>/* accept NULL pointer */<b>
                  They will be used to compress next frame.
                  Resetting session never fails.
  - The parameters : changes all parameters back to "default".
-                  This removes any reference to any dictionary too.
+                  This also removes any reference to any dictionary or external sequence producer.
                  Parameters can only be changed between 2 sessions (i.e. no compression is currently ongoing)
                  otherwise the reset fails, and function returns an error value (which can be tested using ZSTD_isError())
  - Both : similar to resetting the session, followed by resetting parameters.
@@ -459,7 +487,8 @@ size_t     ZSTD_freeDCtx(ZSTD_DCtx* dctx);  </b>/* accept NULL pointer */<b>
  Should cctx hold data from a previously unfinished frame, everything about it is forgotten.
  - Compression parameters are pushed into CCtx before starting compression, using ZSTD_CCtx_set*()
  - The function is always blocking, returns when compression is completed.
-  Hint : compression runs faster if `dstCapacity` >=  `ZSTD_compressBound(srcSize)`.
+  NOTE: Providing `dstCapacity >= ZSTD_compressBound(srcSize)` guarantees that zstd will have
+        enough space to successfully compress the data, though it is possible it fails for other reasons.
 @return : compressed size written into `dst` (<= `dstCapacity),
           or an error code if it fails (which can be tested using ZSTD_isError()).
 
@@ -483,13 +512,15 @@ size_t     ZSTD_freeDCtx(ZSTD_DCtx* dctx);  </b>/* accept NULL pointer */<b>
     * ZSTD_d_stableOutBuffer
     * ZSTD_d_forceIgnoreChecksum
     * ZSTD_d_refMultipleDDicts
+     * ZSTD_d_disableHuffmanAssembly
     * Because they are not stable, it's necessary to define ZSTD_STATIC_LINKING_ONLY to access them.
     * note : never ever use experimentalParam? names directly
     */
     ZSTD_d_experimentalParam1=1000,
     ZSTD_d_experimentalParam2=1001,
     ZSTD_d_experimentalParam3=1002,
-     ZSTD_d_experimentalParam4=1003
+     ZSTD_d_experimentalParam4=1003,
+     ZSTD_d_experimentalParam5=1004

 } ZSTD_dParameter;
 </b></pre><BR>
@@ -658,6 +689,9 @@ size_t ZSTD_endStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output);
     ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
     ZSTD_CCtx_refCDict(zcs, NULL); // clear the dictionary (if any)
     ZSTD_CCtx_setParameter(zcs, ZSTD_c_compressionLevel, compressionLevel);
+
+ Note that ZSTD_initCStream() clears any previously set dictionary. Use the new API
+ to compress with a dictionary.
 
 </p></pre><BR>

@@ -693,6 +727,33 @@ size_t ZSTD_endStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output);
 size_t ZSTD_freeDStream(ZSTD_DStream* zds);  </b>/* accept NULL pointer */<b>
 </pre></b><BR>
 <h3>Streaming decompression functions</h3><pre></pre><b><pre></pre></b><BR>
+<pre><b>size_t ZSTD_initDStream(ZSTD_DStream* zds);
+</b><p> Initialize/reset DStream state for new decompression operation.
+ Call before new decompression operation using same DStream.
+
+ Note : This function is redundant with the advanced API and equivalent to:
+     ZSTD_DCtx_reset(zds, ZSTD_reset_session_only);
+     ZSTD_DCtx_refDDict(zds, NULL);
+ 
+</p></pre><BR>
+
+<pre><b>size_t ZSTD_decompressStream(ZSTD_DStream* zds, ZSTD_outBuffer* output, ZSTD_inBuffer* input);
+</b><p> Streaming decompression function.
+ Call repetitively to consume full input updating it as necessary.
+ Function will update both input and output `pos` fields exposing current state via these fields:
+ - `input.pos < input.size`, some input remaining and caller should provide remaining input
+   on the next call.
+ - `output.pos < output.size`, decoder finished and flushed all remaining buffers.
+ - `output.pos == output.size`, potentially uncflushed data present in the internal buffers,
+   call ZSTD_decompressStream() again to flush remaining data to output.
+ Note : with no additional input, amount of data flushed <= ZSTD_BLOCKSIZE_MAX.
+
+ @return : 0 when a frame is completely decoded and fully flushed,
+           or an error code, which can be tested using ZSTD_isError(),
+           or any other value > 0, which means there is some decoding or flushing to do to complete current frame.
+ 
+</p></pre><BR>
+
 <pre><b>size_t ZSTD_DStreamInSize(void);    </b>/*!< recommended size for input buffer */<b>
 </b></pre><BR>
 <pre><b>size_t ZSTD_DStreamOutSize(void);   </b>/*!< recommended size for output buffer. Guarantee to successfully flush at least one complete block in all circumstances. */<b>
@@ -807,9 +868,11 @@ size_t ZSTD_freeDStream(ZSTD_DStream* zds);  </b>/* accept NULL pointer */<b>

 <a name="Chapter13"></a><h2>Advanced dictionary and prefix API (Requires v1.4.0+)</h2><pre>
 This API allows dictionaries to be used with ZSTD_compress2(),
- ZSTD_compressStream2(), and ZSTD_decompressDCtx(). Dictionaries are sticky, and
- only reset with the context is reset with ZSTD_reset_parameters or
- ZSTD_reset_session_and_parameters. Prefixes are single-use.
+ ZSTD_compressStream2(), and ZSTD_decompressDCtx().
+ Dictionaries are sticky, they remain valid when same context is re-used,
+ they only reset when the context is reset
+ with ZSTD_reset_parameters or ZSTD_reset_session_and_parameters.
+ In contrast, Prefixes are single-use.
 <BR></pre>

 <pre><b>size_t ZSTD_CCtx_loadDictionary(ZSTD_CCtx* cctx, const void* dict, size_t dictSize);
@@ -818,8 +881,9 @@ size_t ZSTD_freeDStream(ZSTD_DStream* zds);  </b>/* accept NULL pointer */<b>
 @result : 0, or an error code (which can be tested with ZSTD_isError()).
  Special: Loading a NULL (or 0-size) dictionary invalidates previous dictionary,
           meaning "return to no-dictionary mode".
-  Note 1 : Dictionary is sticky, it will be used for all future compressed frames.
-           To return to "no-dictionary" situation, load a NULL dictionary (or reset parameters).
+  Note 1 : Dictionary is sticky, it will be used for all future compressed frames,
+           until parameters are reset, a new dictionary is loaded, or the dictionary
+           is explicitly invalidated by loading a NULL dictionary.
  Note 2 : Loading a dictionary involves building tables.
           It's also a CPU consuming operation, with non-negligible impact on latency.
           Tables are dependent on compression parameters, and for this reason,
@@ -828,11 +892,15 @@ size_t ZSTD_freeDStream(ZSTD_DStream* zds);  </b>/* accept NULL pointer */<b>
           Use experimental ZSTD_CCtx_loadDictionary_byReference() to reference content instead.
           In such a case, dictionary buffer must outlive its users.
  Note 4 : Use ZSTD_CCtx_loadDictionary_advanced()
-           to precisely select how dictionary content must be interpreted. 
+           to precisely select how dictionary content must be interpreted.
+  Note 5 : This method does not benefit from LDM (long distance mode).
+           If you want to employ LDM on some large dictionary content,
+           prefer employing ZSTD_CCtx_refPrefix() described below.
+ 
 </p></pre><BR>

 <pre><b>size_t ZSTD_CCtx_refCDict(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict);
-</b><p>  Reference a prepared dictionary, to be used for all next compressed frames.
+</b><p>  Reference a prepared dictionary, to be used for all future compressed frames.
  Note that compression parameters are enforced from within CDict,
  and supersede any compression parameter previously set within CCtx.
  The parameters ignored are labelled as "superseded-by-cdict" in the ZSTD_cParameter enum docs.
@@ -852,6 +920,7 @@ size_t ZSTD_freeDStream(ZSTD_DStream* zds);  </b>/* accept NULL pointer */<b>
  Decompression will need same prefix to properly regenerate data.
  Compressing with a prefix is similar in outcome as performing a diff and compressing it,
  but performs much faster, especially during decompression (compression speed is tunable with compression level).
+  This method is compatible with LDM (long distance mode).
 @result : 0, or an error code (which can be tested with ZSTD_isError()).
  Special: Adding any prefix (including NULL) invalidates any previous prefix or dictionary
  Note 1 : Prefix buffer is referenced. It **must** outlive compression.
@@ -867,9 +936,9 @@ size_t ZSTD_freeDStream(ZSTD_DStream* zds);  </b>/* accept NULL pointer */<b>
 </p></pre><BR>

 <pre><b>size_t ZSTD_DCtx_loadDictionary(ZSTD_DCtx* dctx, const void* dict, size_t dictSize);
-</b><p>  Create an internal DDict from dict buffer,
-  to be used to decompress next frames.
-  The dictionary remains valid for all future frames, until explicitly invalidated.
+</b><p>  Create an internal DDict from dict buffer, to be used to decompress all future frames.
+  The dictionary remains valid for all future frames, until explicitly invalidated, or
+  a new dictionary is loaded.
 @result : 0, or an error code (which can be tested with ZSTD_isError()).
  Special : Adding a NULL (or 0-size) dictionary invalidates any previous dictionary,
            meaning "return to no-dictionary mode".
@@ -893,9 +962,10 @@ size_t ZSTD_freeDStream(ZSTD_DStream* zds);  </b>/* accept NULL pointer */<b>
  The memory for the table is allocated on the first call to refDDict, and can be
  freed with ZSTD_freeDCtx().

+  If called with ZSTD_d_refMultipleDDicts disabled (the default), only one dictionary
+  will be managed, and referencing a dictionary effectively "discards" any previous one.
+
 @result : 0, or an error code (which can be tested with ZSTD_isError()).
-  Note 1 : Currently, only one dictionary can be managed.
-           Referencing a new dictionary effectively "discards" any previous one.
  Special: referencing a NULL DDict means "return to no-dictionary mode".
  Note 2 : DDict is just referenced, its lifetime must outlive its usage from DCtx.
 
@@ -1085,7 +1155,7 @@ size_t ZSTD_sizeof_DDict(const ZSTD_DDict* ddict);
  ZSTD_ps_disable = 2       </b>/* Do not use the feature */<b>
 } ZSTD_paramSwitch_e;
 </b></pre><BR>
-<a name="Chapter15"></a><h2>Frame size functions</h2><pre></pre>
+<a name="Chapter15"></a><h2>Frame header and size functions</h2><pre></pre>

 <pre><b>ZSTDLIB_STATIC_API unsigned long long ZSTD_findDecompressedSize(const void* src, size_t srcSize);
 </b><p>  `src` should point to the start of a series of ZSTD encoded and/or skippable frames
@@ -1131,11 +1201,90 @@ size_t ZSTD_sizeof_DDict(const ZSTD_DDict* ddict);
           or an error code (if srcSize is too small) 
 </p></pre><BR>

+<pre><b>typedef enum { ZSTD_frame, ZSTD_skippableFrame } ZSTD_frameType_e;
+</b></pre><BR>
+<pre><b>typedef struct {
+    unsigned long long frameContentSize; </b>/* if == ZSTD_CONTENTSIZE_UNKNOWN, it means this field is not available. 0 means "empty" */<b>
+    unsigned long long windowSize;       </b>/* can be very large, up to <= frameContentSize */<b>
+    unsigned blockSizeMax;
+    ZSTD_frameType_e frameType;          </b>/* if == ZSTD_skippableFrame, frameContentSize is the size of skippable content */<b>
+    unsigned headerSize;
+    unsigned dictID;
+    unsigned checksumFlag;
+    unsigned _reserved1;
+    unsigned _reserved2;
+} ZSTD_frameHeader;
+</b></pre><BR>
+<pre><b>ZSTDLIB_STATIC_API size_t ZSTD_getFrameHeader(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize);   </b>/**< doesn't consume input */<b>
+</b>/*! ZSTD_getFrameHeader_advanced() :<b>
+ *  same as ZSTD_getFrameHeader(),
+ *  with added capability to select a format (like ZSTD_f_zstd1_magicless) */
+ZSTDLIB_STATIC_API size_t ZSTD_getFrameHeader_advanced(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize, ZSTD_format_e format);
+</b><p>  decode Frame Header, or requires larger `srcSize`.
+ @return : 0, `zfhPtr` is correctly filled,
+          >0, `srcSize` is too small, value is wanted `srcSize` amount,
+           or an error code, which can be tested using ZSTD_isError() 
+</p></pre><BR>
+
+<pre><b>ZSTDLIB_STATIC_API size_t ZSTD_decompressionMargin(const void* src, size_t srcSize);
+</b><p> Zstd supports in-place decompression, where the input and output buffers overlap.
+ In this case, the output buffer must be at least (Margin + Output_Size) bytes large,
+ and the input buffer must be at the end of the output buffer.
+
+  _______________________ Output Buffer ________________________
+ |                                                              |
+ |                                        ____ Input Buffer ____|
+ |                                       |                      |
+ v                                       v                      v
+ |---------------------------------------|-----------|----------|
+ ^                                                   ^          ^
+ |___________________ Output_Size ___________________|_ Margin _|
+
+ NOTE: See also ZSTD_DECOMPRESSION_MARGIN().
+ NOTE: This applies only to single-pass decompression through ZSTD_decompress() or
+ ZSTD_decompressDCtx().
+ NOTE: This function supports multi-frame input.
+
+ @param src The compressed frame(s)
+ @param srcSize The size of the compressed frame(s)
+ @returns The decompression margin or an error that can be checked with ZSTD_isError().
+ 
+</p></pre><BR>
+
+<pre><b>#define ZSTD_DECOMPRESSION_MARGIN(originalSize, blockSize) ((size_t)(                                              \
+        ZSTD_FRAMEHEADERSIZE_MAX                                                              </b>/* Frame header */ + \<b>
+        4                                                                                         </b>/* checksum */ + \<b>
+        ((originalSize) == 0 ? 0 : 3 * (((originalSize) + (blockSize) - 1) / blockSize)) </b>/* 3 bytes per block */ + \<b>
+        (blockSize)                                                                    </b>/* One block of margin */   \<b>
+    ))
+</b><p> Similar to ZSTD_decompressionMargin(), but instead of computing the margin from
+ the compressed frame, compute it from the original size and the blockSizeLog.
+ See ZSTD_decompressionMargin() for details.
+
+ WARNING: This macro does not support multi-frame input, the input must be a single
+ zstd frame. If you need that support use the function, or implement it yourself.
+
+ @param originalSize The original uncompressed size of the data.
+ @param blockSize    The block size == MIN(windowSize, ZSTD_BLOCKSIZE_MAX).
+                     Unless you explicitly set the windowLog smaller than
+                     ZSTD_BLOCKSIZELOG_MAX you can just use ZSTD_BLOCKSIZE_MAX.
+ 
+</p></pre><BR>
+
 <pre><b>typedef enum {
  ZSTD_sf_noBlockDelimiters = 0,         </b>/* Representation of ZSTD_Sequence has no block delimiters, sequences only */<b>
  ZSTD_sf_explicitBlockDelimiters = 1    </b>/* Representation of ZSTD_Sequence contains explicit block delimiters */<b>
 } ZSTD_sequenceFormat_e;
 </b></pre><BR>
+<pre><b>ZSTDLIB_STATIC_API size_t ZSTD_sequenceBound(size_t srcSize);
+</b><p> `srcSize` : size of the input buffer
+  @return : upper-bound for the number of sequences that can be generated
+            from a buffer of srcSize bytes
+
+  note : returns number of sequences - to get bytes, multiply by sizeof(ZSTD_Sequence).
+ 
+</p></pre><BR>
+
 <pre><b></b><p> Generate sequences using ZSTD_compress2(), given a source buffer.

 Each block will end with a dummy sequence
@@ -1260,8 +1409,11 @@ ZSTDLIB_STATIC_API size_t ZSTD_estimateDCtxSize(void);
  and ZSTD_estimateCCtxSize_usingCCtxParams(), which can be used in tandem with ZSTD_CCtxParams_setParameter().
  Both can be used to estimate memory using custom compression parameters and arbitrary srcSize limits.

-  Note 2 : only single-threaded compression is supported.
+  Note : only single-threaded compression is supported.
  ZSTD_estimateCCtxSize_usingCCtxParams() will return an error code if ZSTD_c_nbWorkers is >= 1.
+
+  Note 2 : ZSTD_estimateCCtxSize* functions are not compatible with the Block-Level Sequence Producer API at this time.
+  Size estimates assume that no external sequence producer is registered.
 
 </p></pre><BR>

@@ -1281,7 +1433,12 @@ ZSTDLIB_STATIC_API size_t ZSTD_estimateDStreamSize_fromFrame(const void* src, si
  or deducted from a valid frame Header, using ZSTD_estimateDStreamSize_fromFrame();
  Note : if streaming is init with function ZSTD_init?Stream_usingDict(),
         an internal ?Dict will be created, which additional size is not estimated here.
-         In this case, get total size by adding ZSTD_estimate?DictSize 
+         In this case, get total size by adding ZSTD_estimate?DictSize
+  Note 2 : only single-threaded compression is supported.
+  ZSTD_estimateCStreamSize_usingCCtxParams() will return an error code if ZSTD_c_nbWorkers is >= 1.
+  Note 3 : ZSTD_estimateCStreamSize* functions are not compatible with the Block-Level Sequence Producer API at this time.
+  Size estimates assume that no external sequence producer is registered.
+ 
 </p></pre><BR>

 <pre><b>ZSTDLIB_STATIC_API size_t ZSTD_estimateCDictSize(size_t dictSize, int compressionLevel);
@@ -1381,20 +1538,41 @@ ZSTDLIB_STATIC_API size_t ZSTD_CCtx_refThreadPool(ZSTD_CCtx* cctx, ZSTD_threadPo
  This function never fails (wide contract) 
 </p></pre><BR>

-<pre><b>ZSTDLIB_STATIC_API
-ZSTD_DEPRECATED("use ZSTD_compress2")
+<pre><b>ZSTDLIB_STATIC_API size_t ZSTD_CCtx_setCParams(ZSTD_CCtx* cctx, ZSTD_compressionParameters cparams);
+</b><p>  Set all parameters provided within @p cparams into the working @p cctx.
+  Note : if modifying parameters during compression (MT mode only),
+         note that changes to the .windowLog parameter will be ignored.
+ @return 0 on success, or an error code (can be checked with ZSTD_isError()).
+         On failure, no parameters are updated.
+ 
+</p></pre><BR>
+
+<pre><b>ZSTDLIB_STATIC_API size_t ZSTD_CCtx_setFParams(ZSTD_CCtx* cctx, ZSTD_frameParameters fparams);
+</b><p>  Set all parameters provided within @p fparams into the working @p cctx.
+ @return 0 on success, or an error code (can be checked with ZSTD_isError()).
+ 
+</p></pre><BR>
+
+<pre><b>ZSTDLIB_STATIC_API size_t ZSTD_CCtx_setParams(ZSTD_CCtx* cctx, ZSTD_parameters params);
+</b><p>  Set all parameters provided within @p params into the working @p cctx.
+ @return 0 on success, or an error code (can be checked with ZSTD_isError()).
+ 
+</p></pre><BR>
+
+<pre><b>ZSTD_DEPRECATED("use ZSTD_compress2")
+ZSTDLIB_STATIC_API
 size_t ZSTD_compress_advanced(ZSTD_CCtx* cctx,
-                              void* dst, size_t dstCapacity,
-                        const void* src, size_t srcSize,
-                        const void* dict,size_t dictSize,
-                              ZSTD_parameters params);
+                  void* dst, size_t dstCapacity,
+            const void* src, size_t srcSize,
+            const void* dict,size_t dictSize,
+                  ZSTD_parameters params);
 </b><p>  Note : this function is now DEPRECATED.
         It can be replaced by ZSTD_compress2(), in combination with ZSTD_CCtx_setParameter() and other parameter setters.
  This prototype will generate compilation warnings. 
 </p></pre><BR>

-<pre><b>ZSTDLIB_STATIC_API
-ZSTD_DEPRECATED("use ZSTD_compress2 with ZSTD_CCtx_loadDictionary")
+<pre><b>ZSTD_DEPRECATED("use ZSTD_compress2 with ZSTD_CCtx_loadDictionary")
+ZSTDLIB_STATIC_API
 size_t ZSTD_compress_usingCDict_advanced(ZSTD_CCtx* cctx,
                                  void* dst, size_t dstCapacity,
                            const void* src, size_t srcSize,
@@ -1555,8 +1733,8 @@ ZSTDLIB_STATIC_API size_t ZSTD_freeCCtxParams(ZSTD_CCtx_params* params);  </b>/*
 
 </p></pre><BR>

-<pre><b>ZSTDLIB_STATIC_API
-ZSTD_DEPRECATED("use ZSTD_DCtx_setParameter() instead")
+<pre><b>ZSTD_DEPRECATED("use ZSTD_DCtx_setParameter() instead")
+ZSTDLIB_STATIC_API
 size_t ZSTD_DCtx_setFormat(ZSTD_DCtx* dctx, ZSTD_format_e format);
 </b><p>  This function is REDUNDANT. Prefer ZSTD_DCtx_setParameter().
  Instruct the decoder context about what kind of data to decode next.
@@ -1582,8 +1760,8 @@ size_t ZSTD_DCtx_setFormat(ZSTD_DCtx* dctx, ZSTD_format_e format);
 <BR></pre>

 <h3>Advanced Streaming compression functions</h3><pre></pre><b><pre></pre></b><BR>
-<pre><b>ZSTDLIB_STATIC_API
-ZSTD_DEPRECATED("use ZSTD_CCtx_reset, see zstd.h for detailed instructions")
+<pre><b>ZSTD_DEPRECATED("use ZSTD_CCtx_reset, see zstd.h for detailed instructions")
+ZSTDLIB_STATIC_API
 size_t ZSTD_initCStream_srcSize(ZSTD_CStream* zcs,
             int compressionLevel,
             unsigned long long pledgedSrcSize);
@@ -1600,8 +1778,8 @@ size_t ZSTD_initCStream_srcSize(ZSTD_CStream* zcs,
 
 </p></pre><BR>

-<pre><b>ZSTDLIB_STATIC_API
-ZSTD_DEPRECATED("use ZSTD_CCtx_reset, see zstd.h for detailed instructions")
+<pre><b>ZSTD_DEPRECATED("use ZSTD_CCtx_reset, see zstd.h for detailed instructions")
+ZSTDLIB_STATIC_API
 size_t ZSTD_initCStream_usingDict(ZSTD_CStream* zcs,
         const void* dict, size_t dictSize,
               int compressionLevel);
@@ -1618,18 +1796,15 @@ size_t ZSTD_initCStream_usingDict(ZSTD_CStream* zcs,
 
 </p></pre><BR>

-<pre><b>ZSTDLIB_STATIC_API
-ZSTD_DEPRECATED("use ZSTD_CCtx_reset, see zstd.h for detailed instructions")
+<pre><b>ZSTD_DEPRECATED("use ZSTD_CCtx_reset, see zstd.h for detailed instructions")
+ZSTDLIB_STATIC_API
 size_t ZSTD_initCStream_advanced(ZSTD_CStream* zcs,
        const void* dict, size_t dictSize,
              ZSTD_parameters params,
              unsigned long long pledgedSrcSize);
-</b><p> This function is DEPRECATED, and is approximately equivalent to:
+</b><p> This function is DEPRECATED, and is equivalent to:
     ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
-     // Pseudocode: Set each zstd parameter and leave the rest as-is.
-     for ((param, value) : params) {
-         ZSTD_CCtx_setParameter(zcs, param, value);
-     }
+     ZSTD_CCtx_setParams(zcs, params);
     ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize);
     ZSTD_CCtx_loadDictionary(zcs, dict, dictSize);

@@ -1640,8 +1815,8 @@ size_t ZSTD_initCStream_advanced(ZSTD_CStream* zcs,
 
 </p></pre><BR>

-<pre><b>ZSTDLIB_STATIC_API
-ZSTD_DEPRECATED("use ZSTD_CCtx_reset and ZSTD_CCtx_refCDict, see zstd.h for detailed instructions")
+<pre><b>ZSTD_DEPRECATED("use ZSTD_CCtx_reset and ZSTD_CCtx_refCDict, see zstd.h for detailed instructions")
+ZSTDLIB_STATIC_API
 size_t ZSTD_initCStream_usingCDict(ZSTD_CStream* zcs, const ZSTD_CDict* cdict);
 </b><p> This function is DEPRECATED, and equivalent to:
     ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
@@ -1652,18 +1827,15 @@ size_t ZSTD_initCStream_usingCDict(ZSTD_CStream* zcs, const ZSTD_CDict* cdict);
 
 </p></pre><BR>

-<pre><b>ZSTDLIB_STATIC_API
-ZSTD_DEPRECATED("use ZSTD_CCtx_reset and ZSTD_CCtx_refCDict, see zstd.h for detailed instructions")
+<pre><b>ZSTD_DEPRECATED("use ZSTD_CCtx_reset and ZSTD_CCtx_refCDict, see zstd.h for detailed instructions")
+ZSTDLIB_STATIC_API
 size_t ZSTD_initCStream_usingCDict_advanced(ZSTD_CStream* zcs,
                   const ZSTD_CDict* cdict,
                         ZSTD_frameParameters fParams,
                         unsigned long long pledgedSrcSize);
-</b><p>   This function is DEPRECATED, and is approximately equivalent to:
+</b><p>   This function is DEPRECATED, and is equivalent to:
     ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
-     // Pseudocode: Set each zstd frame parameter and leave the rest as-is.
-     for ((fParam, value) : fParams) {
-         ZSTD_CCtx_setParameter(zcs, fParam, value);
-     }
+     ZSTD_CCtx_setFParams(zcs, fParams);
     ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize);
     ZSTD_CCtx_refCDict(zcs, cdict);

@@ -1674,8 +1846,8 @@ size_t ZSTD_initCStream_usingCDict_advanced(ZSTD_CStream* zcs,
 
 </p></pre><BR>

-<pre><b>ZSTDLIB_STATIC_API
-ZSTD_DEPRECATED("use ZSTD_CCtx_reset, see zstd.h for detailed instructions")
+<pre><b>ZSTD_DEPRECATED("use ZSTD_CCtx_reset, see zstd.h for detailed instructions")
+ZSTDLIB_STATIC_API
 size_t ZSTD_resetCStream(ZSTD_CStream* zcs, unsigned long long pledgedSrcSize);
 </b><p> This function is DEPRECATED, and is equivalent to:
     ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
@@ -1721,39 +1893,71 @@ size_t ZSTD_resetCStream(ZSTD_CStream* zcs, unsigned long long pledgedSrcSize);
 </p></pre><BR>

 <h3>Advanced Streaming decompression functions</h3><pre></pre><b><pre></pre></b><BR>
-<pre><b>ZSTDLIB_STATIC_API size_t ZSTD_initDStream_usingDict(ZSTD_DStream* zds, const void* dict, size_t dictSize);
+<pre><b>ZSTD_DEPRECATED("use ZSTD_DCtx_reset + ZSTD_DCtx_loadDictionary, see zstd.h for detailed instructions")
+ZSTDLIB_STATIC_API size_t ZSTD_initDStream_usingDict(ZSTD_DStream* zds, const void* dict, size_t dictSize);
 </b><p>
     ZSTD_DCtx_reset(zds, ZSTD_reset_session_only);
     ZSTD_DCtx_loadDictionary(zds, dict, dictSize);

 note: no dictionary will be used if dict == NULL or dictSize < 8
- Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
 
 </p></pre><BR>

-<pre><b>ZSTDLIB_STATIC_API size_t ZSTD_initDStream_usingDDict(ZSTD_DStream* zds, const ZSTD_DDict* ddict);
+<pre><b>ZSTD_DEPRECATED("use ZSTD_DCtx_reset + ZSTD_DCtx_refDDict, see zstd.h for detailed instructions")
+ZSTDLIB_STATIC_API size_t ZSTD_initDStream_usingDDict(ZSTD_DStream* zds, const ZSTD_DDict* ddict);
 </b><p>
     ZSTD_DCtx_reset(zds, ZSTD_reset_session_only);
     ZSTD_DCtx_refDDict(zds, ddict);

 note : ddict is referenced, it must outlive decompression session
- Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
 
 </p></pre><BR>

-<pre><b>ZSTDLIB_STATIC_API size_t ZSTD_resetDStream(ZSTD_DStream* zds);
+<pre><b>ZSTD_DEPRECATED("use ZSTD_DCtx_reset, see zstd.h for detailed instructions")
+ZSTDLIB_STATIC_API size_t ZSTD_resetDStream(ZSTD_DStream* zds);
 </b><p>
     ZSTD_DCtx_reset(zds, ZSTD_reset_session_only);

 re-use decompression parameters from previous init; saves dictionary loading
- Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
 
 </p></pre><BR>

-<a name="Chapter20"></a><h2>Buffer-less and synchronous inner streaming functions</h2><pre>
-  This is an advanced API, giving full control over buffer management, for users which need direct control over memory.
-  But it's also a complex one, with several restrictions, documented below.
-  Prefer normal streaming API for an easier experience.
+<pre><b>ZSTDLIB_STATIC_API void
+ZSTD_registerSequenceProducer(
+  ZSTD_CCtx* cctx,
+  void* sequenceProducerState,
+  ZSTD_sequenceProducer_F* sequenceProducer
+);
+</b><p> Instruct zstd to use a block-level external sequence producer function.
+
+ The sequenceProducerState must be initialized by the caller, and the caller is
+ responsible for managing its lifetime. This parameter is sticky across
+ compressions. It will remain set until the user explicitly resets compression
+ parameters.
+
+ Sequence producer registration is considered to be an "advanced parameter",
+ part of the "advanced API". This means it will only have an effect on compression
+ APIs which respect advanced parameters, such as compress2() and compressStream2().
+ Older compression APIs such as compressCCtx(), which predate the introduction of
+ "advanced parameters", will ignore any external sequence producer setting.
+
+ The sequence producer can be "cleared" by registering a NULL function pointer. This
+ removes all limitations described above in the "LIMITATIONS" section of the API docs.
+
+ The user is strongly encouraged to read the full API documentation (above) before
+ calling this function. 
+</p></pre><BR>
+
+<a name="Chapter20"></a><h2>Buffer-less and synchronous inner streaming functions (DEPRECATED)</h2><pre>
+  This API is deprecated, and will be removed in a future version.
+  It allows streaming (de)compression with user allocated buffers.
+  However, it is hard to use, and not as well tested as the rest of
+  our API.
+
+  Please use the normal streaming API instead: ZSTD_compressStream2,
+  and ZSTD_decompressStream.
+  If there is functionality that you need, but it doesn't provide,
+  please open an issue on our GitHub.
 
 <BR></pre>

@@ -1764,7 +1968,6 @@ size_t ZSTD_resetCStream(ZSTD_CStream* zcs, unsigned long long pledgedSrcSize);

  Start by initializing a context.
  Use ZSTD_compressBegin(), or ZSTD_compressBegin_usingDict() for dictionary compression.
-  It's also possible to duplicate a reference context which has already been initialized, using ZSTD_copyCCtx()

  Then, consume your input using ZSTD_compressContinue().
  There are some important considerations to keep in mind when using this advanced function :
@@ -1785,11 +1988,15 @@ size_t ZSTD_resetCStream(ZSTD_CStream* zcs, unsigned long long pledgedSrcSize);
  `ZSTD_CCtx` object can be re-used (ZSTD_compressBegin()) to compress again.
 <BR></pre>

-<h3>Buffer-less streaming compression functions</h3><pre></pre><b><pre>ZSTDLIB_STATIC_API size_t ZSTD_compressBegin(ZSTD_CCtx* cctx, int compressionLevel);
+<h3>Buffer-less streaming compression functions</h3><pre></pre><b><pre>ZSTD_DEPRECATED("The buffer-less API is deprecated in favor of the normal streaming API. See docs.")
+ZSTDLIB_STATIC_API size_t ZSTD_compressBegin(ZSTD_CCtx* cctx, int compressionLevel);
+ZSTD_DEPRECATED("The buffer-less API is deprecated in favor of the normal streaming API. See docs.")
 ZSTDLIB_STATIC_API size_t ZSTD_compressBegin_usingDict(ZSTD_CCtx* cctx, const void* dict, size_t dictSize, int compressionLevel);
+ZSTD_DEPRECATED("The buffer-less API is deprecated in favor of the normal streaming API. See docs.")
 ZSTDLIB_STATIC_API size_t ZSTD_compressBegin_usingCDict(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict); </b>/**< note: fails if cdict==NULL */<b>
-ZSTDLIB_STATIC_API size_t ZSTD_copyCCtx(ZSTD_CCtx* cctx, const ZSTD_CCtx* preparedCCtx, unsigned long long pledgedSrcSize); </b>/**<  note: if pledgedSrcSize is not known, use ZSTD_CONTENTSIZE_UNKNOWN */<b>
 </pre></b><BR>
+<pre><b>size_t ZSTD_copyCCtx(ZSTD_CCtx* cctx, const ZSTD_CCtx* preparedCCtx, unsigned long long pledgedSrcSize); </b>/**<  note: if pledgedSrcSize is not known, use ZSTD_CONTENTSIZE_UNKNOWN */<b>
+</b></pre><BR>
 <pre><b>size_t ZSTD_compressBegin_advanced(ZSTD_CCtx* cctx, const void* dict, size_t dictSize, ZSTD_parameters params, unsigned long long pledgedSrcSize); </b>/**< pledgedSrcSize : If srcSize is not known at init time, use ZSTD_CONTENTSIZE_UNKNOWN */<b>
 </b></pre><BR>
 <a name="Chapter22"></a><h2>Buffer-less streaming decompression (synchronous mode)</h2><pre>
@@ -1801,8 +2008,8 @@ ZSTDLIB_STATIC_API size_t ZSTD_copyCCtx(ZSTD_CCtx* cctx, const ZSTD_CCtx* prepar
  Frame header is extracted from the beginning of compressed frame, so providing only the frame's beginning is enough.
  Data fragment must be large enough to ensure successful decoding.
 `ZSTD_frameHeaderSize_max` bytes is guaranteed to always be large enough.
-  @result : 0 : successful decoding, the `ZSTD_frameHeader` structure is correctly filled.
-           >0 : `srcSize` is too small, please provide at least @result bytes on next attempt.
+  result  : 0 : successful decoding, the `ZSTD_frameHeader` structure is correctly filled.
+           >0 : `srcSize` is too small, please provide at least result bytes on next attempt.
           errorCode, which can be tested using ZSTD_isError().

  It fills a ZSTD_frameHeader structure with important information to correctly decode the frame,
@@ -1821,7 +2028,7 @@ ZSTDLIB_STATIC_API size_t ZSTD_copyCCtx(ZSTD_CCtx* cctx, const ZSTD_CCtx* prepar

  The most memory efficient way is to use a round buffer of sufficient size.
  Sufficient size is determined by invoking ZSTD_decodingBufferSize_min(),
-  which can @return an error code if required value is too large for current system (in 32-bits mode).
+  which can return an error code if required value is too large for current system (in 32-bits mode).
  In a round buffer methodology, ZSTD_decompressContinue() decompresses each block next to previous one,
  up to the moment there is not enough room left in the buffer to guarantee decoding another full block,
  which maximum size is provided in `ZSTD_frameHeader` structure, field `blockSizeMax`.
@@ -1841,7 +2048,7 @@ ZSTDLIB_STATIC_API size_t ZSTD_copyCCtx(ZSTD_CCtx* cctx, const ZSTD_CCtx* prepar
  ZSTD_nextSrcSizeToDecompress() tells how many bytes to provide as 'srcSize' to ZSTD_decompressContinue().
  ZSTD_decompressContinue() requires this _exact_ amount of bytes, or it will fail.

- @result of ZSTD_decompressContinue() is the number of bytes regenerated within 'dst' (necessarily <= dstCapacity).
+  result of ZSTD_decompressContinue() is the number of bytes regenerated within 'dst' (necessarily <= dstCapacity).
  It can be zero : it just means ZSTD_decompressContinue() has decoded some metadata item.
  It can also be an error code, which can be tested with ZSTD_isError().

@@ -1863,34 +2070,25 @@ ZSTDLIB_STATIC_API size_t ZSTD_copyCCtx(ZSTD_CCtx* cctx, const ZSTD_CCtx* prepar
  For skippable frames ZSTD_decompressContinue() always returns 0 : it only skips the content.
 <BR></pre>

-<h3>Buffer-less streaming decompression functions</h3><pre></pre><b><pre>typedef enum { ZSTD_frame, ZSTD_skippableFrame } ZSTD_frameType_e;
-typedef struct {
-    unsigned long long frameContentSize; </b>/* if == ZSTD_CONTENTSIZE_UNKNOWN, it means this field is not available. 0 means "empty" */<b>
-    unsigned long long windowSize;       </b>/* can be very large, up to <= frameContentSize */<b>
-    unsigned blockSizeMax;
-    ZSTD_frameType_e frameType;          </b>/* if == ZSTD_skippableFrame, frameContentSize is the size of skippable content */<b>
-    unsigned headerSize;
-    unsigned dictID;
-    unsigned checksumFlag;
-} ZSTD_frameHeader;
-</pre></b><BR>
-<pre><b>ZSTDLIB_STATIC_API size_t ZSTD_getFrameHeader(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize);   </b>/**< doesn't consume input */<b>
-</b>/*! ZSTD_getFrameHeader_advanced() :<b>
- *  same as ZSTD_getFrameHeader(),
- *  with added capability to select a format (like ZSTD_f_zstd1_magicless) */
-ZSTDLIB_STATIC_API size_t ZSTD_getFrameHeader_advanced(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize, ZSTD_format_e format);
-ZSTDLIB_STATIC_API size_t ZSTD_decodingBufferSize_min(unsigned long long windowSize, unsigned long long frameContentSize);  </b>/**< when frame content size is not known, pass in frameContentSize == ZSTD_CONTENTSIZE_UNKNOWN */<b>
-</b><p>  decode Frame Header, or requires larger `srcSize`.
- @return : 0, `zfhPtr` is correctly filled,
-          >0, `srcSize` is too small, value is wanted `srcSize` amount,
-           or an error code, which can be tested using ZSTD_isError() 
-</p></pre><BR>
-
+<h3>Buffer-less streaming decompression functions</h3><pre></pre><b><pre></pre></b><BR>
+<pre><b>ZSTDLIB_STATIC_API size_t ZSTD_decodingBufferSize_min(unsigned long long windowSize, unsigned long long frameContentSize);  </b>/**< when frame content size is not known, pass in frameContentSize == ZSTD_CONTENTSIZE_UNKNOWN */<b>
+</b></pre><BR>
 <pre><b>typedef enum { ZSTDnit_frameHeader, ZSTDnit_blockHeader, ZSTDnit_block, ZSTDnit_lastBlock, ZSTDnit_checksum, ZSTDnit_skippableFrame } ZSTD_nextInputType_e;
 </b></pre><BR>
-<a name="Chapter23"></a><h2>Block level API</h2><pre></pre>
+<a name="Chapter23"></a><h2>Block level API (DEPRECATED)</h2><pre></pre>

-<pre><b></b><p>    Frame metadata cost is typically ~12 bytes, which can be non-negligible for very small blocks (< 100 bytes).
+<pre><b></b><p>    You can get the frame header down to 2 bytes by setting:
+      - ZSTD_c_format = ZSTD_f_zstd1_magicless
+      - ZSTD_c_contentSizeFlag = 0
+      - ZSTD_c_checksumFlag = 0
+      - ZSTD_c_dictIDFlag = 0
+
+    This API is not as well tested as our normal API, so we recommend not using it.
+    We will be removing it in a future version. If the normal API doesn't provide
+    the functionality you need, please open a GitHub issue.
+
+    Block functions produce and decode raw zstd blocks, without frame metadata.
+    Frame metadata cost is typically ~12 bytes, which can be non-negligible for very small blocks (< 100 bytes).
    But users will have to take in charge needed metadata to regenerate data, such as compressed and content sizes.

    A few rules to respect :
@@ -1899,7 +2097,6 @@ ZSTDLIB_STATIC_API size_t ZSTD_decodingBufferSize_min(unsigned long long windowS
    - It is necessary to init context before starting
      + compression : any ZSTD_compressBegin*() variant, including with dictionary
      + decompression : any ZSTD_decompressBegin*() variant, including with dictionary
-      + copyCCtx() and copyDCtx() can be used too
    - Block size is limited, it must be <= ZSTD_getBlockSize() <= ZSTD_BLOCKSIZE_MAX == 128 KB
      + If input is larger than a block size, it's necessary to split input data into multiple blocks
      + For inputs larger than a single block, consider using regular ZSTD_compress() instead.
@@ -1915,9 +2112,13 @@ ZSTDLIB_STATIC_API size_t ZSTD_decodingBufferSize_min(unsigned long long windowS
        Use ZSTD_insertBlock() for such a case.
 </p></pre><BR>

-<h3>Raw zstd block functions</h3><pre></pre><b><pre>ZSTDLIB_STATIC_API size_t ZSTD_getBlockSize   (const ZSTD_CCtx* cctx);
+<h3>Raw zstd block functions</h3><pre></pre><b><pre>ZSTD_DEPRECATED("The block API is deprecated in favor of the normal compression API. See docs.")
+ZSTDLIB_STATIC_API size_t ZSTD_getBlockSize   (const ZSTD_CCtx* cctx);
+ZSTD_DEPRECATED("The block API is deprecated in favor of the normal compression API. See docs.")
 ZSTDLIB_STATIC_API size_t ZSTD_compressBlock  (ZSTD_CCtx* cctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
+ZSTD_DEPRECATED("The block API is deprecated in favor of the normal compression API. See docs.")
 ZSTDLIB_STATIC_API size_t ZSTD_decompressBlock(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
+ZSTD_DEPRECATED("The block API is deprecated in favor of the normal compression API. See docs.")
 ZSTDLIB_STATIC_API size_t ZSTD_insertBlock    (ZSTD_DCtx* dctx, const void* blockStart, size_t blockSize);  </b>/**< insert uncompressed block into `dctx` history. Useful for multi-blocks decompression. */<b>
 </pre></b><BR>
 </html>
@@ -0,0 +1,55 @@
+/*
+ * Copyright (c) Meta Platforms, Inc. and affiliates.
+ * All rights reserved.
+ *
+ * This source code is licensed under both the BSD-style license (found in the
+ * LICENSE file in the root directory of this source tree) and the GPLv2 (found
+ * in the COPYING file in the root directory of this source tree).
+ * You may select, at your option, one of the above-listed licenses.
+ */
+
+/* This file provides custom allocation primitives
+ */
+
+#define ZSTD_DEPS_NEED_MALLOC
+#include "zstd_deps.h"   /* ZSTD_malloc, ZSTD_calloc, ZSTD_free, ZSTD_memset */
+
+#include "mem.h" /* MEM_STATIC */
+#define ZSTD_STATIC_LINKING_ONLY
+#include "../zstd.h" /* ZSTD_customMem */
+
+#ifndef ZSTD_ALLOCATIONS_H
+#define ZSTD_ALLOCATIONS_H
+
+/* custom memory allocation functions */
+
+MEM_STATIC void* ZSTD_customMalloc(size_t size, ZSTD_customMem customMem)
+{
+    if (customMem.customAlloc)
+        return customMem.customAlloc(customMem.opaque, size);
+    return ZSTD_malloc(size);
+}
+
+MEM_STATIC void* ZSTD_customCalloc(size_t size, ZSTD_customMem customMem)
+{
+    if (customMem.customAlloc) {
+        /* calloc implemented as malloc+memset;
+         * not as efficient as calloc, but next best guess for custom malloc */
+        void* const ptr = customMem.customAlloc(customMem.opaque, size);
+        ZSTD_memset(ptr, 0, size);
+        return ptr;
+    }
+    return ZSTD_calloc(1, size);
+}
+
+MEM_STATIC void ZSTD_customFree(void* ptr, ZSTD_customMem customMem)
+{
+    if (ptr!=NULL) {
+        if (customMem.customFree)
+            customMem.customFree(customMem.opaque, ptr);
+        else
+            ZSTD_free(ptr);
+    }
+}
+
+#endif /* ZSTD_ALLOCATIONS_H */
@@ -17,7 +17,7 @@ MEM_STATIC unsigned ZSTD_countTrailingZeros32_fallback(U32 val)
 {
    assert(val != 0);
    {
-        static const int DeBruijnBytePos[32] = {0, 1, 28, 2, 29, 14, 24, 3,
+        static const U32 DeBruijnBytePos[32] = {0, 1, 28, 2, 29, 14, 24, 3,
                                                30, 22, 20, 15, 25, 17, 4, 8,
                                                31, 27, 13, 23, 21, 19, 16, 7,
                                                26, 12, 18, 6, 11, 5, 10, 9};
@@ -30,7 +30,7 @@ MEM_STATIC unsigned ZSTD_countTrailingZeros32(U32 val)
    assert(val != 0);
 #   if defined(_MSC_VER)
 #       if STATIC_BMI2 == 1
-            return _tzcnt_u32(val);
+            return (unsigned)_tzcnt_u32(val);
 #       else
            if (val != 0) {
                unsigned long r;
@@ -69,7 +69,7 @@ MEM_STATIC unsigned ZSTD_countLeadingZeros32(U32 val)
    assert(val != 0);
 #   if defined(_MSC_VER)
 #       if STATIC_BMI2 == 1
-            return _lzcnt_u32(val);
+            return (unsigned)_lzcnt_u32(val);
 #       else
            if (val != 0) {
                unsigned long r;
@@ -92,7 +92,7 @@ MEM_STATIC unsigned ZSTD_countTrailingZeros64(U64 val)
    assert(val != 0);
 #   if defined(_MSC_VER) && defined(_WIN64)
 #       if STATIC_BMI2 == 1
-            return _tzcnt_u64(val);
+            return (unsigned)_tzcnt_u64(val);
 #       else
            if (val != 0) {
                unsigned long r;
@@ -123,7 +123,7 @@ MEM_STATIC unsigned ZSTD_countLeadingZeros64(U64 val)
    assert(val != 0);
 #   if defined(_MSC_VER) && defined(_WIN64)
 #       if STATIC_BMI2 == 1
-            return _lzcnt_u64(val);
+            return (unsigned)_lzcnt_u64(val);
 #       else
            if (val != 0) {
                unsigned long r;
@@ -172,4 +172,29 @@ MEM_STATIC unsigned ZSTD_highbit32(U32 val)   /* compress, dictBuilder, decodeCo
    return 31 - ZSTD_countLeadingZeros32(val);
 }

+/* ZSTD_rotateRight_*():
+ * Rotates a bitfield to the right by "count" bits.
+ * https://en.wikipedia.org/w/index.php?title=Circular_shift&oldid=991635599#Implementing_circular_shifts
+ */
+MEM_STATIC
+U64 ZSTD_rotateRight_U64(U64 const value, U32 count) {
+    assert(count < 64);
+    count &= 0x3F; /* for fickle pattern recognition */
+    return (value >> count) | (U64)(value << ((0U - count) & 0x3F));
+}
+
+MEM_STATIC
+U32 ZSTD_rotateRight_U32(U32 const value, U32 count) {
+    assert(count < 32);
+    count &= 0x1F; /* for fickle pattern recognition */
+    return (value >> count) | (U32)(value << ((0U - count) & 0x1F));
+}
+
+MEM_STATIC
+U16 ZSTD_rotateRight_U16(U16 const value, U32 count) {
+    assert(count < 16);
+    count &= 0x0F; /* for fickle pattern recognition */
+    return (value >> count) | (U16)(value << ((0U - count) & 0x0F));
+}
+
 #endif /* ZSTD_BITS_H */
@@ -396,7 +396,7 @@ MEM_STATIC BIT_DStream_status BIT_reloadDStreamFast(BIT_DStream_t* bitD)
 *  This function is safe, it guarantees it will not read beyond src buffer.
 * @return : status of `BIT_DStream_t` internal register.
 *           when status == BIT_DStream_unfinished, internal register is filled with at least 25 or 57 bits */
-MEM_STATIC BIT_DStream_status BIT_reloadDStream(BIT_DStream_t* bitD)
+MEM_STATIC FORCE_INLINE_ATTR BIT_DStream_status BIT_reloadDStream(BIT_DStream_t* bitD)
 {
    if (bitD->bitsConsumed > (sizeof(bitD->bitContainer)*8))  /* overflow detected, like end of stream */
        return BIT_DStream_overflow;
@@ -281,7 +281,18 @@
 *  Sanitizer
 *****************************************************************/

-#if ZSTD_MEMORY_SANITIZER
+/* Issue #3240 reports an ASAN failure on an llvm-mingw build. Out of an
+ * abundance of caution, disable our custom poisoning on mingw. */
+#ifdef __MINGW32__
+#ifndef ZSTD_ASAN_DONT_POISON_WORKSPACE
+#define ZSTD_ASAN_DONT_POISON_WORKSPACE 1
+#endif
+#ifndef ZSTD_MSAN_DONT_POISON_WORKSPACE
+#define ZSTD_MSAN_DONT_POISON_WORKSPACE 1
+#endif
+#endif
+
+#if ZSTD_MEMORY_SANITIZER && !defined(ZSTD_MSAN_DONT_POISON_WORKSPACE)
 /* Not all platforms that support msan provide sanitizers/msan_interface.h.
 * We therefore declare the functions we need ourselves, rather than trying to
 * include the header file... */
@@ -300,9 +311,13 @@ void __msan_poison(const volatile void *a, size_t size);
 /* Returns the offset of the first (at least partially) poisoned byte in the
   memory range, or -1 if the whole range is good. */
 intptr_t __msan_test_shadow(const volatile void *x, size_t size);
+
+/* Print shadow and origin for the memory range to stderr in a human-readable
+   format. */
+void __msan_print_shadow(const volatile void *x, size_t size);
 #endif

-#if ZSTD_ADDRESS_SANITIZER
+#if ZSTD_ADDRESS_SANITIZER && !defined(ZSTD_ASAN_DONT_POISON_WORKSPACE)
 /* Not all platforms that support asan provide sanitizers/asan_interface.h.
 * We therefore declare the functions we need ourselves, rather than trying to
 * include the header file... */
@@ -19,7 +19,6 @@
 #include "error_private.h"       /* ERR_*, ERROR */
 #define FSE_STATIC_LINKING_ONLY  /* FSE_MIN_TABLELOG */
 #include "fse.h"
-#define HUF_STATIC_LINKING_ONLY  /* HUF_TABLELOG_ABSOLUTEMAX */
 #include "huf.h"
 #include "bits.h"                /* ZSDT_highbit32, ZSTD_countTrailingZeros32 */

@@ -237,7 +236,7 @@ size_t HUF_readStats(BYTE* huffWeight, size_t hwSize, U32* rankStats,
                     const void* src, size_t srcSize)
 {
    U32 wksp[HUF_READ_STATS_WORKSPACE_SIZE_U32];
-    return HUF_readStats_wksp(huffWeight, hwSize, rankStats, nbSymbolsPtr, tableLogPtr, src, srcSize, wksp, sizeof(wksp), /* bmi2 */ 0);
+    return HUF_readStats_wksp(huffWeight, hwSize, rankStats, nbSymbolsPtr, tableLogPtr, src, srcSize, wksp, sizeof(wksp), /* flags */ 0);
 }

 FORCE_INLINE_TEMPLATE size_t
@@ -329,13 +328,13 @@ size_t HUF_readStats_wksp(BYTE* huffWeight, size_t hwSize, U32* rankStats,
                     U32* nbSymbolsPtr, U32* tableLogPtr,
                     const void* src, size_t srcSize,
                     void* workSpace, size_t wkspSize,
-                     int bmi2)
+                     int flags)
 {
 #if DYNAMIC_BMI2
-    if (bmi2) {
+    if (flags & HUF_flags_bmi2) {
        return HUF_readStats_body_bmi2(huffWeight, hwSize, rankStats, nbSymbolsPtr, tableLogPtr, src, srcSize, workSpace, wkspSize);
    }
 #endif
-    (void)bmi2;
+    (void)flags;
    return HUF_readStats_body_default(huffWeight, hwSize, rankStats, nbSymbolsPtr, tableLogPtr, src, srcSize, workSpace, wkspSize);
 }
@@ -47,12 +47,15 @@ const char* ERR_getErrorString(ERR_enum code)
    case PREFIX(dstSize_tooSmall): return "Destination buffer is too small";
    case PREFIX(srcSize_wrong): return "Src size is incorrect";
    case PREFIX(dstBuffer_null): return "Operation on NULL destination buffer";
+    case PREFIX(noForwardProgress_destFull): return "Operation made no progress over multiple calls, due to output buffer being full";
+    case PREFIX(noForwardProgress_inputEmpty): return "Operation made no progress over multiple calls, due to input being empty";
        /* following error codes are not stable and may be removed or changed in a future version */
    case PREFIX(frameIndex_tooLarge): return "Frame index is too large";
    case PREFIX(seekableIO): return "An I/O error occurred when reading/seeking";
    case PREFIX(dstBuffer_wrong): return "Destination buffer is wrong";
    case PREFIX(srcBuffer_wrong): return "Source buffer is wrong";
-    case PREFIX(externalMatchFinder_failed): return "External matchfinder returned an error code";
+    case PREFIX(sequenceProducer_failed): return "Block-level external sequence producer returned an error code";
+    case PREFIX(externalSequences_invalid): return "External sequences are not valid";
    case PREFIX(maxCode):
    default: return notErrorCode;
    }
@@ -53,34 +53,6 @@ extern "C" {
 FSE_PUBLIC_API unsigned FSE_versionNumber(void);   /**< library version number; to be used when checking dll version */


-/*-****************************************
-*  FSE simple functions
-******************************************/
-/*! FSE_compress() :
-    Compress content of buffer 'src', of size 'srcSize', into destination buffer 'dst'.
-    'dst' buffer must be already allocated. Compression runs faster is dstCapacity >= FSE_compressBound(srcSize).
-    @return : size of compressed data (<= dstCapacity).
-    Special values : if return == 0, srcData is not compressible => Nothing is stored within dst !!!
-                     if return == 1, srcData is a single byte symbol * srcSize times. Use RLE compression instead.
-                     if FSE_isError(return), compression failed (more details using FSE_getErrorName())
-*/
-FSE_PUBLIC_API size_t FSE_compress(void* dst, size_t dstCapacity,
-                             const void* src, size_t srcSize);
-
-/*! FSE_decompress():
-    Decompress FSE data from buffer 'cSrc', of size 'cSrcSize',
-    into already allocated destination buffer 'dst', of size 'dstCapacity'.
-    @return : size of regenerated data (<= maxDstSize),
-              or an error code, which can be tested using FSE_isError() .
-
-    ** Important ** : FSE_decompress() does not decompress non-compressible nor RLE data !!!
-    Why ? : making this distinction requires a header.
-    Header management is intentionally delegated to the user layer, which can better manage special cases.
-*/
-FSE_PUBLIC_API size_t FSE_decompress(void* dst,  size_t dstCapacity,
-                               const void* cSrc, size_t cSrcSize);
-
-
 /*-*****************************************
 *  Tool functions
 ******************************************/
@@ -91,20 +63,6 @@ FSE_PUBLIC_API unsigned    FSE_isError(size_t code);        /* tells if a return
 FSE_PUBLIC_API const char* FSE_getErrorName(size_t code);   /* provides error code string (useful for debugging) */


-/*-*****************************************
-*  FSE advanced functions
-******************************************/
-/*! FSE_compress2() :
-    Same as FSE_compress(), but allows the selection of 'maxSymbolValue' and 'tableLog'
-    Both parameters can be defined as '0' to mean : use default value
-    @return : size of compressed data
-    Special values : if return == 0, srcData is not compressible => Nothing is stored within cSrc !!!
-                     if return == 1, srcData is a single byte symbol * srcSize times. Use RLE compression.
-                     if FSE_isError(return), it's an error code.
-*/
-FSE_PUBLIC_API size_t FSE_compress2 (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog);
-
-
 /*-*****************************************
 *  FSE detailed API
 ******************************************/
@@ -164,8 +122,6 @@ FSE_PUBLIC_API size_t FSE_writeNCount (void* buffer, size_t bufferSize,
 /*! Constructor and Destructor of FSE_CTable.
    Note that FSE_CTable size depends on 'tableLog' and 'maxSymbolValue' */
 typedef unsigned FSE_CTable;   /* don't allocate that. It's only meant to be more restrictive than void* */
-FSE_PUBLIC_API FSE_CTable* FSE_createCTable (unsigned maxSymbolValue, unsigned tableLog);
-FSE_PUBLIC_API void        FSE_freeCTable (FSE_CTable* ct);

 /*! FSE_buildCTable():
    Builds `ct`, which must be already allocated, using FSE_createCTable().
@@ -241,23 +197,7 @@ FSE_PUBLIC_API size_t FSE_readNCount_bmi2(short* normalizedCounter,
                           unsigned* maxSymbolValuePtr, unsigned* tableLogPtr,
                           const void* rBuffer, size_t rBuffSize, int bmi2);

-/*! Constructor and Destructor of FSE_DTable.
-    Note that its size depends on 'tableLog' */
 typedef unsigned FSE_DTable;   /* don't allocate that. It's just a way to be more restrictive than void* */
-FSE_PUBLIC_API FSE_DTable* FSE_createDTable(unsigned tableLog);
-FSE_PUBLIC_API void        FSE_freeDTable(FSE_DTable* dt);
-
-/*! FSE_buildDTable():
-    Builds 'dt', which must be already allocated, using FSE_createDTable().
-    return : 0, or an errorCode, which can be tested using FSE_isError() */
-FSE_PUBLIC_API size_t FSE_buildDTable (FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog);
-
-/*! FSE_decompress_usingDTable():
-    Decompress compressed source `cSrc` of size `cSrcSize` using `dt`
-    into `dst` which must be already allocated.
-    @return : size of regenerated data (necessarily <= `dstCapacity`),
-              or an errorCode, which can be tested using FSE_isError() */
-FSE_PUBLIC_API size_t FSE_decompress_usingDTable(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, const FSE_DTable* dt);

 /*!
 Tutorial :
@@ -320,16 +260,6 @@ If there is an error, the function will return an error code, which can be teste
 unsigned FSE_optimalTableLog_internal(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue, unsigned minus);
 /**< same as FSE_optimalTableLog(), which used `minus==2` */

-/* FSE_compress_wksp() :
- * Same as FSE_compress2(), but using an externally allocated scratch buffer (`workSpace`).
- * FSE_COMPRESS_WKSP_SIZE_U32() provides the minimum size required for `workSpace` as a table of FSE_CTable.
- */
-#define FSE_COMPRESS_WKSP_SIZE_U32(maxTableLog, maxSymbolValue)   ( FSE_CTABLE_SIZE_U32(maxTableLog, maxSymbolValue) + ((maxTableLog > 12) ? (1 << (maxTableLog - 2)) : 1024) )
-size_t FSE_compress_wksp (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize);
-
-size_t FSE_buildCTable_raw (FSE_CTable* ct, unsigned nbBits);
-/**< build a fake FSE_CTable, designed for a flat distribution, where each symbol uses nbBits */
-
 size_t FSE_buildCTable_rle (FSE_CTable* ct, unsigned char symbolValue);
 /**< build a fake FSE_CTable, designed to compress always the same symbolValue */

@@ -347,19 +277,11 @@ size_t FSE_buildCTable_wksp(FSE_CTable* ct, const short* normalizedCounter, unsi
 FSE_PUBLIC_API size_t FSE_buildDTable_wksp(FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize);
 /**< Same as FSE_buildDTable(), using an externally allocated `workspace` produced with `FSE_BUILD_DTABLE_WKSP_SIZE_U32(maxSymbolValue)` */

-size_t FSE_buildDTable_raw (FSE_DTable* dt, unsigned nbBits);
-/**< build a fake FSE_DTable, designed to read a flat distribution where each symbol uses nbBits */
-
-size_t FSE_buildDTable_rle (FSE_DTable* dt, unsigned char symbolValue);
-/**< build a fake FSE_DTable, designed to always generate the same symbolValue */
-
 #define FSE_DECOMPRESS_WKSP_SIZE_U32(maxTableLog, maxSymbolValue) (FSE_DTABLE_SIZE_U32(maxTableLog) + 1 + FSE_BUILD_DTABLE_WKSP_SIZE_U32(maxTableLog, maxSymbolValue) + (FSE_MAX_SYMBOL_VALUE + 1) / 2 + 1)
 #define FSE_DECOMPRESS_WKSP_SIZE(maxTableLog, maxSymbolValue) (FSE_DECOMPRESS_WKSP_SIZE_U32(maxTableLog, maxSymbolValue) * sizeof(unsigned))
-size_t FSE_decompress_wksp(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize);
-/**< same as FSE_decompress(), using an externally allocated `workSpace` produced with `FSE_DECOMPRESS_WKSP_SIZE_U32(maxLog, maxSymbolValue)` */
-
 size_t FSE_decompress_wksp_bmi2(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize, int bmi2);
-/**< Same as FSE_decompress_wksp() but with dynamic BMI2 support. Pass 1 if your CPU supports BMI2 or 0 if it doesn't. */
+/**< same as FSE_decompress(), using an externally allocated `workSpace` produced with `FSE_DECOMPRESS_WKSP_SIZE_U32(maxLog, maxSymbolValue)`.
+ * Set bmi2 to 1 if your CPU supports BMI2 or 0 if it doesn't */

 typedef enum {
   FSE_repeat_none,  /**< Cannot use the previous table */
@@ -56,19 +56,6 @@
 #define FSE_FUNCTION_NAME(X,Y) FSE_CAT(X,Y)
 #define FSE_TYPE_NAME(X,Y) FSE_CAT(X,Y)

-
-/* Function templates */
-FSE_DTable* FSE_createDTable (unsigned tableLog)
-{
-    if (tableLog > FSE_TABLELOG_ABSOLUTE_MAX) tableLog = FSE_TABLELOG_ABSOLUTE_MAX;
-    return (FSE_DTable*)ZSTD_malloc( FSE_DTABLE_SIZE_U32(tableLog) * sizeof (U32) );
-}
-
-void FSE_freeDTable (FSE_DTable* dt)
-{
-    ZSTD_free(dt);
-}
-
 static size_t FSE_buildDTable_internal(FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize)
 {
    void* const tdPtr = dt+1;   /* because *dt is unsigned, 32-bits aligned on 32-bits */
@@ -185,49 +172,6 @@ size_t FSE_buildDTable_wksp(FSE_DTable* dt, const short* normalizedCounter, unsi
 /*-*******************************************************
 *  Decompression (Byte symbols)
 *********************************************************/
-size_t FSE_buildDTable_rle (FSE_DTable* dt, BYTE symbolValue)
-{
-    void* ptr = dt;
-    FSE_DTableHeader* const DTableH = (FSE_DTableHeader*)ptr;
-    void* dPtr = dt + 1;
-    FSE_decode_t* const cell = (FSE_decode_t*)dPtr;
-
-    DTableH->tableLog = 0;
-    DTableH->fastMode = 0;
-
-    cell->newState = 0;
-    cell->symbol = symbolValue;
-    cell->nbBits = 0;
-
-    return 0;
-}
-
-
-size_t FSE_buildDTable_raw (FSE_DTable* dt, unsigned nbBits)
-{
-    void* ptr = dt;
-    FSE_DTableHeader* const DTableH = (FSE_DTableHeader*)ptr;
-    void* dPtr = dt + 1;
-    FSE_decode_t* const dinfo = (FSE_decode_t*)dPtr;
-    const unsigned tableSize = 1 << nbBits;
-    const unsigned tableMask = tableSize - 1;
-    const unsigned maxSV1 = tableMask+1;
-    unsigned s;
-
-    /* Sanity checks */
-    if (nbBits < 1) return ERROR(GENERIC);         /* min size */
-
-    /* Build Decoding Table */
-    DTableH->tableLog = (U16)nbBits;
-    DTableH->fastMode = 1;
-    for (s=0; s<maxSV1; s++) {
-        dinfo[s].newState = 0;
-        dinfo[s].symbol = (BYTE)s;
-        dinfo[s].nbBits = (BYTE)nbBits;
-    }
-
-    return 0;
-}

 FORCE_INLINE_TEMPLATE size_t FSE_decompress_usingDTable_generic(
          void* dst, size_t maxDstSize,
@@ -291,26 +235,6 @@ FORCE_INLINE_TEMPLATE size_t FSE_decompress_usingDTable_generic(
    return op-ostart;
 }

-
-size_t FSE_decompress_usingDTable(void* dst, size_t originalSize,
-                            const void* cSrc, size_t cSrcSize,
-                            const FSE_DTable* dt)
-{
-    const void* ptr = dt;
-    const FSE_DTableHeader* DTableH = (const FSE_DTableHeader*)ptr;
-    const U32 fastMode = DTableH->fastMode;
-
-    /* select fast mode (static) */
-    if (fastMode) return FSE_decompress_usingDTable_generic(dst, originalSize, cSrc, cSrcSize, dt, 1);
-    return FSE_decompress_usingDTable_generic(dst, originalSize, cSrc, cSrcSize, dt, 0);
-}
-
-
-size_t FSE_decompress_wksp(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, unsigned maxLog, void* workSpace, size_t wkspSize)
-{
-    return FSE_decompress_wksp_bmi2(dst, dstCapacity, cSrc, cSrcSize, maxLog, workSpace, wkspSize, /* bmi2 */ 0);
-}
-
 typedef struct {
    short ncount[FSE_MAX_SYMBOL_VALUE + 1];
    FSE_DTable dtable[1]; /* Dynamically sized */
@@ -384,22 +308,4 @@ size_t FSE_decompress_wksp_bmi2(void* dst, size_t dstCapacity, const void* cSrc,
    return FSE_decompress_wksp_body_default(dst, dstCapacity, cSrc, cSrcSize, maxLog, workSpace, wkspSize);
 }

-
-typedef FSE_DTable DTable_max_t[FSE_DTABLE_SIZE_U32(FSE_MAX_TABLELOG)];
-
-#ifndef ZSTD_NO_UNUSED_FUNCTIONS
-size_t FSE_buildDTable(FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog) {
-    U32 wksp[FSE_BUILD_DTABLE_WKSP_SIZE_U32(FSE_TABLELOG_ABSOLUTE_MAX, FSE_MAX_SYMBOL_VALUE)];
-    return FSE_buildDTable_wksp(dt, normalizedCounter, maxSymbolValue, tableLog, wksp, sizeof(wksp));
-}
-
-size_t FSE_decompress(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize)
-{
-    /* Static analyzer seems unable to understand this table will be properly initialized later */
-    U32 wksp[FSE_DECOMPRESS_WKSP_SIZE_U32(FSE_MAX_TABLELOG, FSE_MAX_SYMBOL_VALUE)];
-    return FSE_decompress_wksp(dst, dstCapacity, cSrc, cSrcSize, FSE_MAX_TABLELOG, wksp, sizeof(wksp));
-}
-#endif
-
-
 #endif   /* FSE_COMMONDEFS_ONLY */
@@ -21,102 +21,23 @@ extern "C" {

 /* *** Dependencies *** */
 #include "zstd_deps.h"    /* size_t */
-
-
-/* *** library symbols visibility *** */
-/* Note : when linking with -fvisibility=hidden on gcc, or by default on Visual,
- *        HUF symbols remain "private" (internal symbols for library only).
- *        Set macro FSE_DLL_EXPORT to 1 if you want HUF symbols visible on DLL interface */
-#if defined(FSE_DLL_EXPORT) && (FSE_DLL_EXPORT==1) && defined(__GNUC__) && (__GNUC__ >= 4)
-#  define HUF_PUBLIC_API __attribute__ ((visibility ("default")))
-#elif defined(FSE_DLL_EXPORT) && (FSE_DLL_EXPORT==1)   /* Visual expected */
-#  define HUF_PUBLIC_API __declspec(dllexport)
-#elif defined(FSE_DLL_IMPORT) && (FSE_DLL_IMPORT==1)
-#  define HUF_PUBLIC_API __declspec(dllimport)  /* not required, just to generate faster code (saves a function pointer load from IAT and an indirect jump) */
-#else
-#  define HUF_PUBLIC_API
-#endif
-
-
-/* ========================== */
-/* ***  simple functions  *** */
-/* ========================== */
-
-/** HUF_compress() :
- *  Compress content from buffer 'src', of size 'srcSize', into buffer 'dst'.
- * 'dst' buffer must be already allocated.
- *  Compression runs faster if `dstCapacity` >= HUF_compressBound(srcSize).
- * `srcSize` must be <= `HUF_BLOCKSIZE_MAX` == 128 KB.
- * @return : size of compressed data (<= `dstCapacity`).
- *  Special values : if return == 0, srcData is not compressible => Nothing is stored within dst !!!
- *                   if HUF_isError(return), compression failed (more details using HUF_getErrorName())
- */
-HUF_PUBLIC_API size_t HUF_compress(void* dst, size_t dstCapacity,
-                             const void* src, size_t srcSize);
-
-/** HUF_decompress() :
- *  Decompress HUF data from buffer 'cSrc', of size 'cSrcSize',
- *  into already allocated buffer 'dst', of minimum size 'dstSize'.
- * `originalSize` : **must** be the ***exact*** size of original (uncompressed) data.
- *  Note : in contrast with FSE, HUF_decompress can regenerate
- *         RLE (cSrcSize==1) and uncompressed (cSrcSize==dstSize) data,
- *         because it knows size to regenerate (originalSize).
- * @return : size of regenerated data (== originalSize),
- *           or an error code, which can be tested using HUF_isError()
- */
-HUF_PUBLIC_API size_t HUF_decompress(void* dst,  size_t originalSize,
-                               const void* cSrc, size_t cSrcSize);
-
-
-/* ***   Tool functions *** */
-#define HUF_BLOCKSIZE_MAX (128 * 1024)                  /**< maximum input size for a single block compressed with HUF_compress */
-HUF_PUBLIC_API size_t HUF_compressBound(size_t size);   /**< maximum compressed size (worst case) */
-
-/* Error Management */
-HUF_PUBLIC_API unsigned    HUF_isError(size_t code);       /**< tells if a return value is an error code */
-HUF_PUBLIC_API const char* HUF_getErrorName(size_t code);  /**< provides error code string (useful for debugging) */
-
-
-/* ***   Advanced function   *** */
-
-/** HUF_compress2() :
- *  Same as HUF_compress(), but offers control over `maxSymbolValue` and `tableLog`.
- * `maxSymbolValue` must be <= HUF_SYMBOLVALUE_MAX .
- * `tableLog` must be `<= HUF_TABLELOG_MAX` . */
-HUF_PUBLIC_API size_t HUF_compress2 (void* dst, size_t dstCapacity,
-                               const void* src, size_t srcSize,
-                               unsigned maxSymbolValue, unsigned tableLog);
-
-/** HUF_compress4X_wksp() :
- *  Same as HUF_compress2(), but uses externally allocated @workSpace.
- * @workSpace's size, aka @wkspSize, must be >= HUF_WORKSPACE_SIZE
- * @srcSize must be >= 6
- */
-#define HUF_WORKSPACE_SIZE ((8 << 10) + 512 /* sorting scratch space */)
-#define HUF_WORKSPACE_SIZE_U64 (HUF_WORKSPACE_SIZE / sizeof(U64))
-HUF_PUBLIC_API size_t HUF_compress4X_wksp (void* dst, size_t dstCapacity,
-                                     const void* src, size_t srcSize,
-                                     unsigned maxSymbolValue, unsigned tableLog,
-                                     void* workSpace, size_t wkspSize);
-
-#endif   /* HUF_H_298734234 */
-
-/* ******************************************************************
- *  WARNING !!
- *  The following section contains advanced and experimental definitions
- *  which shall never be used in the context of a dynamic library,
- *  because they are not guaranteed to remain stable in the future.
- *  Only consider them in association with static linking.
- * *****************************************************************/
-#if defined(HUF_STATIC_LINKING_ONLY) && !defined(HUF_H_HUF_STATIC_LINKING_ONLY)
-#define HUF_H_HUF_STATIC_LINKING_ONLY
-
-/* *** Dependencies *** */
-#include "mem.h"   /* U32 */
+#include "mem.h"          /* U32 */
 #define FSE_STATIC_LINKING_ONLY
 #include "fse.h"


+/* ***   Tool functions *** */
+#define HUF_BLOCKSIZE_MAX (128 * 1024)   /**< maximum input size for a single block compressed with HUF_compress */
+size_t HUF_compressBound(size_t size);   /**< maximum compressed size (worst case) */
+
+/* Error Management */
+unsigned    HUF_isError(size_t code);       /**< tells if a return value is an error code */
+const char* HUF_getErrorName(size_t code);  /**< provides error code string (useful for debugging) */
+
+
+#define HUF_WORKSPACE_SIZE ((8 << 10) + 512 /* sorting scratch space */)
+#define HUF_WORKSPACE_SIZE_U64 (HUF_WORKSPACE_SIZE / sizeof(U64))
+
 /* *** Constants *** */
 #define HUF_TABLELOG_MAX      12      /* max runtime value of tableLog (due to static allocation); can be modified up to HUF_TABLELOG_ABSOLUTEMAX */
 #define HUF_TABLELOG_DEFAULT  11      /* default tableLog value when none specified */
@@ -156,30 +77,49 @@ typedef U32 HUF_DTable;
 /* ****************************************
 *  Advanced decompression functions
 ******************************************/
-size_t HUF_decompress4X1 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /**< single-symbol decoder */
-#ifndef HUF_FORCE_DECOMPRESS_X1
-size_t HUF_decompress4X2 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /**< double-symbols decoder */
-#endif

-size_t HUF_decompress4X_DCtx (HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /**< decodes RLE and uncompressed */
-size_t HUF_decompress4X_hufOnly(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /**< considers RLE and uncompressed as errors */
-size_t HUF_decompress4X_hufOnly_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize); /**< considers RLE and uncompressed as errors */
-size_t HUF_decompress4X1_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /**< single-symbol decoder */
-size_t HUF_decompress4X1_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize);   /**< single-symbol decoder */
-#ifndef HUF_FORCE_DECOMPRESS_X1
-size_t HUF_decompress4X2_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /**< double-symbols decoder */
-size_t HUF_decompress4X2_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize);   /**< double-symbols decoder */
-#endif
+/**
+ * Huffman flags bitset.
+ * For all flags, 0 is the default value.
+ */
+typedef enum {
+    /**
+     * If compiled with DYNAMIC_BMI2: Set flag only if the CPU supports BMI2 at runtime.
+     * Otherwise: Ignored.
+     */
+    HUF_flags_bmi2 = (1 << 0),
+    /**
+     * If set: Test possible table depths to find the one that produces the smallest header + encoded size.
+     * If unset: Use heuristic to find the table depth.
+     */
+    HUF_flags_optimalDepth = (1 << 1),
+    /**
+     * If set: If the previous table can encode the input, always reuse the previous table.
+     * If unset: If the previous table can encode the input, reuse the previous table if it results in a smaller output.
+     */
+    HUF_flags_preferRepeat = (1 << 2),
+    /**
+     * If set: Sample the input and check if the sample is uncompressible, if it is then don't attempt to compress.
+     * If unset: Always histogram the entire input.
+     */
+    HUF_flags_suspectUncompressible = (1 << 3),
+    /**
+     * If set: Don't use assembly implementations
+     * If unset: Allow using assembly implementations
+     */
+    HUF_flags_disableAsm = (1 << 4),
+    /**
+     * If set: Don't use the fast decoding loop, always use the fallback decoding loop.
+     * If unset: Use the fast decoding loop when possible.
+     */
+    HUF_flags_disableFast = (1 << 5)
+} HUF_flags_e;


 /* ****************************************
 *  HUF detailed API
 * ****************************************/
 #define HUF_OPTIMAL_DEPTH_THRESHOLD ZSTD_btultra
-typedef enum {
-   HUF_depth_fast, /** Use heuristic to find the table depth**/
-   HUF_depth_optimal /** Test possible table depths to find the one that produces the smallest header + encoded size**/
- } HUF_depth_mode;

 /*! HUF_compress() does the following:
 *  1. count symbol occurrence from source[] into table count[] using FSE_count() (exposed within "fse.h")
@@ -195,12 +135,9 @@ typedef enum {
 unsigned HUF_minTableLog(unsigned symbolCardinality);
 unsigned HUF_cardinality(const unsigned* count, unsigned maxSymbolValue);
 unsigned HUF_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue, void* workSpace,
- size_t wkspSize, HUF_CElt* table, const unsigned* count, HUF_depth_mode depthMode); /* table is used as scratch space for building and testing tables, not a return value */
-size_t HUF_buildCTable (HUF_CElt* CTable, const unsigned* count, unsigned maxSymbolValue, unsigned maxNbBits);   /* @return : maxNbBits; CTable and count can overlap. In which case, CTable will overwrite count content */
-size_t HUF_writeCTable (void* dst, size_t maxDstSize, const HUF_CElt* CTable, unsigned maxSymbolValue, unsigned huffLog);
+ size_t wkspSize, HUF_CElt* table, const unsigned* count, int flags); /* table is used as scratch space for building and testing tables, not a return value */
 size_t HUF_writeCTable_wksp(void* dst, size_t maxDstSize, const HUF_CElt* CTable, unsigned maxSymbolValue, unsigned huffLog, void* workspace, size_t workspaceSize);
-size_t HUF_compress4X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable);
-size_t HUF_compress4X_usingCTable_bmi2(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable, int bmi2);
+size_t HUF_compress4X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable, int flags);
 size_t HUF_estimateCompressedSize(const HUF_CElt* CTable, const unsigned* count, unsigned maxSymbolValue);
 int HUF_validateCTable(const HUF_CElt* CTable, const unsigned* count, unsigned maxSymbolValue);

@@ -220,14 +157,13 @@ size_t HUF_compress4X_repeat(void* dst, size_t dstSize,
                       const void* src, size_t srcSize,
                       unsigned maxSymbolValue, unsigned tableLog,
                       void* workSpace, size_t wkspSize,    /**< `workSpace` must be aligned on 4-bytes boundaries, `wkspSize` must be >= HUF_WORKSPACE_SIZE */
-                       HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2,
-                       unsigned suspectUncompressible, HUF_depth_mode depthMode);
+                       HUF_CElt* hufTable, HUF_repeat* repeat, int flags);

 /** HUF_buildCTable_wksp() :
 *  Same as HUF_buildCTable(), but using externally allocated scratch buffer.
 * `workSpace` must be aligned on 4-bytes boundaries, and its size must be >= HUF_CTABLE_WORKSPACE_SIZE.
 */
-#define HUF_CTABLE_WORKSPACE_SIZE_U32 (2*HUF_SYMBOLVALUE_MAX +1 +1)
+#define HUF_CTABLE_WORKSPACE_SIZE_U32 ((4 * (HUF_SYMBOLVALUE_MAX + 1)) + 192)
 #define HUF_CTABLE_WORKSPACE_SIZE (HUF_CTABLE_WORKSPACE_SIZE_U32 * sizeof(unsigned))
 size_t HUF_buildCTable_wksp (HUF_CElt* tree,
                       const unsigned* count, U32 maxSymbolValue, U32 maxNbBits,
@@ -253,7 +189,7 @@ size_t HUF_readStats_wksp(BYTE* huffWeight, size_t hwSize,
                          U32* rankStats, U32* nbSymbolsPtr, U32* tableLogPtr,
                          const void* src, size_t srcSize,
                          void* workspace, size_t wkspSize,
-                          int bmi2);
+                          int flags);

 /** HUF_readCTable() :
 *  Loading a CTable saved with HUF_writeCTable() */
@@ -291,32 +227,12 @@ U32 HUF_selectDecoder (size_t dstSize, size_t cSrcSize);
 #define HUF_DECOMPRESS_WORKSPACE_SIZE ((2 << 10) + (1 << 9))
 #define HUF_DECOMPRESS_WORKSPACE_SIZE_U32 (HUF_DECOMPRESS_WORKSPACE_SIZE / sizeof(U32))

-#ifndef HUF_FORCE_DECOMPRESS_X2
-size_t HUF_readDTableX1 (HUF_DTable* DTable, const void* src, size_t srcSize);
-size_t HUF_readDTableX1_wksp (HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize);
-#endif
-#ifndef HUF_FORCE_DECOMPRESS_X1
-size_t HUF_readDTableX2 (HUF_DTable* DTable, const void* src, size_t srcSize);
-size_t HUF_readDTableX2_wksp (HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize);
-#endif
-
-size_t HUF_decompress4X_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
-#ifndef HUF_FORCE_DECOMPRESS_X2
-size_t HUF_decompress4X1_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
-#endif
-#ifndef HUF_FORCE_DECOMPRESS_X1
-size_t HUF_decompress4X2_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
-#endif
-

 /* ====================== */
 /* single stream variants */
 /* ====================== */

-size_t HUF_compress1X (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog);
-size_t HUF_compress1X_wksp (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize);  /**< `workSpace` must be a table of at least HUF_WORKSPACE_SIZE_U64 U64 */
-size_t HUF_compress1X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable);
-size_t HUF_compress1X_usingCTable_bmi2(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable, int bmi2);
+size_t HUF_compress1X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable, int flags);
 /** HUF_compress1X_repeat() :
 *  Same as HUF_compress1X_wksp(), but considers using hufTable if *repeat != HUF_repeat_none.
 *  If it uses hufTable it does not modify hufTable or repeat.
@@ -327,50 +243,30 @@ size_t HUF_compress1X_repeat(void* dst, size_t dstSize,
                       const void* src, size_t srcSize,
                       unsigned maxSymbolValue, unsigned tableLog,
                       void* workSpace, size_t wkspSize,   /**< `workSpace` must be aligned on 4-bytes boundaries, `wkspSize` must be >= HUF_WORKSPACE_SIZE */
-                       HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2,
-                       unsigned suspectUncompressible, HUF_depth_mode depthMode);
+                       HUF_CElt* hufTable, HUF_repeat* repeat, int flags);

-size_t HUF_decompress1X1 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /* single-symbol decoder */
+size_t HUF_decompress1X_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize, int flags);
 #ifndef HUF_FORCE_DECOMPRESS_X1
-size_t HUF_decompress1X2 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /* double-symbol decoder */
-#endif
-
-size_t HUF_decompress1X_DCtx (HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);
-size_t HUF_decompress1X_DCtx_wksp (HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize);
-#ifndef HUF_FORCE_DECOMPRESS_X2
-size_t HUF_decompress1X1_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /**< single-symbol decoder */
-size_t HUF_decompress1X1_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize);   /**< single-symbol decoder */
-#endif
-#ifndef HUF_FORCE_DECOMPRESS_X1
-size_t HUF_decompress1X2_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);   /**< double-symbols decoder */
-size_t HUF_decompress1X2_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize);   /**< double-symbols decoder */
-#endif
-
-size_t HUF_decompress1X_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);   /**< automatic selection of sing or double symbol decoder, based on DTable */
-#ifndef HUF_FORCE_DECOMPRESS_X2
-size_t HUF_decompress1X1_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
-#endif
-#ifndef HUF_FORCE_DECOMPRESS_X1
-size_t HUF_decompress1X2_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
+size_t HUF_decompress1X2_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize, int flags);   /**< double-symbols decoder */
 #endif

 /* BMI2 variants.
 * If the CPU has BMI2 support, pass bmi2=1, otherwise pass bmi2=0.
 */
-size_t HUF_decompress1X_usingDTable_bmi2(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int bmi2);
+size_t HUF_decompress1X_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int flags);
 #ifndef HUF_FORCE_DECOMPRESS_X2
-size_t HUF_decompress1X1_DCtx_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize, int bmi2);
+size_t HUF_decompress1X1_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize, int flags);
 #endif
-size_t HUF_decompress4X_usingDTable_bmi2(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int bmi2);
-size_t HUF_decompress4X_hufOnly_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize, int bmi2);
+size_t HUF_decompress4X_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int flags);
+size_t HUF_decompress4X_hufOnly_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize, int flags);
 #ifndef HUF_FORCE_DECOMPRESS_X2
-size_t HUF_readDTableX1_wksp_bmi2(HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize, int bmi2);
+size_t HUF_readDTableX1_wksp(HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize, int flags);
 #endif
 #ifndef HUF_FORCE_DECOMPRESS_X1
-size_t HUF_readDTableX2_wksp_bmi2(HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize, int bmi2);
+size_t HUF_readDTableX2_wksp(HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize, int flags);
 #endif

-#endif /* HUF_STATIC_LINKING_ONLY */
+#endif   /* HUF_H_298734234 */

 #if defined (__cplusplus)
 }
@@ -10,9 +10,9 @@


 /* ======   Dependencies   ======= */
+#include "../common/allocations.h"  /* ZSTD_customCalloc, ZSTD_customFree */
 #include "zstd_deps.h" /* size_t */
 #include "debug.h"     /* assert */
-#include "zstd_internal.h"  /* ZSTD_customCalloc, ZSTD_customFree */
 #include "pool.h"

 /* ======   Compiler specifics   ====== */
@@ -271,7 +271,9 @@ static int isQueueFull(POOL_ctx const* ctx) {
 static void
 POOL_add_internal(POOL_ctx* ctx, POOL_function function, void *opaque)
 {
-    POOL_job const job = {function, opaque};
+    POOL_job job;
+    job.function = function;
+    job.opaque = opaque;
    assert(ctx != NULL);
    if (ctx->shutdown) return;

@@ -137,12 +137,20 @@
 /*
 * For x86 ELF targets, add .note.gnu.property section for Intel CET in
 * assembly sources when CET is enabled.
+ *
+ * Additionally, any function that may be called indirectly must begin
+ * with ZSTD_CET_ENDBRANCH.
 */
 #if defined(__ELF__) && (defined(__x86_64__) || defined(__i386__)) \
    && defined(__has_include)
 # if __has_include(<cet.h>)
 #  include <cet.h>
+#  define ZSTD_CET_ENDBRANCH _CET_ENDBR
 # endif
 #endif

+#ifndef ZSTD_CET_ENDBRANCH
+# define ZSTD_CET_ENDBRANCH
+#endif
+
 #endif /* ZSTD_PORTABILITY_MACROS_H */
@@ -47,7 +47,7 @@ static unsigned __stdcall worker(void *arg)
    void* (*start_routine)(void*);
    void* thread_arg;

-    /* Inialized thread_arg and start_routine and signal main thread that we don't need it
+    /* Initialized thread_arg and start_routine and signal main thread that we don't need it
     * to wait any longer.
     */
    {
@@ -14,7 +14,6 @@
 *  Dependencies
 ***************************************/
 #define ZSTD_DEPS_NEED_MALLOC
-#include "zstd_deps.h"   /* ZSTD_malloc, ZSTD_calloc, ZSTD_free, ZSTD_memset */
 #include "error_private.h"
 #include "zstd_internal.h"

@@ -47,37 +46,3 @@ ZSTD_ErrorCode ZSTD_getErrorCode(size_t code) { return ERR_getErrorCode(code); }
 /*! ZSTD_getErrorString() :
 *  provides error code string from enum */
 const char* ZSTD_getErrorString(ZSTD_ErrorCode code) { return ERR_getErrorString(code); }
-
-
-
-/*=**************************************************************
-*  Custom allocator
-****************************************************************/
-void* ZSTD_customMalloc(size_t size, ZSTD_customMem customMem)
-{
-    if (customMem.customAlloc)
-        return customMem.customAlloc(customMem.opaque, size);
-    return ZSTD_malloc(size);
-}
-
-void* ZSTD_customCalloc(size_t size, ZSTD_customMem customMem)
-{
-    if (customMem.customAlloc) {
-        /* calloc implemented as malloc+memset;
-         * not as efficient as calloc, but next best guess for custom malloc */
-        void* const ptr = customMem.customAlloc(customMem.opaque, size);
-        ZSTD_memset(ptr, 0, size);
-        return ptr;
-    }
-    return ZSTD_calloc(1, size);
-}
-
-void ZSTD_customFree(void* ptr, ZSTD_customMem customMem)
-{
-    if (ptr!=NULL) {
-        if (customMem.customFree)
-            customMem.customFree(customMem.opaque, ptr);
-        else
-            ZSTD_free(ptr);
-    }
-}
@@ -28,7 +28,6 @@
 #include "../zstd.h"
 #define FSE_STATIC_LINKING_ONLY
 #include "fse.h"
-#define HUF_STATIC_LINKING_ONLY
 #include "huf.h"
 #ifndef XXH_STATIC_LINKING_ONLY
 #  define XXH_STATIC_LINKING_ONLY  /* XXH64_state_t */
@@ -114,6 +113,8 @@ typedef enum { set_basic, set_rle, set_compressed, set_repeat } symbolEncodingTy
 #define LLFSELog    9
 #define OffFSELog   8
 #define MaxFSELog  MAX(MAX(MLFSELog, LLFSELog), OffFSELog)
+#define MaxMLBits 16
+#define MaxLLBits 16

 #define ZSTD_MAX_HUF_HEADER_SIZE 128 /* header + <= 127 byte tree description */
 /* Each table cannot take more than #symbols * FSELog bits */
@@ -341,17 +342,13 @@ MEM_STATIC ZSTD_sequenceLength ZSTD_getSequenceLength(seqStore_t const* seqStore
 *          `decompressedBound != ZSTD_CONTENTSIZE_ERROR`
 */
 typedef struct {
+    size_t nbBlocks;
    size_t compressedSize;
    unsigned long long decompressedBound;
 } ZSTD_frameSizeInfo;   /* decompress & legacy */

 const seqStore_t* ZSTD_getSeqStore(const ZSTD_CCtx* ctx);   /* compress & dictBuilder */
-void ZSTD_seqToCodes(const seqStore_t* seqStorePtr);   /* compress, dictBuilder, decodeCorpus (shouldn't get its definition from here) */
-
-/* custom memory allocation functions */
-void* ZSTD_customMalloc(size_t size, ZSTD_customMem customMem);
-void* ZSTD_customCalloc(size_t size, ZSTD_customMem customMem);
-void ZSTD_customFree(void* ptr, ZSTD_customMem customMem);
+int ZSTD_seqToCodes(const seqStore_t* seqStorePtr);   /* compress, dictBuilder, decodeCorpus (shouldn't get its definition from here) */


 /* ZSTD_invalidateRepCodes() :
@@ -343,16 +343,6 @@ size_t FSE_writeNCount (void* buffer, size_t bufferSize,
 *  FSE Compression Code
 ****************************************************************/

-FSE_CTable* FSE_createCTable (unsigned maxSymbolValue, unsigned tableLog)
-{
-    size_t size;
-    if (tableLog > FSE_TABLELOG_ABSOLUTE_MAX) tableLog = FSE_TABLELOG_ABSOLUTE_MAX;
-    size = FSE_CTABLE_SIZE_U32 (tableLog, maxSymbolValue) * sizeof(U32);
-    return (FSE_CTable*)ZSTD_malloc(size);
-}
-
-void FSE_freeCTable (FSE_CTable* ct) { ZSTD_free(ct); }
-
 /* provides the minimum logSize to safely represent a distribution */
 static unsigned FSE_minTableLog(size_t srcSize, unsigned maxSymbolValue)
 {
@@ -533,40 +523,6 @@ size_t FSE_normalizeCount (short* normalizedCounter, unsigned tableLog,
    return tableLog;
 }

-
-/* fake FSE_CTable, for raw (uncompressed) input */
-size_t FSE_buildCTable_raw (FSE_CTable* ct, unsigned nbBits)
-{
-    const unsigned tableSize = 1 << nbBits;
-    const unsigned tableMask = tableSize - 1;
-    const unsigned maxSymbolValue = tableMask;
-    void* const ptr = ct;
-    U16* const tableU16 = ( (U16*) ptr) + 2;
-    void* const FSCT = ((U32*)ptr) + 1 /* header */ + (tableSize>>1);   /* assumption : tableLog >= 1 */
-    FSE_symbolCompressionTransform* const symbolTT = (FSE_symbolCompressionTransform*) (FSCT);
-    unsigned s;
-
-    /* Sanity checks */
-    if (nbBits < 1) return ERROR(GENERIC);             /* min size */
-
-    /* header */
-    tableU16[-2] = (U16) nbBits;
-    tableU16[-1] = (U16) maxSymbolValue;
-
-    /* Build table */
-    for (s=0; s<tableSize; s++)
-        tableU16[s] = (U16)(tableSize + s);
-
-    /* Build Symbol Transformation Table */
-    {   const U32 deltaNbBits = (nbBits << 16) - (1 << nbBits);
-        for (s=0; s<=maxSymbolValue; s++) {
-            symbolTT[s].deltaNbBits = deltaNbBits;
-            symbolTT[s].deltaFindState = s-1;
-    }   }
-
-    return 0;
-}
-
 /* fake FSE_CTable, for rle input (always same symbol) */
 size_t FSE_buildCTable_rle (FSE_CTable* ct, BYTE symbolValue)
 {
@@ -665,78 +621,4 @@ size_t FSE_compress_usingCTable (void* dst, size_t dstSize,

 size_t FSE_compressBound(size_t size) { return FSE_COMPRESSBOUND(size); }

-#ifndef ZSTD_NO_UNUSED_FUNCTIONS
-/* FSE_compress_wksp() :
- * Same as FSE_compress2(), but using an externally allocated scratch buffer (`workSpace`).
- * `wkspSize` size must be `(1<<tableLog)`.
- */
-size_t FSE_compress_wksp (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize)
-{
-    BYTE* const ostart = (BYTE*) dst;
-    BYTE* op = ostart;
-    BYTE* const oend = ostart + dstSize;
-
-    unsigned count[FSE_MAX_SYMBOL_VALUE+1];
-    S16   norm[FSE_MAX_SYMBOL_VALUE+1];
-    FSE_CTable* CTable = (FSE_CTable*)workSpace;
-    size_t const CTableSize = FSE_CTABLE_SIZE_U32(tableLog, maxSymbolValue);
-    void* scratchBuffer = (void*)(CTable + CTableSize);
-    size_t const scratchBufferSize = wkspSize - (CTableSize * sizeof(FSE_CTable));
-
-    /* init conditions */
-    if (wkspSize < FSE_COMPRESS_WKSP_SIZE_U32(tableLog, maxSymbolValue)) return ERROR(tableLog_tooLarge);
-    if (srcSize <= 1) return 0;  /* Not compressible */
-    if (!maxSymbolValue) maxSymbolValue = FSE_MAX_SYMBOL_VALUE;
-    if (!tableLog) tableLog = FSE_DEFAULT_TABLELOG;
-
-    /* Scan input and build symbol stats */
-    {   CHECK_V_F(maxCount, HIST_count_wksp(count, &maxSymbolValue, src, srcSize, scratchBuffer, scratchBufferSize) );
-        if (maxCount == srcSize) return 1;   /* only a single symbol in src : rle */
-        if (maxCount == 1) return 0;         /* each symbol present maximum once => not compressible */
-        if (maxCount < (srcSize >> 7)) return 0;   /* Heuristic : not compressible enough */
-    }
-
-    tableLog = FSE_optimalTableLog(tableLog, srcSize, maxSymbolValue);
-    CHECK_F( FSE_normalizeCount(norm, tableLog, count, srcSize, maxSymbolValue, /* useLowProbCount */ srcSize >= 2048) );
-
-    /* Write table description header */
-    {   CHECK_V_F(nc_err, FSE_writeNCount(op, oend-op, norm, maxSymbolValue, tableLog) );
-        op += nc_err;
-    }
-
-    /* Compress */
-    CHECK_F( FSE_buildCTable_wksp(CTable, norm, maxSymbolValue, tableLog, scratchBuffer, scratchBufferSize) );
-    {   CHECK_V_F(cSize, FSE_compress_usingCTable(op, oend - op, src, srcSize, CTable) );
-        if (cSize == 0) return 0;   /* not enough space for compressed data */
-        op += cSize;
-    }
-
-    /* check compressibility */
-    if ( (size_t)(op-ostart) >= srcSize-1 ) return 0;
-
-    return op-ostart;
-}
-
-typedef struct {
-    FSE_CTable CTable_max[FSE_CTABLE_SIZE_U32(FSE_MAX_TABLELOG, FSE_MAX_SYMBOL_VALUE)];
-    union {
-      U32 hist_wksp[HIST_WKSP_SIZE_U32];
-      BYTE scratchBuffer[1 << FSE_MAX_TABLELOG];
-    } workspace;
-} fseWkspMax_t;
-
-size_t FSE_compress2 (void* dst, size_t dstCapacity, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog)
-{
-    fseWkspMax_t scratchBuffer;
-    DEBUG_STATIC_ASSERT(sizeof(scratchBuffer) >= FSE_COMPRESS_WKSP_SIZE_U32(FSE_MAX_TABLELOG, FSE_MAX_SYMBOL_VALUE));   /* compilation failures here means scratchBuffer is not large enough */
-    if (tableLog > FSE_MAX_TABLELOG) return ERROR(tableLog_tooLarge);
-    return FSE_compress_wksp(dst, dstCapacity, src, srcSize, maxSymbolValue, tableLog, &scratchBuffer, sizeof(scratchBuffer));
-}
-
-size_t FSE_compress (void* dst, size_t dstCapacity, const void* src, size_t srcSize)
-{
-    return FSE_compress2(dst, dstCapacity, src, srcSize, FSE_MAX_SYMBOL_VALUE, FSE_DEFAULT_TABLELOG);
-}
-#endif
-
 #endif   /* FSE_COMMONDEFS_ONLY */
@@ -29,7 +29,6 @@
 #include "hist.h"
 #define FSE_STATIC_LINKING_ONLY   /* FSE_optimalTableLog_internal */
 #include "../common/fse.h"        /* header compression */
-#define HUF_STATIC_LINKING_ONLY
 #include "../common/huf.h"
 #include "../common/error_private.h"
 #include "../common/bits.h"       /* ZSTD_highbit32 */
@@ -236,6 +235,8 @@ size_t HUF_writeCTable_wksp(void* dst, size_t maxDstSize,
    U32 n;
    HUF_WriteCTableWksp* wksp = (HUF_WriteCTableWksp*)HUF_alignUpWorkspace(workspace, &workspaceSize, ZSTD_ALIGNOF(U32));

+    HUF_STATIC_ASSERT(HUF_CTABLE_WORKSPACE_SIZE >= sizeof(HUF_WriteCTableWksp));
+
    /* check conditions */
    if (workspaceSize < sizeof(HUF_WriteCTableWksp)) return ERROR(GENERIC);
    if (maxSymbolValue > HUF_SYMBOLVALUE_MAX) return ERROR(maxSymbolValue_tooLarge);
@@ -265,16 +266,6 @@ size_t HUF_writeCTable_wksp(void* dst, size_t maxDstSize,
    return ((maxSymbolValue+1)/2) + 1;
 }

-/*! HUF_writeCTable() :
-    `CTable` : Huffman tree to save, using huf representation.
-    @return : size of saved CTable */
-size_t HUF_writeCTable (void* dst, size_t maxDstSize,
-                        const HUF_CElt* CTable, unsigned maxSymbolValue, unsigned huffLog)
-{
-    HUF_WriteCTableWksp wksp;
-    return HUF_writeCTable_wksp(dst, maxDstSize, CTable, maxSymbolValue, huffLog, &wksp, sizeof(wksp));
-}
-

 size_t HUF_readCTable (HUF_CElt* CTable, unsigned* maxSymbolValuePtr, const void* src, size_t srcSize, unsigned* hasZeroWeights)
 {
@@ -486,7 +477,7 @@ typedef struct {
    U16 curr;
 } rankPos;

-typedef nodeElt huffNodeTable[HUF_CTABLE_WORKSPACE_SIZE_U32];
+typedef nodeElt huffNodeTable[2 * (HUF_SYMBOLVALUE_MAX + 1)];

 /* Number of buckets available for HUF_sort() */
 #define RANK_POSITION_TABLE_SIZE 192
@@ -745,6 +736,8 @@ HUF_buildCTable_wksp(HUF_CElt* CTable, const unsigned* count, U32 maxSymbolValue
    nodeElt* const huffNode = huffNode0+1;
    int nonNullRank;

+    HUF_STATIC_ASSERT(HUF_CTABLE_WORKSPACE_SIZE == sizeof(HUF_buildCTable_wksp_tables));
+
    DEBUGLOG(5, "HUF_buildCTable_wksp (alphabet size = %u)", maxSymbolValue+1);

    /* safety checks */
@@ -1112,9 +1105,9 @@ HUF_compress1X_usingCTable_internal_default(void* dst, size_t dstSize,
 static size_t
 HUF_compress1X_usingCTable_internal(void* dst, size_t dstSize,
                              const void* src, size_t srcSize,
-                              const HUF_CElt* CTable, const int bmi2)
+                              const HUF_CElt* CTable, const int flags)
 {
-    if (bmi2) {
+    if (flags & HUF_flags_bmi2) {
        return HUF_compress1X_usingCTable_internal_bmi2(dst, dstSize, src, srcSize, CTable);
    }
    return HUF_compress1X_usingCTable_internal_default(dst, dstSize, src, srcSize, CTable);
@@ -1125,28 +1118,23 @@ HUF_compress1X_usingCTable_internal(void* dst, size_t dstSize,
 static size_t
 HUF_compress1X_usingCTable_internal(void* dst, size_t dstSize,
                              const void* src, size_t srcSize,
-                              const HUF_CElt* CTable, const int bmi2)
+                              const HUF_CElt* CTable, const int flags)
 {
-    (void)bmi2;
+    (void)flags;
    return HUF_compress1X_usingCTable_internal_body(dst, dstSize, src, srcSize, CTable);
 }

 #endif

-size_t HUF_compress1X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable)
+size_t HUF_compress1X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable, int flags)
 {
-    return HUF_compress1X_usingCTable_bmi2(dst, dstSize, src, srcSize, CTable, /* bmi2 */ 0);
-}
-
-size_t HUF_compress1X_usingCTable_bmi2(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable, int bmi2)
-{
-    return HUF_compress1X_usingCTable_internal(dst, dstSize, src, srcSize, CTable, bmi2);
+    return HUF_compress1X_usingCTable_internal(dst, dstSize, src, srcSize, CTable, flags);
 }

 static size_t
 HUF_compress4X_usingCTable_internal(void* dst, size_t dstSize,
                              const void* src, size_t srcSize,
-                              const HUF_CElt* CTable, int bmi2)
+                              const HUF_CElt* CTable, int flags)
 {
    size_t const segmentSize = (srcSize+3)/4;   /* first 3 segments */
    const BYTE* ip = (const BYTE*) src;
@@ -1160,7 +1148,7 @@ HUF_compress4X_usingCTable_internal(void* dst, size_t dstSize,
    op += 6;   /* jumpTable */

    assert(op <= oend);
-    {   CHECK_V_F(cSize, HUF_compress1X_usingCTable_internal(op, (size_t)(oend-op), ip, segmentSize, CTable, bmi2) );
+    {   CHECK_V_F(cSize, HUF_compress1X_usingCTable_internal(op, (size_t)(oend-op), ip, segmentSize, CTable, flags) );
        if (cSize == 0 || cSize > 65535) return 0;
        MEM_writeLE16(ostart, (U16)cSize);
        op += cSize;
@@ -1168,7 +1156,7 @@ HUF_compress4X_usingCTable_internal(void* dst, size_t dstSize,

    ip += segmentSize;
    assert(op <= oend);
-    {   CHECK_V_F(cSize, HUF_compress1X_usingCTable_internal(op, (size_t)(oend-op), ip, segmentSize, CTable, bmi2) );
+    {   CHECK_V_F(cSize, HUF_compress1X_usingCTable_internal(op, (size_t)(oend-op), ip, segmentSize, CTable, flags) );
        if (cSize == 0 || cSize > 65535) return 0;
        MEM_writeLE16(ostart+2, (U16)cSize);
        op += cSize;
@@ -1176,7 +1164,7 @@ HUF_compress4X_usingCTable_internal(void* dst, size_t dstSize,

    ip += segmentSize;
    assert(op <= oend);
-    {   CHECK_V_F(cSize, HUF_compress1X_usingCTable_internal(op, (size_t)(oend-op), ip, segmentSize, CTable, bmi2) );
+    {   CHECK_V_F(cSize, HUF_compress1X_usingCTable_internal(op, (size_t)(oend-op), ip, segmentSize, CTable, flags) );
        if (cSize == 0 || cSize > 65535) return 0;
        MEM_writeLE16(ostart+4, (U16)cSize);
        op += cSize;
@@ -1185,7 +1173,7 @@ HUF_compress4X_usingCTable_internal(void* dst, size_t dstSize,
    ip += segmentSize;
    assert(op <= oend);
    assert(ip <= iend);
-    {   CHECK_V_F(cSize, HUF_compress1X_usingCTable_internal(op, (size_t)(oend-op), ip, (size_t)(iend-ip), CTable, bmi2) );
+    {   CHECK_V_F(cSize, HUF_compress1X_usingCTable_internal(op, (size_t)(oend-op), ip, (size_t)(iend-ip), CTable, flags) );
        if (cSize == 0 || cSize > 65535) return 0;
        op += cSize;
    }
@@ -1193,14 +1181,9 @@ HUF_compress4X_usingCTable_internal(void* dst, size_t dstSize,
    return (size_t)(op-ostart);
 }

-size_t HUF_compress4X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable)
+size_t HUF_compress4X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable, int flags)
 {
-    return HUF_compress4X_usingCTable_bmi2(dst, dstSize, src, srcSize, CTable, /* bmi2 */ 0);
-}
-
-size_t HUF_compress4X_usingCTable_bmi2(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable, int bmi2)
-{
-    return HUF_compress4X_usingCTable_internal(dst, dstSize, src, srcSize, CTable, bmi2);
+    return HUF_compress4X_usingCTable_internal(dst, dstSize, src, srcSize, CTable, flags);
 }

 typedef enum { HUF_singleStream, HUF_fourStreams } HUF_nbStreams_e;
@@ -1208,11 +1191,11 @@ typedef enum { HUF_singleStream, HUF_fourStreams } HUF_nbStreams_e;
 static size_t HUF_compressCTable_internal(
                BYTE* const ostart, BYTE* op, BYTE* const oend,
                const void* src, size_t srcSize,
-                HUF_nbStreams_e nbStreams, const HUF_CElt* CTable, const int bmi2)
+                HUF_nbStreams_e nbStreams, const HUF_CElt* CTable, const int flags)
 {
    size_t const cSize = (nbStreams==HUF_singleStream) ?
-                         HUF_compress1X_usingCTable_internal(op, (size_t)(oend - op), src, srcSize, CTable, bmi2) :
-                         HUF_compress4X_usingCTable_internal(op, (size_t)(oend - op), src, srcSize, CTable, bmi2);
+                         HUF_compress1X_usingCTable_internal(op, (size_t)(oend - op), src, srcSize, CTable, flags) :
+                         HUF_compress4X_usingCTable_internal(op, (size_t)(oend - op), src, srcSize, CTable, flags);
    if (HUF_isError(cSize)) { return cSize; }
    if (cSize==0) { return 0; }   /* uncompressible */
    op += cSize;
@@ -1260,12 +1243,12 @@ unsigned HUF_optimalTableLog(
            void* workSpace, size_t wkspSize,
            HUF_CElt* table,
      const unsigned* count,
-            HUF_depth_mode depthMode)
+            int flags)
 {
    assert(srcSize > 1); /* Not supported, RLE should be used instead */
    assert(wkspSize >= sizeof(HUF_buildCTable_wksp_tables));

-    if (depthMode != HUF_depth_optimal) {
+    if (!(flags & HUF_flags_optimalDepth)) {
        /* cheap evaluation, based on FSE */
        return FSE_optimalTableLog_internal(maxTableLog, srcSize, maxSymbolValue, 1);
    }
@@ -1317,8 +1300,7 @@ HUF_compress_internal (void* dst, size_t dstSize,
                       unsigned maxSymbolValue, unsigned huffLog,
                       HUF_nbStreams_e nbStreams,
                       void* workSpace, size_t wkspSize,
-                       HUF_CElt* oldHufTable, HUF_repeat* repeat, int preferRepeat,
-                 const int bmi2, unsigned suspectUncompressible, HUF_depth_mode depthMode)
+                       HUF_CElt* oldHufTable, HUF_repeat* repeat, int flags)
 {
    HUF_compress_tables_t* const table = (HUF_compress_tables_t*)HUF_alignUpWorkspace(workSpace, &wkspSize, ZSTD_ALIGNOF(size_t));
    BYTE* const ostart = (BYTE*)dst;
@@ -1339,15 +1321,15 @@ HUF_compress_internal (void* dst, size_t dstSize,
    if (!huffLog) huffLog = HUF_TABLELOG_DEFAULT;

    /* Heuristic : If old table is valid, use it for small inputs */
-    if (preferRepeat && repeat && *repeat == HUF_repeat_valid) {
+    if ((flags & HUF_flags_preferRepeat) && repeat && *repeat == HUF_repeat_valid) {
        return HUF_compressCTable_internal(ostart, op, oend,
                                           src, srcSize,
-                                           nbStreams, oldHufTable, bmi2);
+                                           nbStreams, oldHufTable, flags);
    }

    /* If uncompressible data is suspected, do a smaller sampling first */
    DEBUG_STATIC_ASSERT(SUSPECT_INCOMPRESSIBLE_SAMPLE_RATIO >= 2);
-    if (suspectUncompressible && srcSize >= (SUSPECT_INCOMPRESSIBLE_SAMPLE_SIZE * SUSPECT_INCOMPRESSIBLE_SAMPLE_RATIO)) {
+    if ((flags & HUF_flags_suspectUncompressible) && srcSize >= (SUSPECT_INCOMPRESSIBLE_SAMPLE_SIZE * SUSPECT_INCOMPRESSIBLE_SAMPLE_RATIO)) {
        size_t largestTotal = 0;
        DEBUGLOG(5, "input suspected incompressible : sampling to check");
        {   unsigned maxSymbolValueBegin = maxSymbolValue;
@@ -1375,14 +1357,14 @@ HUF_compress_internal (void* dst, size_t dstSize,
        *repeat = HUF_repeat_none;
    }
    /* Heuristic : use existing table for small inputs */
-    if (preferRepeat && repeat && *repeat != HUF_repeat_none) {
+    if ((flags & HUF_flags_preferRepeat) && repeat && *repeat != HUF_repeat_none) {
        return HUF_compressCTable_internal(ostart, op, oend,
                                           src, srcSize,
-                                           nbStreams, oldHufTable, bmi2);
+                                           nbStreams, oldHufTable, flags);
    }

    /* Build Huffman Tree */
-    huffLog = HUF_optimalTableLog(huffLog, srcSize, maxSymbolValue, &table->wksps, sizeof(table->wksps), table->CTable, table->count, depthMode);
+    huffLog = HUF_optimalTableLog(huffLog, srcSize, maxSymbolValue, &table->wksps, sizeof(table->wksps), table->CTable, table->count, flags);
    {   size_t const maxBits = HUF_buildCTable_wksp(table->CTable, table->count,
                                            maxSymbolValue, huffLog,
                                            &table->wksps.buildCTable_wksp, sizeof(table->wksps.buildCTable_wksp));
@@ -1407,7 +1389,7 @@ HUF_compress_internal (void* dst, size_t dstSize,
            if (oldSize <= hSize + newSize || hSize + 12 >= srcSize) {
                return HUF_compressCTable_internal(ostart, op, oend,
                                                   src, srcSize,
-                                                   nbStreams, oldHufTable, bmi2);
+                                                   nbStreams, oldHufTable, flags);
        }   }

        /* Use the new huffman table */
@@ -1419,48 +1401,20 @@ HUF_compress_internal (void* dst, size_t dstSize,
    }
    return HUF_compressCTable_internal(ostart, op, oend,
                                       src, srcSize,
-                                       nbStreams, table->CTable, bmi2);
-}
-
-
-size_t HUF_compress1X_wksp (void* dst, size_t dstSize,
-                      const void* src, size_t srcSize,
-                      unsigned maxSymbolValue, unsigned huffLog,
-                      void* workSpace, size_t wkspSize)
-{
-    return HUF_compress_internal(dst, dstSize, src, srcSize,
-                                 maxSymbolValue, huffLog, HUF_singleStream,
-                                 workSpace, wkspSize,
-                                 NULL, NULL, 0, 0 /*bmi2*/, 0, HUF_depth_fast);
+                                       nbStreams, table->CTable, flags);
 }

 size_t HUF_compress1X_repeat (void* dst, size_t dstSize,
                      const void* src, size_t srcSize,
                      unsigned maxSymbolValue, unsigned huffLog,
                      void* workSpace, size_t wkspSize,
-                      HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat,
-                      int bmi2, unsigned suspectUncompressible, HUF_depth_mode depthMode)
+                      HUF_CElt* hufTable, HUF_repeat* repeat, int flags)
 {
    DEBUGLOG(5, "HUF_compress1X_repeat (srcSize = %zu)", srcSize);
    return HUF_compress_internal(dst, dstSize, src, srcSize,
                                 maxSymbolValue, huffLog, HUF_singleStream,
                                 workSpace, wkspSize, hufTable,
-                                 repeat, preferRepeat, bmi2, suspectUncompressible, depthMode);
-}
-
-/* HUF_compress4X_repeat():
- * compress input using 4 streams.
- * provide workspace to generate compression tables */
-size_t HUF_compress4X_wksp (void* dst, size_t dstSize,
-                      const void* src, size_t srcSize,
-                      unsigned maxSymbolValue, unsigned huffLog,
-                      void* workSpace, size_t wkspSize)
-{
-    DEBUGLOG(5, "HUF_compress4X_wksp (srcSize = %zu)", srcSize);
-    return HUF_compress_internal(dst, dstSize, src, srcSize,
-                                 maxSymbolValue, huffLog, HUF_fourStreams,
-                                 workSpace, wkspSize,
-                                 NULL, NULL, 0, 0 /*bmi2*/, 0, HUF_depth_fast);
+                                 repeat, flags);
 }

 /* HUF_compress4X_repeat():
@@ -1471,45 +1425,11 @@ size_t HUF_compress4X_repeat (void* dst, size_t dstSize,
                      const void* src, size_t srcSize,
                      unsigned maxSymbolValue, unsigned huffLog,
                      void* workSpace, size_t wkspSize,
-                      HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2,
-                      unsigned suspectUncompressible, HUF_depth_mode depthMode)
+                      HUF_CElt* hufTable, HUF_repeat* repeat, int flags)
 {
    DEBUGLOG(5, "HUF_compress4X_repeat (srcSize = %zu)", srcSize);
    return HUF_compress_internal(dst, dstSize, src, srcSize,
                                 maxSymbolValue, huffLog, HUF_fourStreams,
                                 workSpace, wkspSize,
-                                 hufTable, repeat, preferRepeat, bmi2, suspectUncompressible, depthMode);
+                                 hufTable, repeat, flags);
 }
-
-#ifndef ZSTD_NO_UNUSED_FUNCTIONS
-/** HUF_buildCTable() :
- * @return : maxNbBits
- *  Note : count is used before tree is written, so they can safely overlap
- */
-size_t HUF_buildCTable (HUF_CElt* tree, const unsigned* count, unsigned maxSymbolValue, unsigned maxNbBits)
-{
-    HUF_buildCTable_wksp_tables workspace;
-    return HUF_buildCTable_wksp(tree, count, maxSymbolValue, maxNbBits, &workspace, sizeof(workspace));
-}
-
-size_t HUF_compress1X (void* dst, size_t dstSize,
-                 const void* src, size_t srcSize,
-                 unsigned maxSymbolValue, unsigned huffLog)
-{
-    U64 workSpace[HUF_WORKSPACE_SIZE_U64];
-    return HUF_compress1X_wksp(dst, dstSize, src, srcSize, maxSymbolValue, huffLog, workSpace, sizeof(workSpace));
-}
-
-size_t HUF_compress2 (void* dst, size_t dstSize,
-                const void* src, size_t srcSize,
-                unsigned maxSymbolValue, unsigned huffLog)
-{
-    U64 workSpace[HUF_WORKSPACE_SIZE_U64];
-    return HUF_compress4X_wksp(dst, dstSize, src, srcSize, maxSymbolValue, huffLog, workSpace, sizeof(workSpace));
-}
-
-size_t HUF_compress (void* dst, size_t maxDstSize, const void* src, size_t srcSize)
-{
-    return HUF_compress2(dst, maxDstSize, src, srcSize, 255, HUF_TABLELOG_DEFAULT);
-}
-#endif
@@ -226,8 +226,10 @@ struct ZSTD_matchState_t {
    U32 hashLog3;           /* dispatch table for matches of len==3 : larger == faster, more memory */

    U32 rowHashLog;                          /* For row-based matchfinder: Hashlog based on nb of rows in the hashTable.*/
-    U16* tagTable;                           /* For row-based matchFinder: A row-based table containing the hashes and head index. */
+    BYTE* tagTable;                          /* For row-based matchFinder: A row-based table containing the hashes and head index. */
    U32 hashCache[ZSTD_ROW_HASH_CACHE_SIZE]; /* For row-based matchFinder: a cache of hashes to improve speed */
+    U64 hashSalt;                            /* For row-based matchFinder: salts the hash for re-use of tag table */
+    U32 hashSaltEntropy;                     /* For row-based matchFinder: collects entropy for salt generation */

    U32* hashTable;
    U32* hashTable3;
@@ -247,6 +249,13 @@ struct ZSTD_matchState_t {
     * This behavior is controlled from the cctx ms.
     * This parameter has no effect in the cdict ms. */
    int prefetchCDictTables;
+
+    /* When == 0, lazy match finders insert every position.
+     * When != 0, lazy match finders only insert positions they search.
+     * This allows them to skip much faster over incompressible data,
+     * at a small cost to compression ratio.
+     */
+    int lazySkipping;
 };

 typedef struct {
@@ -353,8 +362,14 @@ struct ZSTD_CCtx_params_s {

    /* Indicates whether an external matchfinder has been referenced.
     * Users can't set this externally.
-     * It is set internally in ZSTD_registerExternalMatchFinder(). */
-    int useExternalMatchFinder;
+     * It is set internally in ZSTD_registerSequenceProducer(). */
+    int useSequenceProducer;
+
+    /* Adjust the max block size*/
+    size_t maxBlockSize;
+
+    /* Controls repcode search in external sequence parsing */
+    ZSTD_paramSwitch_e searchForExternalRepcodes;
 };  /* typedef'd to ZSTD_CCtx_params within "zstd.h" */

 #define COMPRESS_SEQUENCES_WORKSPACE_SIZE (sizeof(unsigned) * (MaxSeq + 2))
@@ -389,7 +404,7 @@ typedef struct {
 /* Context for block-level external matchfinder API */
 typedef struct {
  void* mState;
-  ZSTD_externalMatchFinder_F* mFinder;
+  ZSTD_sequenceProducer_F* mFinder;
  ZSTD_Sequence* seqBuffer;
  size_t seqBufferCapacity;
 } ZSTD_externalMatchCtx;
@@ -491,7 +506,7 @@ typedef enum {
                                 * In this mode we take both the source size and the dictionary size
                                 * into account when selecting and adjusting the parameters.
                                 */
-    ZSTD_cpm_unknown = 3,       /* ZSTD_getCParams, ZSTD_getParams, ZSTD_adjustParams.
+    ZSTD_cpm_unknown = 3        /* ZSTD_getCParams, ZSTD_getParams, ZSTD_adjustParams.
                                 * We don't know what these parameters are for. We default to the legacy
                                 * behavior of taking both the source size and the dict size into account
                                 * when selecting and adjusting parameters.
@@ -781,28 +796,35 @@ ZSTD_count_2segments(const BYTE* ip, const BYTE* match,
 *  Hashes
 ***************************************/
 static const U32 prime3bytes = 506832829U;
-static U32    ZSTD_hash3(U32 u, U32 h) { assert(h <= 32); return ((u << (32-24)) * prime3bytes)  >> (32-h) ; }
-MEM_STATIC size_t ZSTD_hash3Ptr(const void* ptr, U32 h) { return ZSTD_hash3(MEM_readLE32(ptr), h); } /* only in zstd_opt.h */
+static U32    ZSTD_hash3(U32 u, U32 h, U32 s) { assert(h <= 32); return (((u << (32-24)) * prime3bytes) ^ s)  >> (32-h) ; }
+MEM_STATIC size_t ZSTD_hash3Ptr(const void* ptr, U32 h) { return ZSTD_hash3(MEM_readLE32(ptr), h, 0); } /* only in zstd_opt.h */
+MEM_STATIC size_t ZSTD_hash3PtrS(const void* ptr, U32 h, U32 s) { return ZSTD_hash3(MEM_readLE32(ptr), h, s); }

 static const U32 prime4bytes = 2654435761U;
-static U32    ZSTD_hash4(U32 u, U32 h) { assert(h <= 32); return (u * prime4bytes) >> (32-h) ; }
-static size_t ZSTD_hash4Ptr(const void* ptr, U32 h) { return ZSTD_hash4(MEM_readLE32(ptr), h); }
+static U32    ZSTD_hash4(U32 u, U32 h, U32 s) { assert(h <= 32); return ((u * prime4bytes) ^ s) >> (32-h) ; }
+static size_t ZSTD_hash4Ptr(const void* ptr, U32 h) { return ZSTD_hash4(MEM_readLE32(ptr), h, 0); }
+static size_t ZSTD_hash4PtrS(const void* ptr, U32 h, U32 s) { return ZSTD_hash4(MEM_readLE32(ptr), h, s); }

 static const U64 prime5bytes = 889523592379ULL;
-static size_t ZSTD_hash5(U64 u, U32 h) { assert(h <= 64); return (size_t)(((u  << (64-40)) * prime5bytes) >> (64-h)) ; }
-static size_t ZSTD_hash5Ptr(const void* p, U32 h) { return ZSTD_hash5(MEM_readLE64(p), h); }
+static size_t ZSTD_hash5(U64 u, U32 h, U64 s) { assert(h <= 64); return (size_t)((((u  << (64-40)) * prime5bytes) ^ s) >> (64-h)) ; }
+static size_t ZSTD_hash5Ptr(const void* p, U32 h) { return ZSTD_hash5(MEM_readLE64(p), h, 0); }
+static size_t ZSTD_hash5PtrS(const void* p, U32 h, U64 s) { return ZSTD_hash5(MEM_readLE64(p), h, s); }

 static const U64 prime6bytes = 227718039650203ULL;
-static size_t ZSTD_hash6(U64 u, U32 h) { assert(h <= 64); return (size_t)(((u  << (64-48)) * prime6bytes) >> (64-h)) ; }
-static size_t ZSTD_hash6Ptr(const void* p, U32 h) { return ZSTD_hash6(MEM_readLE64(p), h); }
+static size_t ZSTD_hash6(U64 u, U32 h, U64 s) { assert(h <= 64); return (size_t)((((u  << (64-48)) * prime6bytes) ^ s) >> (64-h)) ; }
+static size_t ZSTD_hash6Ptr(const void* p, U32 h) { return ZSTD_hash6(MEM_readLE64(p), h, 0); }
+static size_t ZSTD_hash6PtrS(const void* p, U32 h, U64 s) { return ZSTD_hash6(MEM_readLE64(p), h, s); }

 static const U64 prime7bytes = 58295818150454627ULL;
-static size_t ZSTD_hash7(U64 u, U32 h) { assert(h <= 64); return (size_t)(((u  << (64-56)) * prime7bytes) >> (64-h)) ; }
-static size_t ZSTD_hash7Ptr(const void* p, U32 h) { return ZSTD_hash7(MEM_readLE64(p), h); }
+static size_t ZSTD_hash7(U64 u, U32 h, U64 s) { assert(h <= 64); return (size_t)((((u  << (64-56)) * prime7bytes) ^ s) >> (64-h)) ; }
+static size_t ZSTD_hash7Ptr(const void* p, U32 h) { return ZSTD_hash7(MEM_readLE64(p), h, 0); }
+static size_t ZSTD_hash7PtrS(const void* p, U32 h, U64 s) { return ZSTD_hash7(MEM_readLE64(p), h, s); }

 static const U64 prime8bytes = 0xCF1BBCDCB7A56463ULL;
-static size_t ZSTD_hash8(U64 u, U32 h) { assert(h <= 64); return (size_t)(((u) * prime8bytes) >> (64-h)) ; }
-static size_t ZSTD_hash8Ptr(const void* p, U32 h) { return ZSTD_hash8(MEM_readLE64(p), h); }
+static size_t ZSTD_hash8(U64 u, U32 h, U64 s) { assert(h <= 64); return (size_t)((((u) * prime8bytes)  ^ s) >> (64-h)) ; }
+static size_t ZSTD_hash8Ptr(const void* p, U32 h) { return ZSTD_hash8(MEM_readLE64(p), h, 0); }
+static size_t ZSTD_hash8PtrS(const void* p, U32 h, U64 s) { return ZSTD_hash8(MEM_readLE64(p), h, s); }
+

 MEM_STATIC FORCE_INLINE_ATTR
 size_t ZSTD_hashPtr(const void* p, U32 hBits, U32 mls)
@@ -822,6 +844,24 @@ size_t ZSTD_hashPtr(const void* p, U32 hBits, U32 mls)
    }
 }

+MEM_STATIC FORCE_INLINE_ATTR
+size_t ZSTD_hashPtrSalted(const void* p, U32 hBits, U32 mls, const U64 hashSalt) {
+    /* Although some of these hashes do support hBits up to 64, some do not.
+     * To be on the safe side, always avoid hBits > 32. */
+    assert(hBits <= 32);
+
+    switch(mls)
+    {
+        default:
+        case 4: return ZSTD_hash4PtrS(p, hBits, (U32)hashSalt);
+        case 5: return ZSTD_hash5PtrS(p, hBits, hashSalt);
+        case 6: return ZSTD_hash6PtrS(p, hBits, hashSalt);
+        case 7: return ZSTD_hash7PtrS(p, hBits, hashSalt);
+        case 8: return ZSTD_hash8PtrS(p, hBits, hashSalt);
+    }
+}
+
+
 /** ZSTD_ipow() :
 * Return base^exponent.
 */
@@ -1169,10 +1209,15 @@ ZSTD_checkDictValidity(const ZSTD_window_t* window,
                    (unsigned)blockEndIdx, (unsigned)maxDist, (unsigned)loadedDictEnd);
        assert(blockEndIdx >= loadedDictEnd);

-        if (blockEndIdx > loadedDictEnd + maxDist) {
+        if (blockEndIdx > loadedDictEnd + maxDist || loadedDictEnd != window->dictLimit) {
            /* On reaching window size, dictionaries are invalidated.
             * For simplification, if window size is reached anywhere within next block,
             * the dictionary is invalidated for the full block.
+             *
+             * We also have to invalidate the dictionary if ZSTD_window_update() has detected
+             * non-contiguous segments, which means that loadedDictEnd != window->dictLimit.
+             * loadedDictEnd may be 0, if forceWindow is true, but in that case we never use
+             * dictMatchState, so setting it to NULL is not a problem.
             */
            DEBUGLOG(6, "invalidating dictionary for current block (distance > windowSize)");
            *loadedDictEndPtr = 0;
@@ -1445,7 +1490,7 @@ size_t
 ZSTD_copySequencesToSeqStoreExplicitBlockDelim(ZSTD_CCtx* cctx,
                                              ZSTD_sequencePosition* seqPos,
                                        const ZSTD_Sequence* const inSeqs, size_t inSeqsSize,
-                                        const void* src, size_t blockSize);
+                                        const void* src, size_t blockSize, ZSTD_paramSwitch_e externalRepSearch);

 /* Returns the number of bytes to move the current read position back by.
 * Only non-zero if we ended up splitting a sequence.
@@ -1462,6 +1507,26 @@ ZSTD_copySequencesToSeqStoreExplicitBlockDelim(ZSTD_CCtx* cctx,
 size_t
 ZSTD_copySequencesToSeqStoreNoBlockDelim(ZSTD_CCtx* cctx, ZSTD_sequencePosition* seqPos,
                                   const ZSTD_Sequence* const inSeqs, size_t inSeqsSize,
-                                   const void* src, size_t blockSize);
+                                   const void* src, size_t blockSize, ZSTD_paramSwitch_e externalRepSearch);
+
+
+/* ===============================================================
+ * Deprecated definitions that are still used internally to avoid
+ * deprecation warnings. These functions are exactly equivalent to
+ * their public variants, but avoid the deprecation warnings.
+ * =============================================================== */
+
+size_t ZSTD_compressBegin_usingCDict_deprecated(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict);
+
+size_t ZSTD_compressContinue_public(ZSTD_CCtx* cctx,
+                                    void* dst, size_t dstCapacity,
+                              const void* src, size_t srcSize);
+
+size_t ZSTD_compressEnd_public(ZSTD_CCtx* cctx,
+                               void* dst, size_t dstCapacity,
+                         const void* src, size_t srcSize);
+
+size_t ZSTD_compressBlock_deprecated(ZSTD_CCtx* cctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
+

 #endif /* ZSTD_COMPRESS_H */
@@ -24,9 +24,9 @@ static size_t showHexa(const void* src, size_t srcSize)
    const BYTE* const ip = (const BYTE*)src;
    size_t u;
    for (u=0; u<srcSize; u++) {
-        RAWLOG(6, " %02X", ip[u]); (void)ip;
+        RAWLOG(5, " %02X", ip[u]); (void)ip;
    }
-    RAWLOG(6, " \n");
+    RAWLOG(5, " \n");
    return srcSize;
 }

@@ -65,12 +65,26 @@ size_t ZSTD_noCompressLiterals (void* dst, size_t dstCapacity, const void* src,
    return srcSize + flSize;
 }

+static int allBytesIdentical(const void* src, size_t srcSize)
+{
+    assert(srcSize >= 1);
+    assert(src != NULL);
+    {   const BYTE b = ((const BYTE*)src)[0];
+        size_t p;
+        for (p=1; p<srcSize; p++) {
+            if (((const BYTE*)src)[p] != b) return 0;
+        }
+        return 1;
+    }
+}
+
 size_t ZSTD_compressRleLiteralsBlock (void* dst, size_t dstCapacity, const void* src, size_t srcSize)
 {
    BYTE* const ostart = (BYTE*)dst;
    U32   const flSize = 1 + (srcSize>31) + (srcSize>4095);

-    (void)dstCapacity;  /* dstCapacity already guaranteed to be >=4, hence large enough */
+    assert(dstCapacity >= 4); (void)dstCapacity;
+    assert(allBytesIdentical(src, srcSize));

    switch(flSize)
    {
@@ -88,7 +102,7 @@ size_t ZSTD_compressRleLiteralsBlock (void* dst, size_t dstCapacity, const void*
    }

    ostart[flSize] = *(const BYTE*)src;
-    DEBUGLOG(5, "RLE literals: %u -> %u", (U32)srcSize, (U32)flSize + 1);
+    DEBUGLOG(5, "RLE : Repeated Literal (%02X: %u times) -> %u bytes encoded", ((const BYTE*)src)[0], (U32)srcSize, (U32)flSize + 1);
    return flSize+1;
 }

@@ -105,8 +119,8 @@ ZSTD_minLiteralsToCompress(ZSTD_strategy strategy, HUF_repeat huf_repeat)
    /* btultra2 : min 8 bytes;
     * then 2x larger for each successive compression strategy
     * max threshold 64 bytes */
-    {   int const shift = MIN(9-strategy, 3);
-        size_t const mintc = (huf_repeat == HUF_repeat_valid) ? 6 : 8 << shift;
+    {   int const shift = MIN(9-(int)strategy, 3);
+        size_t const mintc = (huf_repeat == HUF_repeat_valid) ? 6 : (size_t)8 << shift;
        DEBUGLOG(7, "minLiteralsToCompress = %zu", mintc);
        return mintc;
    }
@@ -146,9 +160,13 @@ size_t ZSTD_compressLiterals (

    RETURN_ERROR_IF(dstCapacity < lhSize+1, dstSize_tooSmall, "not enough space for compression");
    {   HUF_repeat repeat = prevHuf->repeatMode;
-        int const preferRepeat = (strategy < ZSTD_lazy) ? srcSize <= 1024 : 0;
-        HUF_depth_mode const depthMode = (strategy >= HUF_OPTIMAL_DEPTH_THRESHOLD) ? HUF_depth_optimal : HUF_depth_fast;
-        typedef size_t (*huf_compress_f)(void*, size_t, const void*, size_t, unsigned, unsigned, void*, size_t, HUF_CElt*, HUF_repeat*, int, int, unsigned, HUF_depth_mode);
+        int const flags = 0
+            | (bmi2 ? HUF_flags_bmi2 : 0)
+            | (strategy < ZSTD_lazy && srcSize <= 1024 ? HUF_flags_preferRepeat : 0)
+            | (strategy >= HUF_OPTIMAL_DEPTH_THRESHOLD ? HUF_flags_optimalDepth : 0)
+            | (suspectUncompressible ? HUF_flags_suspectUncompressible : 0);
+
+        typedef size_t (*huf_compress_f)(void*, size_t, const void*, size_t, unsigned, unsigned, void*, size_t, HUF_CElt*, HUF_repeat*, int);
        huf_compress_f huf_compress;
        if (repeat == HUF_repeat_valid && lhSize == 3) singleStream = 1;
        huf_compress = singleStream ? HUF_compress1X_repeat : HUF_compress4X_repeat;
@@ -157,11 +175,11 @@ size_t ZSTD_compressLiterals (
                                HUF_SYMBOLVALUE_MAX, LitHufLog,
                                entropyWorkspace, entropyWorkspaceSize,
                                (HUF_CElt*)nextHuf->CTable,
-                                &repeat, preferRepeat,
-                                bmi2, suspectUncompressible, depthMode);
+                                &repeat, flags);
+        DEBUGLOG(5, "%zu literals compressed into %zu bytes (before header)", srcSize, cLitSize);
        if (repeat != HUF_repeat_none) {
            /* reused the existing table */
-            DEBUGLOG(5, "Reusing previous huffman table");
+            DEBUGLOG(5, "reusing statistics from previous huffman block");
            hType = set_repeat;
        }
    }
@@ -172,9 +190,15 @@ size_t ZSTD_compressLiterals (
            return ZSTD_noCompressLiterals(dst, dstCapacity, src, srcSize);
    }   }
    if (cLitSize==1) {
-        ZSTD_memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
-        return ZSTD_compressRleLiteralsBlock(dst, dstCapacity, src, srcSize);
-    }
+        /* A return value of 1 signals that the alphabet consists of a single symbol.
+         * However, in some rare circumstances, it could be the compressed size (a single byte).
+         * For that outcome to have a chance to happen, it's necessary that `srcSize < 8`.
+         * (it's also necessary to not generate statistics).
+         * Therefore, in such a case, actively check that all bytes are identical. */
+        if ((srcSize >= 8) || allBytesIdentical(src, srcSize)) {
+            ZSTD_memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
+            return ZSTD_compressRleLiteralsBlock(dst, dstCapacity, src, srcSize);
+    }   }

    if (hType == set_compressed) {
        /* using a newly constructed table */
@@ -16,6 +16,10 @@

 size_t ZSTD_noCompressLiterals (void* dst, size_t dstCapacity, const void* src, size_t srcSize);

+/* ZSTD_compressRleLiteralsBlock() :
+ * Conditions :
+ * - All bytes in @src are identical
+ * - dstCapacity >= 4 */
 size_t ZSTD_compressRleLiteralsBlock (void* dst, size_t dstCapacity, const void* src, size_t srcSize);

 /* ZSTD_compressLiterals():
@@ -54,8 +54,6 @@ ZSTD_compressSubBlock_literal(const HUF_CElt* hufTable,
    symbolEncodingType_e hType = writeEntropy ? hufMetadata->hType : set_repeat;
    size_t cLitSize = 0;

-    (void)bmi2; /* TODO bmi2... */
-
    DEBUGLOG(5, "ZSTD_compressSubBlock_literal (litSize=%zu, lhSize=%zu, writeEntropy=%d)", litSize, lhSize, writeEntropy);

    *entropyWritten = 0;
@@ -77,9 +75,9 @@ ZSTD_compressSubBlock_literal(const HUF_CElt* hufTable,
        DEBUGLOG(5, "ZSTD_compressSubBlock_literal (hSize=%zu)", hufMetadata->hufDesSize);
    }

-    /* TODO bmi2 */
-    {   const size_t cSize = singleStream ? HUF_compress1X_usingCTable(op, oend-op, literals, litSize, hufTable)
-                                          : HUF_compress4X_usingCTable(op, oend-op, literals, litSize, hufTable);
+    {   int const flags = bmi2 ? HUF_flags_bmi2 : 0;
+        const size_t cSize = singleStream ? HUF_compress1X_usingCTable(op, oend-op, literals, litSize, hufTable, flags)
+                                          : HUF_compress4X_usingCTable(op, oend-op, literals, litSize, hufTable, flags);
        op += cSize;
        cLitSize += cSize;
        if (cSize == 0 || ERR_isError(cSize)) {
@@ -14,7 +14,9 @@
 /*-*************************************
 *  Dependencies
 ***************************************/
+#include "../common/allocations.h"  /* ZSTD_customMalloc, ZSTD_customFree */
 #include "../common/zstd_internal.h"
+#include "../common/portability_macros.h"

 #if defined (__cplusplus)
 extern "C" {
@@ -44,8 +46,9 @@ extern "C" {
 ***************************************/
 typedef enum {
    ZSTD_cwksp_alloc_objects,
-    ZSTD_cwksp_alloc_buffers,
-    ZSTD_cwksp_alloc_aligned
+    ZSTD_cwksp_alloc_aligned_init_once,
+    ZSTD_cwksp_alloc_aligned,
+    ZSTD_cwksp_alloc_buffers
 } ZSTD_cwksp_alloc_phase_e;

 /**
@@ -98,8 +101,8 @@ typedef enum {
 *
 * Workspace Layout:
 *
- * [                        ... workspace ...                         ]
- * [objects][tables ... ->] free space [<- ... aligned][<- ... buffers]
+ * [                        ... workspace ...                           ]
+ * [objects][tables ->] free space [<- buffers][<- aligned][<- init once]
 *
 * The various objects that live in the workspace are divided into the
 * following categories, and are allocated separately:
@@ -123,9 +126,18 @@ typedef enum {
 *   uint32_t arrays, all of whose values are between 0 and (nextSrc - base).
 *   Their sizes depend on the cparams. These tables are 64-byte aligned.
 *
- * - Aligned: these buffers are used for various purposes that require 4 byte
- *   alignment, but don't require any initialization before they're used. These
- *   buffers are each aligned to 64 bytes.
+ * - Init once: these buffers require to be initialized at least once before
+ *   use. They should be used when we want to skip memory initialization
+ *   while not triggering memory checkers (like Valgrind) when reading from
+ *   from this memory without writing to it first.
+ *   These buffers should be used carefully as they might contain data
+ *   from previous compressions.
+ *   Buffers are aligned to 64 bytes.
+ *
+ * - Aligned: these buffers don't require any initialization before they're
+ *   used. The user of the buffer should make sure they write into a buffer
+ *   location before reading from it.
+ *   Buffers are aligned to 64 bytes.
 *
 * - Buffers: these buffers are used for various purposes that don't require
 *   any alignment or initialization before they're used. This means they can
@@ -137,8 +149,9 @@ typedef enum {
 * correctly packed into the workspace buffer. That order is:
 *
 * 1. Objects
- * 2. Buffers
- * 3. Aligned/Tables
+ * 2. Init once / Tables
+ * 3. Aligned / Tables
+ * 4. Buffers / Tables
 *
 * Attempts to reserve objects of different types out of order will fail.
 */
@@ -150,6 +163,7 @@ typedef struct {
    void* tableEnd;
    void* tableValidEnd;
    void* allocStart;
+    void* initOnceStart;

    BYTE allocFailed;
    int workspaceOversizedDuration;
@@ -162,6 +176,7 @@ typedef struct {
 ***************************************/

 MEM_STATIC size_t ZSTD_cwksp_available_space(ZSTD_cwksp* ws);
+MEM_STATIC void*  ZSTD_cwksp_initialAllocStart(ZSTD_cwksp* ws);

 MEM_STATIC void ZSTD_cwksp_assert_internal_consistency(ZSTD_cwksp* ws) {
    (void)ws;
@@ -171,6 +186,20 @@ MEM_STATIC void ZSTD_cwksp_assert_internal_consistency(ZSTD_cwksp* ws) {
    assert(ws->tableEnd <= ws->allocStart);
    assert(ws->tableValidEnd <= ws->allocStart);
    assert(ws->allocStart <= ws->workspaceEnd);
+    assert(ws->initOnceStart <= ZSTD_cwksp_initialAllocStart(ws));
+    assert(ws->workspace <= ws->initOnceStart);
+#if ZSTD_MEMORY_SANITIZER
+    {
+        intptr_t const offset = __msan_test_shadow(ws->initOnceStart,
+            (U8*)ZSTD_cwksp_initialAllocStart(ws) - (U8*)ws->initOnceStart);
+#if defined(ZSTD_MSAN_PRINT)
+        if(offset!=-1) {
+            __msan_print_shadow((U8*)ws->initOnceStart + offset - 8, 32);
+        }
+#endif
+        assert(offset==-1);
+    };
+#endif
 }

 /**
@@ -217,14 +246,10 @@ MEM_STATIC size_t ZSTD_cwksp_aligned_alloc_size(size_t size) {
 * for internal purposes (currently only alignment).
 */
 MEM_STATIC size_t ZSTD_cwksp_slack_space_required(void) {
-    /* For alignment, the wksp will always allocate an additional n_1=[1, 64] bytes
-     * to align the beginning of tables section, as well as another n_2=[0, 63] bytes
-     * to align the beginning of the aligned section.
-     *
-     * n_1 + n_2 == 64 bytes if the cwksp is freshly allocated, due to tables and
-     * aligneds being sized in multiples of 64 bytes.
+    /* For alignment, the wksp will always allocate an additional 2*ZSTD_CWKSP_ALIGNMENT_BYTES
+     * bytes to align the beginning of tables section and end of buffers;
     */
-    size_t const slackSpace = ZSTD_CWKSP_ALIGNMENT_BYTES;
+    size_t const slackSpace = ZSTD_CWKSP_ALIGNMENT_BYTES * 2;
    return slackSpace;
 }

@@ -237,10 +262,18 @@ MEM_STATIC size_t ZSTD_cwksp_bytes_to_align_ptr(void* ptr, const size_t alignByt
    size_t const alignBytesMask = alignBytes - 1;
    size_t const bytes = (alignBytes - ((size_t)ptr & (alignBytesMask))) & alignBytesMask;
    assert((alignBytes & alignBytesMask) == 0);
-    assert(bytes != ZSTD_CWKSP_ALIGNMENT_BYTES);
+    assert(bytes < alignBytes);
    return bytes;
 }

+/**
+ * Returns the initial value for allocStart which is used to determine the position from
+ * which we can allocate from the end of the workspace.
+ */
+MEM_STATIC void*  ZSTD_cwksp_initialAllocStart(ZSTD_cwksp* ws) {
+    return (void*)((size_t)ws->workspaceEnd & ~(ZSTD_CWKSP_ALIGNMENT_BYTES-1));
+}
+
 /**
 * Internal function. Do not use directly.
 * Reserves the given number of bytes within the aligned/buffer segment of the wksp,
@@ -281,27 +314,16 @@ ZSTD_cwksp_internal_advance_phase(ZSTD_cwksp* ws, ZSTD_cwksp_alloc_phase_e phase
 {
    assert(phase >= ws->phase);
    if (phase > ws->phase) {
-        /* Going from allocating objects to allocating buffers */
-        if (ws->phase < ZSTD_cwksp_alloc_buffers &&
-                phase >= ZSTD_cwksp_alloc_buffers) {
+        /* Going from allocating objects to allocating initOnce / tables */
+        if (ws->phase < ZSTD_cwksp_alloc_aligned_init_once &&
+            phase >= ZSTD_cwksp_alloc_aligned_init_once) {
            ws->tableValidEnd = ws->objectEnd;
-        }
+            ws->initOnceStart = ZSTD_cwksp_initialAllocStart(ws);

-        /* Going from allocating buffers to allocating aligneds/tables */
-        if (ws->phase < ZSTD_cwksp_alloc_aligned &&
-                phase >= ZSTD_cwksp_alloc_aligned) {
-            {   /* Align the start of the "aligned" to 64 bytes. Use [1, 64] bytes. */
-                size_t const bytesToAlign =
-                    ZSTD_CWKSP_ALIGNMENT_BYTES - ZSTD_cwksp_bytes_to_align_ptr(ws->allocStart, ZSTD_CWKSP_ALIGNMENT_BYTES);
-                DEBUGLOG(5, "reserving aligned alignment addtl space: %zu", bytesToAlign);
-                ZSTD_STATIC_ASSERT((ZSTD_CWKSP_ALIGNMENT_BYTES & (ZSTD_CWKSP_ALIGNMENT_BYTES - 1)) == 0); /* power of 2 */
-                RETURN_ERROR_IF(!ZSTD_cwksp_reserve_internal_buffer_space(ws, bytesToAlign),
-                                memory_allocation, "aligned phase - alignment initial allocation failed!");
-            }
            {   /* Align the start of the tables to 64 bytes. Use [0, 63] bytes */
-                void* const alloc = ws->objectEnd;
+                void *const alloc = ws->objectEnd;
                size_t const bytesToAlign = ZSTD_cwksp_bytes_to_align_ptr(alloc, ZSTD_CWKSP_ALIGNMENT_BYTES);
-                void* const objectEnd = (BYTE*)alloc + bytesToAlign;
+                void *const objectEnd = (BYTE *) alloc + bytesToAlign;
                DEBUGLOG(5, "reserving table alignment addtl space: %zu", bytesToAlign);
                RETURN_ERROR_IF(objectEnd > ws->workspaceEnd, memory_allocation,
                                "table phase - alignment initial allocation failed!");
@@ -309,7 +331,9 @@ ZSTD_cwksp_internal_advance_phase(ZSTD_cwksp* ws, ZSTD_cwksp_alloc_phase_e phase
                ws->tableEnd = objectEnd;  /* table area starts being empty */
                if (ws->tableValidEnd < ws->tableEnd) {
                    ws->tableValidEnd = ws->tableEnd;
-        }   }   }
+                }
+            }
+        }
        ws->phase = phase;
        ZSTD_cwksp_assert_internal_consistency(ws);
    }
@@ -321,7 +345,7 @@ ZSTD_cwksp_internal_advance_phase(ZSTD_cwksp* ws, ZSTD_cwksp_alloc_phase_e phase
 */
 MEM_STATIC int ZSTD_cwksp_owns_buffer(const ZSTD_cwksp* ws, const void* ptr)
 {
-    return (ptr != NULL) && (ws->workspace <= ptr) && (ptr <= ws->workspaceEnd);
+    return (ptr != NULL) && (ws->workspace <= ptr) && (ptr < ws->workspaceEnd);
 }

 /**
@@ -348,7 +372,9 @@ ZSTD_cwksp_reserve_internal(ZSTD_cwksp* ws, size_t bytes, ZSTD_cwksp_alloc_phase
    if (alloc) {
        alloc = (BYTE *)alloc + ZSTD_CWKSP_ASAN_REDZONE_SIZE;
        if (ws->isStatic == ZSTD_cwksp_dynamic_alloc) {
-            __asan_unpoison_memory_region(alloc, bytes);
+            /* We need to keep the redzone poisoned while unpoisoning the bytes that
+             * are actually allocated. */
+            __asan_unpoison_memory_region(alloc, bytes - 2 * ZSTD_CWKSP_ASAN_REDZONE_SIZE);
        }
    }
 #endif
@@ -364,6 +390,36 @@ MEM_STATIC BYTE* ZSTD_cwksp_reserve_buffer(ZSTD_cwksp* ws, size_t bytes)
    return (BYTE*)ZSTD_cwksp_reserve_internal(ws, bytes, ZSTD_cwksp_alloc_buffers);
 }

+/**
+ * Reserves and returns memory sized on and aligned on ZSTD_CWKSP_ALIGNMENT_BYTES (64 bytes).
+ * This memory has been initialized at least once in the past.
+ * This doesn't mean it has been initialized this time, and it might contain data from previous
+ * operations.
+ * The main usage is for algorithms that might need read access into uninitialized memory.
+ * The algorithm must maintain safety under these conditions and must make sure it doesn't
+ * leak any of the past data (directly or in side channels).
+ */
+MEM_STATIC void* ZSTD_cwksp_reserve_aligned_init_once(ZSTD_cwksp* ws, size_t bytes)
+{
+    size_t const alignedBytes = ZSTD_cwksp_align(bytes, ZSTD_CWKSP_ALIGNMENT_BYTES);
+    void* ptr = ZSTD_cwksp_reserve_internal(ws, alignedBytes, ZSTD_cwksp_alloc_aligned_init_once);
+    assert(((size_t)ptr & (ZSTD_CWKSP_ALIGNMENT_BYTES-1))== 0);
+    if(ptr && ptr < ws->initOnceStart) {
+        /* We assume the memory following the current allocation is either:
+         * 1. Not usable as initOnce memory (end of workspace)
+         * 2. Another initOnce buffer that has been allocated before (and so was previously memset)
+         * 3. An ASAN redzone, in which case we don't want to write on it
+         * For these reasons it should be fine to not explicitly zero every byte up to ws->initOnceStart.
+         * Note that we assume here that MSAN and ASAN cannot run in the same time. */
+        ZSTD_memset(ptr, 0, MIN((size_t)((U8*)ws->initOnceStart - (U8*)ptr), alignedBytes));
+        ws->initOnceStart = ptr;
+    }
+#if ZSTD_MEMORY_SANITIZER
+    assert(__msan_test_shadow(ptr, bytes) == -1);
+#endif
+    return ptr;
+}
+
 /**
 * Reserves and returns memory sized on and aligned on ZSTD_CWKSP_ALIGNMENT_BYTES (64 bytes).
 */
@@ -382,13 +438,17 @@ MEM_STATIC void* ZSTD_cwksp_reserve_aligned(ZSTD_cwksp* ws, size_t bytes)
 */
 MEM_STATIC void* ZSTD_cwksp_reserve_table(ZSTD_cwksp* ws, size_t bytes)
 {
-    const ZSTD_cwksp_alloc_phase_e phase = ZSTD_cwksp_alloc_aligned;
+    const ZSTD_cwksp_alloc_phase_e phase = ZSTD_cwksp_alloc_aligned_init_once;
    void* alloc;
    void* end;
    void* top;

-    if (ZSTD_isError(ZSTD_cwksp_internal_advance_phase(ws, phase))) {
-        return NULL;
+    /* We can only start allocating tables after we are done reserving space for objects at the
+     * start of the workspace */
+    if(ws->phase < phase) {
+        if (ZSTD_isError(ZSTD_cwksp_internal_advance_phase(ws, phase))) {
+            return NULL;
+        }
    }
    alloc = ws->tableEnd;
    end = (BYTE *)alloc + bytes;
@@ -467,11 +527,19 @@ MEM_STATIC void ZSTD_cwksp_mark_tables_dirty(ZSTD_cwksp* ws)
 #if ZSTD_MEMORY_SANITIZER && !defined (ZSTD_MSAN_DONT_POISON_WORKSPACE)
    /* To validate that the table re-use logic is sound, and that we don't
     * access table space that we haven't cleaned, we re-"poison" the table
-     * space every time we mark it dirty. */
+     * space every time we mark it dirty.
+     * Since tableValidEnd space and initOnce space may overlap we don't poison
+     * the initOnce portion as it break its promise. This means that this poisoning
+     * check isn't always applied fully. */
    {
        size_t size = (BYTE*)ws->tableValidEnd - (BYTE*)ws->objectEnd;
        assert(__msan_test_shadow(ws->objectEnd, size) == -1);
-        __msan_poison(ws->objectEnd, size);
+        if((BYTE*)ws->tableValidEnd < (BYTE*)ws->initOnceStart) {
+            __msan_poison(ws->objectEnd, size);
+        } else {
+            assert(ws->initOnceStart >= ws->objectEnd);
+            __msan_poison(ws->objectEnd, (BYTE*)ws->initOnceStart - (BYTE*)ws->objectEnd);
+        }
    }
 #endif

@@ -536,11 +604,14 @@ MEM_STATIC void ZSTD_cwksp_clear(ZSTD_cwksp* ws) {
 #if ZSTD_MEMORY_SANITIZER && !defined (ZSTD_MSAN_DONT_POISON_WORKSPACE)
    /* To validate that the context re-use logic is sound, and that we don't
     * access stuff that this compression hasn't initialized, we re-"poison"
-     * the workspace (or at least the non-static, non-table parts of it)
-     * every time we start a new compression. */
+     * the workspace except for the areas in which we expect memory re-use
+     * without initialization (objects, valid tables area and init once
+     * memory). */
    {
-        size_t size = (BYTE*)ws->workspaceEnd - (BYTE*)ws->tableValidEnd;
-        __msan_poison(ws->tableValidEnd, size);
+        if((BYTE*)ws->tableValidEnd < (BYTE*)ws->initOnceStart) {
+            size_t size = (BYTE*)ws->initOnceStart - (BYTE*)ws->tableValidEnd;
+            __msan_poison(ws->tableValidEnd, size);
+        }
    }
 #endif

@@ -556,10 +627,10 @@ MEM_STATIC void ZSTD_cwksp_clear(ZSTD_cwksp* ws) {
 #endif

    ws->tableEnd = ws->objectEnd;
-    ws->allocStart = ws->workspaceEnd;
+    ws->allocStart = ZSTD_cwksp_initialAllocStart(ws);
    ws->allocFailed = 0;
-    if (ws->phase > ZSTD_cwksp_alloc_buffers) {
-        ws->phase = ZSTD_cwksp_alloc_buffers;
+    if (ws->phase > ZSTD_cwksp_alloc_aligned_init_once) {
+        ws->phase = ZSTD_cwksp_alloc_aligned_init_once;
    }
    ZSTD_cwksp_assert_internal_consistency(ws);
 }
@@ -576,6 +647,7 @@ MEM_STATIC void ZSTD_cwksp_init(ZSTD_cwksp* ws, void* start, size_t size, ZSTD_c
    ws->workspaceEnd = (BYTE*)start + size;
    ws->objectEnd = ws->workspace;
    ws->tableValidEnd = ws->objectEnd;
+    ws->initOnceStart = ZSTD_cwksp_initialAllocStart(ws);
    ws->phase = ZSTD_cwksp_alloc_objects;
    ws->isStatic = isStatic;
    ZSTD_cwksp_clear(ws);
@@ -628,17 +700,11 @@ MEM_STATIC int ZSTD_cwksp_reserve_failed(const ZSTD_cwksp* ws) {
 * Returns if the estimated space needed for a wksp is within an acceptable limit of the
 * actual amount of space used.
 */
-MEM_STATIC int ZSTD_cwksp_estimated_space_within_bounds(const ZSTD_cwksp* const ws,
-                                                        size_t const estimatedSpace, int resizedWorkspace) {
-    if (resizedWorkspace) {
-        /* Resized/newly allocated wksp should have exact bounds */
-        return ZSTD_cwksp_used(ws) == estimatedSpace;
-    } else {
-        /* Due to alignment, when reusing a workspace, we can actually consume 63 fewer or more bytes
-         * than estimatedSpace. See the comments in zstd_cwksp.h for details.
-         */
-        return (ZSTD_cwksp_used(ws) >= estimatedSpace - 63) && (ZSTD_cwksp_used(ws) <= estimatedSpace + 63);
-    }
+MEM_STATIC int ZSTD_cwksp_estimated_space_within_bounds(const ZSTD_cwksp *const ws, size_t const estimatedSpace) {
+    /* We have an alignment space between objects and tables between tables and buffers, so we can have up to twice
+     * the alignment bytes difference between estimation and actual usage */
+    return (estimatedSpace - ZSTD_cwksp_slack_space_required()) <= ZSTD_cwksp_used(ws) &&
+           ZSTD_cwksp_used(ws) <= estimatedSpace;
 }


@@ -12,6 +12,8 @@
 #include "zstd_lazy.h"
 #include "../common/bits.h" /* ZSTD_countTrailingZeros64 */

+#define kLazySkippingStep 8
+

 /*-*************************************
 *  Binary Tree search
@@ -618,7 +620,7 @@ size_t ZSTD_dedicatedDictSearch_lazy_search(size_t* offsetPtr, size_t ml, U32 nb
 FORCE_INLINE_TEMPLATE U32 ZSTD_insertAndFindFirstIndex_internal(
                        ZSTD_matchState_t* ms,
                        const ZSTD_compressionParameters* const cParams,
-                        const BYTE* ip, U32 const mls)
+                        const BYTE* ip, U32 const mls, U32 const lazySkipping)
 {
    U32* const hashTable  = ms->hashTable;
    const U32 hashLog = cParams->hashLog;
@@ -633,6 +635,9 @@ FORCE_INLINE_TEMPLATE U32 ZSTD_insertAndFindFirstIndex_internal(
        NEXT_IN_CHAIN(idx, chainMask) = hashTable[h];
        hashTable[h] = idx;
        idx++;
+        /* Stop inserting every position when in the lazy skipping mode. */
+        if (lazySkipping)
+            break;
    }

    ms->nextToUpdate = target;
@@ -641,7 +646,7 @@ FORCE_INLINE_TEMPLATE U32 ZSTD_insertAndFindFirstIndex_internal(

 U32 ZSTD_insertAndFindFirstIndex(ZSTD_matchState_t* ms, const BYTE* ip) {
    const ZSTD_compressionParameters* const cParams = &ms->cParams;
-    return ZSTD_insertAndFindFirstIndex_internal(ms, cParams, ip, ms->cParams.minMatch);
+    return ZSTD_insertAndFindFirstIndex_internal(ms, cParams, ip, ms->cParams.minMatch, /* lazySkipping*/ 0);
 }

 /* inlining is important to hardwire a hot branch (template emulation) */
@@ -685,7 +690,7 @@ size_t ZSTD_HcFindBestMatch(
    }

    /* HC4 match finder */
-    matchIndex = ZSTD_insertAndFindFirstIndex_internal(ms, cParams, ip, mls);
+    matchIndex = ZSTD_insertAndFindFirstIndex_internal(ms, cParams, ip, mls, ms->lazySkipping);

    for ( ; (matchIndex>=lowLimit) & (nbAttempts>0) ; nbAttempts--) {
        size_t currentMl=0;
@@ -758,8 +763,6 @@ size_t ZSTD_HcFindBestMatch(
 * (SIMD) Row-based matchfinder
 ***********************************/
 /* Constants for row-based hash */
-#define ZSTD_ROW_HASH_TAG_OFFSET 16     /* byte offset of hashes in the match state's tagTable from the beginning of a row */
-#define ZSTD_ROW_HASH_TAG_BITS 8        /* nb bits to use for the tag */
 #define ZSTD_ROW_HASH_TAG_MASK ((1u << ZSTD_ROW_HASH_TAG_BITS) - 1)
 #define ZSTD_ROW_HASH_MAX_ENTRIES 64    /* absolute maximum number of entries per row, for all configurations */

@@ -775,39 +778,15 @@ MEM_STATIC U32 ZSTD_VecMask_next(ZSTD_VecMask val) {
    return ZSTD_countTrailingZeros64(val);
 }

-/* ZSTD_rotateRight_*():
- * Rotates a bitfield to the right by "count" bits.
- * https://en.wikipedia.org/w/index.php?title=Circular_shift&oldid=991635599#Implementing_circular_shifts
- */
-FORCE_INLINE_TEMPLATE
-U64 ZSTD_rotateRight_U64(U64 const value, U32 count) {
-    assert(count < 64);
-    count &= 0x3F; /* for fickle pattern recognition */
-    return (value >> count) | (U64)(value << ((0U - count) & 0x3F));
-}
-
-FORCE_INLINE_TEMPLATE
-U32 ZSTD_rotateRight_U32(U32 const value, U32 count) {
-    assert(count < 32);
-    count &= 0x1F; /* for fickle pattern recognition */
-    return (value >> count) | (U32)(value << ((0U - count) & 0x1F));
-}
-
-FORCE_INLINE_TEMPLATE
-U16 ZSTD_rotateRight_U16(U16 const value, U32 count) {
-    assert(count < 16);
-    count &= 0x0F; /* for fickle pattern recognition */
-    return (value >> count) | (U16)(value << ((0U - count) & 0x0F));
-}
-
 /* ZSTD_row_nextIndex():
 * Returns the next index to insert at within a tagTable row, and updates the "head"
- * value to reflect the update. Essentially cycles backwards from [0, {entries per row})
+ * value to reflect the update. Essentially cycles backwards from [1, {entries per row})
 */
 FORCE_INLINE_TEMPLATE U32 ZSTD_row_nextIndex(BYTE* const tagRow, U32 const rowMask) {
-  U32 const next = (*tagRow - 1) & rowMask;
-  *tagRow = (BYTE)next;
-  return next;
+    U32 next = (*tagRow-1) & rowMask;
+    next += (next == 0) ? rowMask : 0; /* skip first position */
+    *tagRow = (BYTE)next;
+    return next;
 }

 /* ZSTD_isAligned():
@@ -821,7 +800,7 @@ MEM_STATIC int ZSTD_isAligned(void const* ptr, size_t align) {
 /* ZSTD_row_prefetch():
 * Performs prefetching for the hashTable and tagTable at a given row.
 */
-FORCE_INLINE_TEMPLATE void ZSTD_row_prefetch(U32 const* hashTable, U16 const* tagTable, U32 const relRow, U32 const rowLog) {
+FORCE_INLINE_TEMPLATE void ZSTD_row_prefetch(U32 const* hashTable, BYTE const* tagTable, U32 const relRow, U32 const rowLog) {
    PREFETCH_L1(hashTable + relRow);
    if (rowLog >= 5) {
        PREFETCH_L1(hashTable + relRow + 16);
@@ -845,13 +824,13 @@ FORCE_INLINE_TEMPLATE void ZSTD_row_fillHashCache(ZSTD_matchState_t* ms, const B
                                   U32 idx, const BYTE* const iLimit)
 {
    U32 const* const hashTable = ms->hashTable;
-    U16 const* const tagTable = ms->tagTable;
+    BYTE const* const tagTable = ms->tagTable;
    U32 const hashLog = ms->rowHashLog;
    U32 const maxElemsToPrefetch = (base + idx) > iLimit ? 0 : (U32)(iLimit - (base + idx) + 1);
    U32 const lim = idx + MIN(ZSTD_ROW_HASH_CACHE_SIZE, maxElemsToPrefetch);

    for (; idx < lim; ++idx) {
-        U32 const hash = (U32)ZSTD_hashPtr(base + idx, hashLog + ZSTD_ROW_HASH_TAG_BITS, mls);
+        U32 const hash = (U32)ZSTD_hashPtrSalted(base + idx, hashLog + ZSTD_ROW_HASH_TAG_BITS, mls, ms->hashSalt);
        U32 const row = (hash >> ZSTD_ROW_HASH_TAG_BITS) << rowLog;
        ZSTD_row_prefetch(hashTable, tagTable, row, rowLog);
        ms->hashCache[idx & ZSTD_ROW_HASH_CACHE_MASK] = hash;
@@ -867,11 +846,12 @@ FORCE_INLINE_TEMPLATE void ZSTD_row_fillHashCache(ZSTD_matchState_t* ms, const B
 * base + idx + ZSTD_ROW_HASH_CACHE_SIZE. Also prefetches the appropriate rows from hashTable and tagTable.
 */
 FORCE_INLINE_TEMPLATE U32 ZSTD_row_nextCachedHash(U32* cache, U32 const* hashTable,
-                                                  U16 const* tagTable, BYTE const* base,
+                                                  BYTE const* tagTable, BYTE const* base,
                                                  U32 idx, U32 const hashLog,
-                                                  U32 const rowLog, U32 const mls)
+                                                  U32 const rowLog, U32 const mls,
+                                                  U64 const hashSalt)
 {
-    U32 const newHash = (U32)ZSTD_hashPtr(base+idx+ZSTD_ROW_HASH_CACHE_SIZE, hashLog + ZSTD_ROW_HASH_TAG_BITS, mls);
+    U32 const newHash = (U32)ZSTD_hashPtrSalted(base+idx+ZSTD_ROW_HASH_CACHE_SIZE, hashLog + ZSTD_ROW_HASH_TAG_BITS, mls, hashSalt);
    U32 const row = (newHash >> ZSTD_ROW_HASH_TAG_BITS) << rowLog;
    ZSTD_row_prefetch(hashTable, tagTable, row, rowLog);
    {   U32 const hash = cache[idx & ZSTD_ROW_HASH_CACHE_MASK];
@@ -889,22 +869,21 @@ FORCE_INLINE_TEMPLATE void ZSTD_row_update_internalImpl(ZSTD_matchState_t* ms,
                                                        U32 const rowMask, U32 const useCache)
 {
    U32* const hashTable = ms->hashTable;
-    U16* const tagTable = ms->tagTable;
+    BYTE* const tagTable = ms->tagTable;
    U32 const hashLog = ms->rowHashLog;
    const BYTE* const base = ms->window.base;

    DEBUGLOG(6, "ZSTD_row_update_internalImpl(): updateStartIdx=%u, updateEndIdx=%u", updateStartIdx, updateEndIdx);
    for (; updateStartIdx < updateEndIdx; ++updateStartIdx) {
-        U32 const hash = useCache ? ZSTD_row_nextCachedHash(ms->hashCache, hashTable, tagTable, base, updateStartIdx, hashLog, rowLog, mls)
-                                  : (U32)ZSTD_hashPtr(base + updateStartIdx, hashLog + ZSTD_ROW_HASH_TAG_BITS, mls);
+        U32 const hash = useCache ? ZSTD_row_nextCachedHash(ms->hashCache, hashTable, tagTable, base, updateStartIdx, hashLog, rowLog, mls, ms->hashSalt)
+                                  : (U32)ZSTD_hashPtrSalted(base + updateStartIdx, hashLog + ZSTD_ROW_HASH_TAG_BITS, mls, ms->hashSalt);
        U32 const relRow = (hash >> ZSTD_ROW_HASH_TAG_BITS) << rowLog;
        U32* const row = hashTable + relRow;
-        BYTE* tagRow = (BYTE*)(tagTable + relRow);  /* Though tagTable is laid out as a table of U16, each tag is only 1 byte.
-                                                       Explicit cast allows us to get exact desired position within each row */
+        BYTE* tagRow = tagTable + relRow;
        U32 const pos = ZSTD_row_nextIndex(tagRow, rowMask);

-        assert(hash == ZSTD_hashPtr(base + updateStartIdx, hashLog + ZSTD_ROW_HASH_TAG_BITS, mls));
-        ((BYTE*)tagRow)[pos + ZSTD_ROW_HASH_TAG_OFFSET] = hash & ZSTD_ROW_HASH_TAG_MASK;
+        assert(hash == ZSTD_hashPtrSalted(base + updateStartIdx, hashLog + ZSTD_ROW_HASH_TAG_BITS, mls, ms->hashSalt));
+        tagRow[pos] = hash & ZSTD_ROW_HASH_TAG_MASK;
        row[pos] = updateStartIdx;
    }
 }
@@ -1060,7 +1039,7 @@ ZSTD_row_getNEONMask(const U32 rowEntries, const BYTE* const src, const BYTE tag
 FORCE_INLINE_TEMPLATE ZSTD_VecMask
 ZSTD_row_getMatchMask(const BYTE* const tagRow, const BYTE tag, const U32 headGrouped, const U32 rowEntries)
 {
-    const BYTE* const src = tagRow + ZSTD_ROW_HASH_TAG_OFFSET;
+    const BYTE* const src = tagRow;
    assert((rowEntries == 16) || (rowEntries == 32) || rowEntries == 64);
    assert(rowEntries <= ZSTD_ROW_HASH_MAX_ENTRIES);
    assert(ZSTD_row_matchMaskGroupWidth(rowEntries) * rowEntries <= sizeof(ZSTD_VecMask) * 8);
@@ -1145,7 +1124,7 @@ size_t ZSTD_RowFindBestMatch(
                        const U32 rowLog)
 {
    U32* const hashTable = ms->hashTable;
-    U16* const tagTable = ms->tagTable;
+    BYTE* const tagTable = ms->tagTable;
    U32* const hashCache = ms->hashCache;
    const U32 hashLog = ms->rowHashLog;
    const ZSTD_compressionParameters* const cParams = &ms->cParams;
@@ -1164,8 +1143,10 @@ size_t ZSTD_RowFindBestMatch(
    const U32 rowMask = rowEntries - 1;
    const U32 cappedSearchLog = MIN(cParams->searchLog, rowLog); /* nb of searches is capped at nb entries per row */
    const U32 groupWidth = ZSTD_row_matchMaskGroupWidth(rowEntries);
+    const U64 hashSalt = ms->hashSalt;
    U32 nbAttempts = 1U << cappedSearchLog;
    size_t ml=4-1;
+    U32 hash;

    /* DMS/DDS variables that may be referenced laster */
    const ZSTD_matchState_t* const dms = ms->dictMatchState;
@@ -1189,7 +1170,7 @@ size_t ZSTD_RowFindBestMatch(
    if (dictMode == ZSTD_dictMatchState) {
        /* Prefetch DMS rows */
        U32* const dmsHashTable = dms->hashTable;
-        U16* const dmsTagTable = dms->tagTable;
+        BYTE* const dmsTagTable = dms->tagTable;
        U32 const dmsHash = (U32)ZSTD_hashPtr(ip, dms->rowHashLog + ZSTD_ROW_HASH_TAG_BITS, mls);
        U32 const dmsRelRow = (dmsHash >> ZSTD_ROW_HASH_TAG_BITS) << rowLog;
        dmsTag = dmsHash & ZSTD_ROW_HASH_TAG_MASK;
@@ -1199,9 +1180,19 @@ size_t ZSTD_RowFindBestMatch(
    }

    /* Update the hashTable and tagTable up to (but not including) ip */
-    ZSTD_row_update_internal(ms, ip, mls, rowLog, rowMask, 1 /* useCache */);
+    if (!ms->lazySkipping) {
+        ZSTD_row_update_internal(ms, ip, mls, rowLog, rowMask, 1 /* useCache */);
+        hash = ZSTD_row_nextCachedHash(hashCache, hashTable, tagTable, base, curr, hashLog, rowLog, mls, hashSalt);
+    } else {
+        /* Stop inserting every position when in the lazy skipping mode.
+         * The hash cache is also not kept up to date in this mode.
+         */
+        hash = (U32)ZSTD_hashPtrSalted(ip, hashLog + ZSTD_ROW_HASH_TAG_BITS, mls, hashSalt);
+        ms->nextToUpdate = curr;
+    }
+    ms->hashSaltEntropy += hash; /* collect salt entropy */
+
    {   /* Get the hash for ip, compute the appropriate row */
-        U32 const hash = ZSTD_row_nextCachedHash(hashCache, hashTable, tagTable, base, curr, hashLog, rowLog, mls);
        U32 const relRow = (hash >> ZSTD_ROW_HASH_TAG_BITS) << rowLog;
        U32 const tag = hash & ZSTD_ROW_HASH_TAG_MASK;
        U32* const row = hashTable + relRow;
@@ -1213,9 +1204,10 @@ size_t ZSTD_RowFindBestMatch(
        ZSTD_VecMask matches = ZSTD_row_getMatchMask(tagRow, (BYTE)tag, headGrouped, rowEntries);

        /* Cycle through the matches and prefetch */
-        for (; (matches > 0) && (nbAttempts > 0); --nbAttempts, matches &= (matches - 1)) {
+        for (; (matches > 0) && (nbAttempts > 0); matches &= (matches - 1)) {
            U32 const matchPos = ((headGrouped + ZSTD_VecMask_next(matches)) / groupWidth) & rowMask;
            U32 const matchIndex = row[matchPos];
+            if(matchPos == 0) continue;
            assert(numMatches < rowEntries);
            if (matchIndex < lowLimit)
                break;
@@ -1225,13 +1217,14 @@ size_t ZSTD_RowFindBestMatch(
                PREFETCH_L1(dictBase + matchIndex);
            }
            matchBuffer[numMatches++] = matchIndex;
+            --nbAttempts;
        }

        /* Speed opt: insert current byte into hashtable too. This allows us to avoid one iteration of the loop
           in ZSTD_row_update_internal() at the next search. */
        {
            U32 const pos = ZSTD_row_nextIndex(tagRow, rowMask);
-            tagRow[pos + ZSTD_ROW_HASH_TAG_OFFSET] = (BYTE)tag;
+            tagRow[pos] = (BYTE)tag;
            row[pos] = ms->nextToUpdate++;
        }

@@ -1282,13 +1275,15 @@ size_t ZSTD_RowFindBestMatch(
            size_t currMatch = 0;
            ZSTD_VecMask matches = ZSTD_row_getMatchMask(dmsTagRow, (BYTE)dmsTag, headGrouped, rowEntries);

-            for (; (matches > 0) && (nbAttempts > 0); --nbAttempts, matches &= (matches - 1)) {
+            for (; (matches > 0) && (nbAttempts > 0); matches &= (matches - 1)) {
                U32 const matchPos = ((headGrouped + ZSTD_VecMask_next(matches)) / groupWidth) & rowMask;
                U32 const matchIndex = dmsRow[matchPos];
+                if(matchPos == 0) continue;
                if (matchIndex < dmsLowestIndex)
                    break;
                PREFETCH_L1(dmsBase + matchIndex);
                matchBuffer[numMatches++] = matchIndex;
+                --nbAttempts;
            }

            /* Return the longest match */
@@ -1545,10 +1540,11 @@ ZSTD_compressBlock_lazy_generic(
        assert(offset_2 <= dictAndPrefixLength);
    }

+    /* Reset the lazy skipping state */
+    ms->lazySkipping = 0;
+
    if (searchMethod == search_rowHash) {
-        ZSTD_row_fillHashCache(ms, base, rowLog,
-                            MIN(ms->cParams.minMatch, 6 /* mls caps out at 6 */),
-                            ms->nextToUpdate, ilimit);
+        ZSTD_row_fillHashCache(ms, base, rowLog, mls, ms->nextToUpdate, ilimit);
    }

    /* Match Loop */
@@ -1592,7 +1588,16 @@ ZSTD_compressBlock_lazy_generic(
        }

        if (matchLength < 4) {
-            ip += ((ip-anchor) >> kSearchStrength) + 1;   /* jump faster over incompressible sections */
+            size_t const step = ((size_t)(ip-anchor) >> kSearchStrength) + 1;   /* jump faster over incompressible sections */;
+            ip += step;
+            /* Enter the lazy skipping mode once we are skipping more than 8 bytes at a time.
+             * In this mode we stop inserting every position into our tables, and only insert
+             * positions that we search, which is one in step positions.
+             * The exact cutoff is flexible, I've just chosen a number that is reasonably high,
+             * so we minimize the compression ratio loss in "normal" scenarios. This mode gets
+             * triggered once we've gone 2KB without finding any matches.
+             */
+            ms->lazySkipping = step > kLazySkippingStep;
            continue;
        }

@@ -1696,6 +1701,13 @@ _storeSequence:
            ZSTD_storeSeq(seqStore, litLength, anchor, iend, (U32)offBase, matchLength);
            anchor = ip = start + matchLength;
        }
+        if (ms->lazySkipping) {
+            /* We've found a match, disable lazy skipping mode, and refill the hash cache. */
+            if (searchMethod == search_rowHash) {
+                ZSTD_row_fillHashCache(ms, base, rowLog, mls, ms->nextToUpdate, ilimit);
+            }
+            ms->lazySkipping = 0;
+        }

        /* check immediate repcode */
        if (isDxS) {
@@ -1913,12 +1925,13 @@ size_t ZSTD_compressBlock_lazy_extDict_generic(

    DEBUGLOG(5, "ZSTD_compressBlock_lazy_extDict_generic (searchFunc=%u)", (U32)searchMethod);

+    /* Reset the lazy skipping state */
+    ms->lazySkipping = 0;
+
    /* init */
    ip += (ip == prefixStart);
    if (searchMethod == search_rowHash) {
-        ZSTD_row_fillHashCache(ms, base, rowLog,
-                               MIN(ms->cParams.minMatch, 6 /* mls caps out at 6 */),
-                               ms->nextToUpdate, ilimit);
+        ZSTD_row_fillHashCache(ms, base, rowLog, mls, ms->nextToUpdate, ilimit);
    }

    /* Match Loop */
@@ -1956,7 +1969,16 @@ size_t ZSTD_compressBlock_lazy_extDict_generic(
        }

        if (matchLength < 4) {
-            ip += ((ip-anchor) >> kSearchStrength) + 1;   /* jump faster over incompressible sections */
+            size_t const step = ((size_t)(ip-anchor) >> kSearchStrength);
+            ip += step + 1;   /* jump faster over incompressible sections */
+            /* Enter the lazy skipping mode once we are skipping more than 8 bytes at a time.
+             * In this mode we stop inserting every position into our tables, and only insert
+             * positions that we search, which is one in step positions.
+             * The exact cutoff is flexible, I've just chosen a number that is reasonably high,
+             * so we minimize the compression ratio loss in "normal" scenarios. This mode gets
+             * triggered once we've gone 2KB without finding any matches.
+             */
+            ms->lazySkipping = step > kLazySkippingStep;
            continue;
        }

@@ -2042,6 +2064,13 @@ _storeSequence:
            ZSTD_storeSeq(seqStore, litLength, anchor, iend, (U32)offBase, matchLength);
            anchor = ip = start + matchLength;
        }
+        if (ms->lazySkipping) {
+            /* We've found a match, disable lazy skipping mode, and refill the hash cache. */
+            if (searchMethod == search_rowHash) {
+                ZSTD_row_fillHashCache(ms, base, rowLog, mls, ms->nextToUpdate, ilimit);
+            }
+            ms->lazySkipping = 0;
+        }

        /* check immediate repcode */
        while (ip <= ilimit) {
@@ -25,6 +25,8 @@ extern "C" {
 */
 #define ZSTD_LAZY_DDSS_BUCKET_LOG 2

+#define ZSTD_ROW_HASH_TAG_BITS 8        /* nb bits to use for the tag */
+
 U32 ZSTD_insertAndFindFirstIndex(ZSTD_matchState_t* ms, const BYTE* ip);
 void ZSTD_row_update(ZSTD_matchState_t* const ms, const BYTE* ip);

@@ -116,7 +118,7 @@ size_t ZSTD_compressBlock_lazy2_extDict_row(
 size_t ZSTD_compressBlock_btlazy2_extDict(
        ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
        void const* src, size_t srcSize);
-        
+

 #if defined (__cplusplus)
 }
@@ -1086,6 +1086,8 @@ ZSTD_compressBlock_opt_generic(ZSTD_matchState_t* ms,
    ZSTD_optimal_t lastSequence;
    ZSTD_optLdm_t optLdm;

+    ZSTD_memset(&lastSequence, 0, sizeof(ZSTD_optimal_t));
+
    optLdm.seqStore = ms->ldmSeqStore ? *ms->ldmSeqStore : kNullRawSeqStore;
    optLdm.endPosInBlock = optLdm.startPosInBlock = optLdm.offset = 0;
    ZSTD_opt_getNextMatchAndUpdateSeqStore(&optLdm, (U32)(ip-istart), (U32)(iend-ip));
@@ -20,6 +20,7 @@


 /* ======   Dependencies   ====== */
+#include "../common/allocations.h"  /* ZSTD_customMalloc, ZSTD_customCalloc, ZSTD_customFree */
 #include "../common/zstd_deps.h"   /* ZSTD_memcpy, ZSTD_memset, INT_MAX, UINT_MAX */
 #include "../common/mem.h"         /* MEM_STATIC */
 #include "../common/pool.h"        /* threadpool */
@@ -719,7 +720,7 @@ static void ZSTDMT_compressionJob(void* jobDescription)
    ZSTDMT_serialState_update(job->serial, cctx, rawSeqStore, job->src, job->jobID);

    if (!job->firstJob) {  /* flush and overwrite frame header when it's not first job */
-        size_t const hSize = ZSTD_compressContinue(cctx, dstBuff.start, dstBuff.capacity, job->src.start, 0);
+        size_t const hSize = ZSTD_compressContinue_public(cctx, dstBuff.start, dstBuff.capacity, job->src.start, 0);
        if (ZSTD_isError(hSize)) JOB_ERROR(hSize);
        DEBUGLOG(5, "ZSTDMT_compressionJob: flush and overwrite %u bytes of frame header (not first job)", (U32)hSize);
        ZSTD_invalidateRepCodes(cctx);
@@ -737,7 +738,7 @@ static void ZSTDMT_compressionJob(void* jobDescription)
        DEBUGLOG(5, "ZSTDMT_compressionJob: compress %u bytes in %i blocks", (U32)job->src.size, nbChunks);
        assert(job->cSize == 0);
        for (chunkNb = 1; chunkNb < nbChunks; chunkNb++) {
-            size_t const cSize = ZSTD_compressContinue(cctx, op, oend-op, ip, chunkSize);
+            size_t const cSize = ZSTD_compressContinue_public(cctx, op, oend-op, ip, chunkSize);
            if (ZSTD_isError(cSize)) JOB_ERROR(cSize);
            ip += chunkSize;
            op += cSize; assert(op < oend);
@@ -757,8 +758,8 @@ static void ZSTDMT_compressionJob(void* jobDescription)
            size_t const lastBlockSize1 = job->src.size & (chunkSize-1);
            size_t const lastBlockSize = ((lastBlockSize1==0) & (job->src.size>=chunkSize)) ? chunkSize : lastBlockSize1;
            size_t const cSize = (job->lastJob) ?
-                 ZSTD_compressEnd     (cctx, op, oend-op, ip, lastBlockSize) :
-                 ZSTD_compressContinue(cctx, op, oend-op, ip, lastBlockSize);
+                 ZSTD_compressEnd_public(cctx, op, oend-op, ip, lastBlockSize) :
+                 ZSTD_compressContinue_public(cctx, op, oend-op, ip, lastBlockSize);
            if (ZSTD_isError(cSize)) JOB_ERROR(cSize);
            lastCBlockSize = cSize;
    }   }
@@ -30,14 +30,14 @@
 * TODO: Support Windows calling convention.
 */

-ZSTD_HIDE_ASM_FUNCTION(HUF_decompress4X1_usingDTable_internal_bmi2_asm_loop)
-ZSTD_HIDE_ASM_FUNCTION(HUF_decompress4X2_usingDTable_internal_bmi2_asm_loop)
-ZSTD_HIDE_ASM_FUNCTION(_HUF_decompress4X2_usingDTable_internal_bmi2_asm_loop)
-ZSTD_HIDE_ASM_FUNCTION(_HUF_decompress4X1_usingDTable_internal_bmi2_asm_loop)
-.global HUF_decompress4X1_usingDTable_internal_bmi2_asm_loop
-.global HUF_decompress4X2_usingDTable_internal_bmi2_asm_loop
-.global _HUF_decompress4X1_usingDTable_internal_bmi2_asm_loop
-.global _HUF_decompress4X2_usingDTable_internal_bmi2_asm_loop
+ZSTD_HIDE_ASM_FUNCTION(HUF_decompress4X1_usingDTable_internal_fast_asm_loop)
+ZSTD_HIDE_ASM_FUNCTION(HUF_decompress4X2_usingDTable_internal_fast_asm_loop)
+ZSTD_HIDE_ASM_FUNCTION(_HUF_decompress4X2_usingDTable_internal_fast_asm_loop)
+ZSTD_HIDE_ASM_FUNCTION(_HUF_decompress4X1_usingDTable_internal_fast_asm_loop)
+.global HUF_decompress4X1_usingDTable_internal_fast_asm_loop
+.global HUF_decompress4X2_usingDTable_internal_fast_asm_loop
+.global _HUF_decompress4X1_usingDTable_internal_fast_asm_loop
+.global _HUF_decompress4X2_usingDTable_internal_fast_asm_loop
 .text

 /* Sets up register mappings for clarity.
@@ -95,8 +95,9 @@ ZSTD_HIDE_ASM_FUNCTION(_HUF_decompress4X1_usingDTable_internal_bmi2_asm_loop)
 /* Define both _HUF_* & HUF_* symbols because MacOS
 * C symbols are prefixed with '_' & Linux symbols aren't.
 */
-_HUF_decompress4X1_usingDTable_internal_bmi2_asm_loop:
-HUF_decompress4X1_usingDTable_internal_bmi2_asm_loop:
+_HUF_decompress4X1_usingDTable_internal_fast_asm_loop:
+HUF_decompress4X1_usingDTable_internal_fast_asm_loop:
+    ZSTD_CET_ENDBRANCH
    /* Save all registers - even if they are callee saved for simplicity. */
    push %rax
    push %rbx
@@ -350,8 +351,9 @@ HUF_decompress4X1_usingDTable_internal_bmi2_asm_loop:
    pop %rax
    ret

-_HUF_decompress4X2_usingDTable_internal_bmi2_asm_loop:
-HUF_decompress4X2_usingDTable_internal_bmi2_asm_loop:
+_HUF_decompress4X2_usingDTable_internal_fast_asm_loop:
+HUF_decompress4X2_usingDTable_internal_fast_asm_loop:
+    ZSTD_CET_ENDBRANCH
    /* Save all registers - even if they are callee saved for simplicity. */
    push %rax
    push %rbx
@@ -14,12 +14,12 @@
 /*-*******************************************************
 *  Dependencies
 *********************************************************/
+#include "../common/allocations.h"  /* ZSTD_customMalloc, ZSTD_customFree */
 #include "../common/zstd_deps.h"   /* ZSTD_memcpy, ZSTD_memmove, ZSTD_memset */
 #include "../common/cpu.h"         /* bmi2 */
 #include "../common/mem.h"         /* low level memory routines */
 #define FSE_STATIC_LINKING_ONLY
 #include "../common/fse.h"
-#define HUF_STATIC_LINKING_ONLY
 #include "../common/huf.h"
 #include "zstd_decompress_internal.h"
 #include "zstd_ddict.h"
@@ -55,11 +55,11 @@
 /*-*******************************************************
 *  Dependencies
 *********************************************************/
+#include "../common/allocations.h"  /* ZSTD_customMalloc, ZSTD_customCalloc, ZSTD_customFree */
 #include "../common/zstd_deps.h"   /* ZSTD_memcpy, ZSTD_memmove, ZSTD_memset */
 #include "../common/mem.h"         /* low level memory routines */
 #define FSE_STATIC_LINKING_ONLY
 #include "../common/fse.h"
-#define HUF_STATIC_LINKING_ONLY
 #include "../common/huf.h"
 #include "../common/xxhash.h" /* XXH64_reset, XXH64_update, XXH64_digest, XXH64 */
 #include "../common/zstd_internal.h"  /* blockProperties_t */
@@ -244,6 +244,7 @@ static void ZSTD_DCtx_resetParameters(ZSTD_DCtx* dctx)
    dctx->outBufferMode = ZSTD_bm_buffered;
    dctx->forceIgnoreChecksum = ZSTD_d_validateChecksum;
    dctx->refMultipleDDicts = ZSTD_rmd_refSingleDDict;
+    dctx->disableHufAsm = 0;
 }

 static void ZSTD_initDCtx_internal(ZSTD_DCtx* dctx)
@@ -588,49 +589,52 @@ static size_t readSkippableFrameSize(void const* src, size_t srcSize)
    sizeU32 = MEM_readLE32((BYTE const*)src + ZSTD_FRAMEIDSIZE);
    RETURN_ERROR_IF((U32)(sizeU32 + ZSTD_SKIPPABLEHEADERSIZE) < sizeU32,
                    frameParameter_unsupported, "");
-    {
-        size_t const skippableSize = skippableHeaderSize + sizeU32;
+    {   size_t const skippableSize = skippableHeaderSize + sizeU32;
        RETURN_ERROR_IF(skippableSize > srcSize, srcSize_wrong, "");
        return skippableSize;
    }
 }

 /*! ZSTD_readSkippableFrame() :
- * Retrieves a zstd skippable frame containing data given by src, and writes it to dst buffer.
+ * Retrieves content of a skippable frame, and writes it to dst buffer.
 *
 * The parameter magicVariant will receive the magicVariant that was supplied when the frame was written,
 * i.e. magicNumber - ZSTD_MAGIC_SKIPPABLE_START.  This can be NULL if the caller is not interested
 * in the magicVariant.
 *
- * Returns an error if destination buffer is not large enough, or if the frame is not skippable.
+ * Returns an error if destination buffer is not large enough, or if this is not a valid skippable frame.
 *
 * @return : number of bytes written or a ZSTD error.
 */
-ZSTDLIB_API size_t ZSTD_readSkippableFrame(void* dst, size_t dstCapacity, unsigned* magicVariant,
-                                            const void* src, size_t srcSize)
+size_t ZSTD_readSkippableFrame(void* dst, size_t dstCapacity,
+                               unsigned* magicVariant,  /* optional, can be NULL */
+                         const void* src, size_t srcSize)
 {
-    U32 const magicNumber = MEM_readLE32(src);
-    size_t skippableFrameSize = readSkippableFrameSize(src, srcSize);
-    size_t skippableContentSize = skippableFrameSize - ZSTD_SKIPPABLEHEADERSIZE;
+    RETURN_ERROR_IF(srcSize < ZSTD_SKIPPABLEHEADERSIZE, srcSize_wrong, "");

-    /* check input validity */
-    RETURN_ERROR_IF(!ZSTD_isSkippableFrame(src, srcSize), frameParameter_unsupported, "");
-    RETURN_ERROR_IF(skippableFrameSize < ZSTD_SKIPPABLEHEADERSIZE || skippableFrameSize > srcSize, srcSize_wrong, "");
-    RETURN_ERROR_IF(skippableContentSize > dstCapacity, dstSize_tooSmall, "");
+    {   U32 const magicNumber = MEM_readLE32(src);
+        size_t skippableFrameSize = readSkippableFrameSize(src, srcSize);
+        size_t skippableContentSize = skippableFrameSize - ZSTD_SKIPPABLEHEADERSIZE;

-    /* deliver payload */
-    if (skippableContentSize > 0  && dst != NULL)
-        ZSTD_memcpy(dst, (const BYTE *)src + ZSTD_SKIPPABLEHEADERSIZE, skippableContentSize);
-    if (magicVariant != NULL)
-        *magicVariant = magicNumber - ZSTD_MAGIC_SKIPPABLE_START;
-    return skippableContentSize;
+        /* check input validity */
+        RETURN_ERROR_IF(!ZSTD_isSkippableFrame(src, srcSize), frameParameter_unsupported, "");
+        RETURN_ERROR_IF(skippableFrameSize < ZSTD_SKIPPABLEHEADERSIZE || skippableFrameSize > srcSize, srcSize_wrong, "");
+        RETURN_ERROR_IF(skippableContentSize > dstCapacity, dstSize_tooSmall, "");
+
+        /* deliver payload */
+        if (skippableContentSize > 0  && dst != NULL)
+            ZSTD_memcpy(dst, (const BYTE *)src + ZSTD_SKIPPABLEHEADERSIZE, skippableContentSize);
+        if (magicVariant != NULL)
+            *magicVariant = magicNumber - ZSTD_MAGIC_SKIPPABLE_START;
+        return skippableContentSize;
+    }
 }

 /** ZSTD_findDecompressedSize() :
- *  compatible with legacy mode
 *  `srcSize` must be the exact length of some number of ZSTD compressed and/or
 *      skippable frames
- *  @return : decompressed size of the frames contained */
+ *  note: compatible with legacy mode
+ * @return : decompressed size of the frames contained */
 unsigned long long ZSTD_findDecompressedSize(const void* src, size_t srcSize)
 {
    unsigned long long totalDstSize = 0;
@@ -640,9 +644,7 @@ unsigned long long ZSTD_findDecompressedSize(const void* src, size_t srcSize)

        if ((magicNumber & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {
            size_t const skippableSize = readSkippableFrameSize(src, srcSize);
-            if (ZSTD_isError(skippableSize)) {
-                return ZSTD_CONTENTSIZE_ERROR;
-            }
+            if (ZSTD_isError(skippableSize)) return ZSTD_CONTENTSIZE_ERROR;
            assert(skippableSize <= srcSize);

            src = (const BYTE *)src + skippableSize;
@@ -650,17 +652,17 @@ unsigned long long ZSTD_findDecompressedSize(const void* src, size_t srcSize)
            continue;
        }

-        {   unsigned long long const ret = ZSTD_getFrameContentSize(src, srcSize);
-            if (ret >= ZSTD_CONTENTSIZE_ERROR) return ret;
+        {   unsigned long long const fcs = ZSTD_getFrameContentSize(src, srcSize);
+            if (fcs >= ZSTD_CONTENTSIZE_ERROR) return fcs;

-            /* check for overflow */
-            if (totalDstSize + ret < totalDstSize) return ZSTD_CONTENTSIZE_ERROR;
-            totalDstSize += ret;
+            if (totalDstSize + fcs < totalDstSize)
+                return ZSTD_CONTENTSIZE_ERROR; /* check for overflow */
+            totalDstSize += fcs;
        }
+        /* skip to next frame */
        {   size_t const frameSrcSize = ZSTD_findFrameCompressedSize(src, srcSize);
-            if (ZSTD_isError(frameSrcSize)) {
-                return ZSTD_CONTENTSIZE_ERROR;
-            }
+            if (ZSTD_isError(frameSrcSize)) return ZSTD_CONTENTSIZE_ERROR;
+            assert(frameSrcSize <= srcSize);

            src = (const BYTE *)src + frameSrcSize;
            srcSize -= frameSrcSize;
@@ -782,6 +784,7 @@ static ZSTD_frameSizeInfo ZSTD_findFrameSizeInfo(const void* src, size_t srcSize
            ip += 4;
        }

+        frameSizeInfo.nbBlocks = nbBlocks;
        frameSizeInfo.compressedSize = (size_t)(ip - ipstart);
        frameSizeInfo.decompressedBound = (zfh.frameContentSize != ZSTD_CONTENTSIZE_UNKNOWN)
                                        ? zfh.frameContentSize
@@ -825,6 +828,48 @@ unsigned long long ZSTD_decompressBound(const void* src, size_t srcSize)
    return bound;
 }

+size_t ZSTD_decompressionMargin(void const* src, size_t srcSize)
+{
+    size_t margin = 0;
+    unsigned maxBlockSize = 0;
+
+    /* Iterate over each frame */
+    while (srcSize > 0) {
+        ZSTD_frameSizeInfo const frameSizeInfo = ZSTD_findFrameSizeInfo(src, srcSize);
+        size_t const compressedSize = frameSizeInfo.compressedSize;
+        unsigned long long const decompressedBound = frameSizeInfo.decompressedBound;
+        ZSTD_frameHeader zfh;
+
+        FORWARD_IF_ERROR(ZSTD_getFrameHeader(&zfh, src, srcSize), "");
+        if (ZSTD_isError(compressedSize) || decompressedBound == ZSTD_CONTENTSIZE_ERROR)
+            return ERROR(corruption_detected);
+
+        if (zfh.frameType == ZSTD_frame) {
+            /* Add the frame header to our margin */
+            margin += zfh.headerSize;
+            /* Add the checksum to our margin */
+            margin += zfh.checksumFlag ? 4 : 0;
+            /* Add 3 bytes per block */
+            margin += 3 * frameSizeInfo.nbBlocks;
+
+            /* Compute the max block size */
+            maxBlockSize = MAX(maxBlockSize, zfh.blockSizeMax);
+        } else {
+            assert(zfh.frameType == ZSTD_skippableFrame);
+            /* Add the entire skippable frame size to our margin. */
+            margin += compressedSize;
+        }
+
+        assert(srcSize >= compressedSize);
+        src = (const BYTE*)src + compressedSize;
+        srcSize -= compressedSize;
+    }
+
+    /* Add the max block size back to the margin. */
+    margin += maxBlockSize;
+
+    return margin;
+}

 /*-*************************************************************
 *   Frame decoding
@@ -850,7 +895,7 @@ static size_t ZSTD_copyRawBlock(void* dst, size_t dstCapacity,
        if (srcSize == 0) return 0;
        RETURN_ERROR(dstBuffer_null, "");
    }
-    ZSTD_memcpy(dst, src, srcSize);
+    ZSTD_memmove(dst, src, srcSize);
    return srcSize;
 }

@@ -928,6 +973,7 @@ static size_t ZSTD_decompressFrame(ZSTD_DCtx* dctx,

    /* Loop on each block */
    while (1) {
+        BYTE* oBlockEnd = oend;
        size_t decodedSize;
        blockProperties_t blockProperties;
        size_t const cBlockSize = ZSTD_getcBlockSize(ip, remainingSrcSize, &blockProperties);
@@ -937,16 +983,34 @@ static size_t ZSTD_decompressFrame(ZSTD_DCtx* dctx,
        remainingSrcSize -= ZSTD_blockHeaderSize;
        RETURN_ERROR_IF(cBlockSize > remainingSrcSize, srcSize_wrong, "");

+        if (ip >= op && ip < oBlockEnd) {
+            /* We are decompressing in-place. Limit the output pointer so that we
+             * don't overwrite the block that we are currently reading. This will
+             * fail decompression if the input & output pointers aren't spaced
+             * far enough apart.
+             *
+             * This is important to set, even when the pointers are far enough
+             * apart, because ZSTD_decompressBlock_internal() can decide to store
+             * literals in the output buffer, after the block it is decompressing.
+             * Since we don't want anything to overwrite our input, we have to tell
+             * ZSTD_decompressBlock_internal to never write past ip.
+             *
+             * See ZSTD_allocateLiteralsBuffer() for reference.
+             */
+            oBlockEnd = op + (ip - op);
+        }
+
        switch(blockProperties.blockType)
        {
        case bt_compressed:
-            decodedSize = ZSTD_decompressBlock_internal(dctx, op, (size_t)(oend-op), ip, cBlockSize, /* frame */ 1, not_streaming);
+            decodedSize = ZSTD_decompressBlock_internal(dctx, op, (size_t)(oBlockEnd-op), ip, cBlockSize, /* frame */ 1, not_streaming);
            break;
        case bt_raw :
+            /* Use oend instead of oBlockEnd because this function is safe to overlap. It uses memmove. */
            decodedSize = ZSTD_copyRawBlock(op, (size_t)(oend-op), ip, cBlockSize);
            break;
        case bt_rle :
-            decodedSize = ZSTD_setRleBlock(op, (size_t)(oend-op), *ip, blockProperties.origSize);
+            decodedSize = ZSTD_setRleBlock(op, (size_t)(oBlockEnd-op), *ip, blockProperties.origSize);
            break;
        case bt_reserved :
        default:
@@ -981,6 +1045,7 @@ static size_t ZSTD_decompressFrame(ZSTD_DCtx* dctx,
    }
    ZSTD_DCtx_trace_end(dctx, (U64)(op-ostart), (U64)(ip-istart), /* streaming */ 0);
    /* Allow caller to get size read */
+    DEBUGLOG(4, "ZSTD_decompressFrame: decompressed frame of size %zi, consuming %zi bytes of input", op-ostart, ip - (const BYTE*)*srcPtr);
    *srcPtr = ip;
    *srcSizePtr = remainingSrcSize;
    return (size_t)(op-ostart);
@@ -1027,17 +1092,18 @@ static size_t ZSTD_decompressMultiFrame(ZSTD_DCtx* dctx,
        }
 #endif

-        {   U32 const magicNumber = MEM_readLE32(src);
-            DEBUGLOG(4, "reading magic number %08X (expecting %08X)",
-                        (unsigned)magicNumber, ZSTD_MAGICNUMBER);
+        if (srcSize >= 4) {
+            U32 const magicNumber = MEM_readLE32(src);
+            DEBUGLOG(5, "reading magic number %08X", (unsigned)magicNumber);
            if ((magicNumber & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {
+                /* skippable frame detected : skip it */
                size_t const skippableSize = readSkippableFrameSize(src, srcSize);
-                FORWARD_IF_ERROR(skippableSize, "readSkippableFrameSize failed");
+                FORWARD_IF_ERROR(skippableSize, "invalid skippable frame");
                assert(skippableSize <= srcSize);

                src = (const BYTE *)src + skippableSize;
                srcSize -= skippableSize;
-                continue;
+                continue; /* check next frame */
        }   }

        if (ddict) {
@@ -1375,11 +1441,11 @@ ZSTD_loadDEntropy(ZSTD_entropyDTables_t* entropy,
        /* in minimal huffman, we always use X1 variants */
        size_t const hSize = HUF_readDTableX1_wksp(entropy->hufTable,
                                                dictPtr, dictEnd - dictPtr,
-                                                workspace, workspaceSize);
+                                                workspace, workspaceSize, /* flags */ 0);
 #else
        size_t const hSize = HUF_readDTableX2_wksp(entropy->hufTable,
                                                dictPtr, (size_t)(dictEnd - dictPtr),
-                                                workspace, workspaceSize);
+                                                workspace, workspaceSize, /* flags */ 0);
 #endif
        RETURN_ERROR_IF(HUF_isError(hSize), dictionary_corrupted, "");
        dictPtr += hSize;
@@ -1656,7 +1722,9 @@ size_t ZSTD_initDStream_usingDict(ZSTD_DStream* zds, const void* dict, size_t di
 size_t ZSTD_initDStream(ZSTD_DStream* zds)
 {
    DEBUGLOG(4, "ZSTD_initDStream");
-    return ZSTD_initDStream_usingDDict(zds, NULL);
+    FORWARD_IF_ERROR(ZSTD_DCtx_reset(zds, ZSTD_reset_session_only), "");
+    FORWARD_IF_ERROR(ZSTD_DCtx_refDDict(zds, NULL), "");
+    return ZSTD_startingInputLength(zds->format);
 }

 /* ZSTD_initDStream_usingDDict() :
@@ -1664,6 +1732,7 @@ size_t ZSTD_initDStream(ZSTD_DStream* zds)
 * this function cannot fail */
 size_t ZSTD_initDStream_usingDDict(ZSTD_DStream* dctx, const ZSTD_DDict* ddict)
 {
+    DEBUGLOG(4, "ZSTD_initDStream_usingDDict");
    FORWARD_IF_ERROR( ZSTD_DCtx_reset(dctx, ZSTD_reset_session_only) , "");
    FORWARD_IF_ERROR( ZSTD_DCtx_refDDict(dctx, ddict) , "");
    return ZSTD_startingInputLength(dctx->format);
@@ -1674,6 +1743,7 @@ size_t ZSTD_initDStream_usingDDict(ZSTD_DStream* dctx, const ZSTD_DDict* ddict)
 * this function cannot fail */
 size_t ZSTD_resetDStream(ZSTD_DStream* dctx)
 {
+    DEBUGLOG(4, "ZSTD_resetDStream");
    FORWARD_IF_ERROR(ZSTD_DCtx_reset(dctx, ZSTD_reset_session_only), "");
    return ZSTD_startingInputLength(dctx->format);
 }
@@ -1745,6 +1815,11 @@ ZSTD_bounds ZSTD_dParam_getBounds(ZSTD_dParameter dParam)
            bounds.lowerBound = (int)ZSTD_rmd_refSingleDDict;
            bounds.upperBound = (int)ZSTD_rmd_refMultipleDDicts;
            return bounds;
+        case ZSTD_d_disableHuffmanAssembly:
+            bounds.lowerBound = 0;
+            bounds.upperBound = 1;
+            return bounds;
+
        default:;
    }
    bounds.error = ERROR(parameter_unsupported);
@@ -1785,6 +1860,9 @@ size_t ZSTD_DCtx_getParameter(ZSTD_DCtx* dctx, ZSTD_dParameter param, int* value
        case ZSTD_d_refMultipleDDicts:
            *value = (int)dctx->refMultipleDDicts;
            return 0;
+        case ZSTD_d_disableHuffmanAssembly:
+            *value = (int)dctx->disableHufAsm;
+            return 0;
        default:;
    }
    RETURN_ERROR(parameter_unsupported, "");
@@ -1818,6 +1896,10 @@ size_t ZSTD_DCtx_setParameter(ZSTD_DCtx* dctx, ZSTD_dParameter dParam, int value
            }
            dctx->refMultipleDDicts = (ZSTD_refMultipleDDicts_e)value;
            return 0;
+        case ZSTD_d_disableHuffmanAssembly:
+            CHECK_DBOUNDS(ZSTD_d_disableHuffmanAssembly, value);
+            dctx->disableHufAsm = value != 0;
+            return 0;
        default:;
    }
    RETURN_ERROR(parameter_unsupported, "");
@@ -2159,7 +2241,7 @@ size_t ZSTD_decompressStream(ZSTD_DStream* zds, ZSTD_outBuffer* output, ZSTD_inB
                int const isSkipFrame = ZSTD_isSkipFrame(zds);
                size_t loadedSize;
                /* At this point we shouldn't be decompressing a block that we can stream. */
-                assert(neededInSize == ZSTD_nextSrcSizeToDecompressWithInputSize(zds, iend - ip));
+                assert(neededInSize == ZSTD_nextSrcSizeToDecompressWithInputSize(zds, (size_t)(iend - ip)));
                if (isSkipFrame) {
                    loadedSize = MIN(toLoad, (size_t)(iend-ip));
                } else {
@@ -2219,8 +2301,8 @@ size_t ZSTD_decompressStream(ZSTD_DStream* zds, ZSTD_outBuffer* output, ZSTD_inB
    if ((ip==istart) && (op==ostart)) {  /* no forward progress */
        zds->noForwardProgress ++;
        if (zds->noForwardProgress >= ZSTD_NO_FORWARD_PROGRESS_MAX) {
-            RETURN_ERROR_IF(op==oend, dstSize_tooSmall, "");
-            RETURN_ERROR_IF(ip==iend, srcSize_wrong, "");
+            RETURN_ERROR_IF(op==oend, noForwardProgress_destFull, "");
+            RETURN_ERROR_IF(ip==iend, noForwardProgress_inputEmpty, "");
            assert(0);
        }
    } else {
@@ -2257,11 +2339,17 @@ size_t ZSTD_decompressStream_simpleArgs (
                            void* dst, size_t dstCapacity, size_t* dstPos,
                      const void* src, size_t srcSize, size_t* srcPos)
 {
-    ZSTD_outBuffer output = { dst, dstCapacity, *dstPos };
-    ZSTD_inBuffer  input  = { src, srcSize, *srcPos };
-    /* ZSTD_compress_generic() will check validity of dstPos and srcPos */
-    size_t const cErr = ZSTD_decompressStream(dctx, &output, &input);
-    *dstPos = output.pos;
-    *srcPos = input.pos;
-    return cErr;
+    ZSTD_outBuffer output;
+    ZSTD_inBuffer  input;
+    output.dst = dst;
+    output.size = dstCapacity;
+    output.pos = *dstPos;
+    input.src = src;
+    input.size = srcSize;
+    input.pos = *srcPos;
+    {   size_t const cErr = ZSTD_decompressStream(dctx, &output, &input);
+        *dstPos = output.pos;
+        *srcPos = input.pos;
+        return cErr;
+    }
 }
@@ -20,7 +20,6 @@
 #include "../common/mem.h"         /* low level memory routines */
 #define FSE_STATIC_LINKING_ONLY
 #include "../common/fse.h"
-#define HUF_STATIC_LINKING_ONLY
 #include "../common/huf.h"
 #include "../common/zstd_internal.h"
 #include "zstd_decompress_internal.h"   /* ZSTD_DCtx */
@@ -142,6 +141,9 @@ size_t ZSTD_decodeLiteralsBlock(ZSTD_DCtx* dctx,
                U32 const lhc = MEM_readLE32(istart);
                size_t hufSuccess;
                size_t expectedWriteSize = MIN(ZSTD_BLOCKSIZE_MAX, dstCapacity);
+                int const flags = 0
+                    | (ZSTD_DCtx_get_bmi2(dctx) ? HUF_flags_bmi2 : 0)
+                    | (dctx->disableHufAsm ? HUF_flags_disableAsm : 0);
                switch(lhlCode)
                {
                case 0: case 1: default:   /* note : default is impossible, since lhlCode into [0..3] */
@@ -181,14 +183,14 @@ size_t ZSTD_decodeLiteralsBlock(ZSTD_DCtx* dctx,

                if (litEncType==set_repeat) {
                    if (singleStream) {
-                        hufSuccess = HUF_decompress1X_usingDTable_bmi2(
+                        hufSuccess = HUF_decompress1X_usingDTable(
                            dctx->litBuffer, litSize, istart+lhSize, litCSize,
-                            dctx->HUFptr, ZSTD_DCtx_get_bmi2(dctx));
+                            dctx->HUFptr, flags);
                    } else {
                        assert(litSize >= MIN_LITERALS_FOR_4_STREAMS);
-                        hufSuccess = HUF_decompress4X_usingDTable_bmi2(
+                        hufSuccess = HUF_decompress4X_usingDTable(
                            dctx->litBuffer, litSize, istart+lhSize, litCSize,
-                            dctx->HUFptr, ZSTD_DCtx_get_bmi2(dctx));
+                            dctx->HUFptr, flags);
                    }
                } else {
                    if (singleStream) {
@@ -196,18 +198,18 @@ size_t ZSTD_decodeLiteralsBlock(ZSTD_DCtx* dctx,
                        hufSuccess = HUF_decompress1X_DCtx_wksp(
                            dctx->entropy.hufTable, dctx->litBuffer, litSize,
                            istart+lhSize, litCSize, dctx->workspace,
-                            sizeof(dctx->workspace));
+                            sizeof(dctx->workspace), flags);
 #else
-                        hufSuccess = HUF_decompress1X1_DCtx_wksp_bmi2(
+                        hufSuccess = HUF_decompress1X1_DCtx_wksp(
                            dctx->entropy.hufTable, dctx->litBuffer, litSize,
                            istart+lhSize, litCSize, dctx->workspace,
-                            sizeof(dctx->workspace), ZSTD_DCtx_get_bmi2(dctx));
+                            sizeof(dctx->workspace), flags);
 #endif
                    } else {
-                        hufSuccess = HUF_decompress4X_hufOnly_wksp_bmi2(
+                        hufSuccess = HUF_decompress4X_hufOnly_wksp(
                            dctx->entropy.hufTable, dctx->litBuffer, litSize,
                            istart+lhSize, litCSize, dctx->workspace,
-                            sizeof(dctx->workspace), ZSTD_DCtx_get_bmi2(dctx));
+                            sizeof(dctx->workspace), flags);
                    }
                }
                if (dctx->litBufferLocation == ZSTD_split)
@@ -1168,7 +1170,7 @@ ZSTD_updateFseStateWithDInfo(ZSTD_fseState* DStatePtr, BIT_DStream_t* bitD, U16
 }

 /* We need to add at most (ZSTD_WINDOWLOG_MAX_32 - 1) bits to read the maximum
- * offset bits. But we can only read at most (STREAM_ACCUMULATOR_MIN_32 - 1)
+ * offset bits. But we can only read at most STREAM_ACCUMULATOR_MIN_32
 * bits before reloading. This value is the maximum number of bytes we read
 * after reloading when we are decoding long offsets.
 */
@@ -1218,6 +1220,10 @@ ZSTD_decodeSequence(seqState_t* seqState, const ZSTD_longOffset_e longOffsets)
        U32 const llnbBits = llDInfo->nbBits;
        U32 const mlnbBits = mlDInfo->nbBits;
        U32 const ofnbBits = ofDInfo->nbBits;
+
+        assert(llBits <= MaxLLBits);
+        assert(mlBits <= MaxMLBits);
+        assert(ofBits <= MaxOff);
        /*
         * As gcc has better branch and block analyzers, sometimes it is only
         * valuable to mark likeliness for clang, it gives around 3-4% of
@@ -1226,20 +1232,19 @@ ZSTD_decodeSequence(seqState_t* seqState, const ZSTD_longOffset_e longOffsets)

        /* sequence */
        {   size_t offset;
-    #if defined(__clang__)
-            if (LIKELY(ofBits > 1)) {
-    #else
            if (ofBits > 1) {
-    #endif
                ZSTD_STATIC_ASSERT(ZSTD_lo_isLongOffset == 1);
                ZSTD_STATIC_ASSERT(LONG_OFFSETS_MAX_EXTRA_BITS_32 == 5);
-                assert(ofBits <= MaxOff);
+                ZSTD_STATIC_ASSERT(STREAM_ACCUMULATOR_MIN_32 > LONG_OFFSETS_MAX_EXTRA_BITS_32);
+                ZSTD_STATIC_ASSERT(STREAM_ACCUMULATOR_MIN_32 - LONG_OFFSETS_MAX_EXTRA_BITS_32 >= MaxMLBits);
                if (MEM_32bits() && longOffsets && (ofBits >= STREAM_ACCUMULATOR_MIN_32)) {
-                    U32 const extraBits = ofBits - MIN(ofBits, 32 - seqState->DStream.bitsConsumed);
+                    /* Always read extra bits, this keeps the logic simple,
+                     * avoids branches, and avoids accidentally reading 0 bits.
+                     */
+                    U32 const extraBits = LONG_OFFSETS_MAX_EXTRA_BITS_32;
                    offset = ofBase + (BIT_readBitsFast(&seqState->DStream, ofBits - extraBits) << extraBits);
                    BIT_reloadDStream(&seqState->DStream);
-                    if (extraBits) offset += BIT_readBitsFast(&seqState->DStream, extraBits);
-                    assert(extraBits <= LONG_OFFSETS_MAX_EXTRA_BITS_32);   /* to avoid another reload */
+                    offset += BIT_readBitsFast(&seqState->DStream, extraBits);
                } else {
                    offset = ofBase + BIT_readBitsFast(&seqState->DStream, ofBits/*>0*/);   /* <=  (ZSTD_WINDOWLOG_MAX-1) bits */
                    if (MEM_32bits()) BIT_reloadDStream(&seqState->DStream);
@@ -1264,11 +1269,7 @@ ZSTD_decodeSequence(seqState_t* seqState, const ZSTD_longOffset_e longOffsets)
            seq.offset = offset;
        }

-    #if defined(__clang__)
-        if (UNLIKELY(mlBits > 0))
-    #else
        if (mlBits > 0)
-    #endif
            seq.matchLength += BIT_readBitsFast(&seqState->DStream, mlBits/*>0*/);

        if (MEM_32bits() && (mlBits+llBits >= STREAM_ACCUMULATOR_MIN_32-LONG_OFFSETS_MAX_EXTRA_BITS_32))
@@ -1278,11 +1279,7 @@ ZSTD_decodeSequence(seqState_t* seqState, const ZSTD_longOffset_e longOffsets)
        /* Ensure there are enough bits to read the rest of data in 64-bit mode. */
        ZSTD_STATIC_ASSERT(16+LLFSELog+MLFSELog+OffFSELog < STREAM_ACCUMULATOR_MIN_64);

-    #if defined(__clang__)
-        if (UNLIKELY(llBits > 0))
-    #else
        if (llBits > 0)
-    #endif
            seq.litLength += BIT_readBitsFast(&seqState->DStream, llBits/*>0*/);

        if (MEM_32bits())
@@ -1977,34 +1974,79 @@ ZSTD_decompressSequencesLong(ZSTD_DCtx* dctx,
 #endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT */


+/**
+ * @returns The total size of the history referenceable by zstd, including
+ * both the prefix and the extDict. At @p op any offset larger than this
+ * is invalid.
+ */
+static size_t ZSTD_totalHistorySize(BYTE* op, BYTE const* virtualStart)
+{
+    return (size_t)(op - virtualStart);
+}

-#if !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT) && \
-    !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG)
-/* ZSTD_getLongOffsetsShare() :
+typedef struct {
+    unsigned longOffsetShare;
+    unsigned maxNbAdditionalBits;
+} ZSTD_OffsetInfo;
+
+/* ZSTD_getOffsetInfo() :
 * condition : offTable must be valid
 * @return : "share" of long offsets (arbitrarily defined as > (1<<23))
- *           compared to maximum possible of (1<<OffFSELog) */
-static unsigned
-ZSTD_getLongOffsetsShare(const ZSTD_seqSymbol* offTable)
+ *           compared to maximum possible of (1<<OffFSELog),
+ *           as well as the maximum number additional bits required.
+ */
+static ZSTD_OffsetInfo
+ZSTD_getOffsetInfo(const ZSTD_seqSymbol* offTable, int nbSeq)
 {
-    const void* ptr = offTable;
-    U32 const tableLog = ((const ZSTD_seqSymbol_header*)ptr)[0].tableLog;
-    const ZSTD_seqSymbol* table = offTable + 1;
-    U32 const max = 1 << tableLog;
-    U32 u, total = 0;
-    DEBUGLOG(5, "ZSTD_getLongOffsetsShare: (tableLog=%u)", tableLog);
+    ZSTD_OffsetInfo info = {0, 0};
+    /* If nbSeq == 0, then the offTable is uninitialized, but we have
+     * no sequences, so both values should be 0.
+     */
+    if (nbSeq != 0) {
+        const void* ptr = offTable;
+        U32 const tableLog = ((const ZSTD_seqSymbol_header*)ptr)[0].tableLog;
+        const ZSTD_seqSymbol* table = offTable + 1;
+        U32 const max = 1 << tableLog;
+        U32 u;
+        DEBUGLOG(5, "ZSTD_getLongOffsetsShare: (tableLog=%u)", tableLog);

-    assert(max <= (1 << OffFSELog));  /* max not too large */
-    for (u=0; u<max; u++) {
-        if (table[u].nbAdditionalBits > 22) total += 1;
+        assert(max <= (1 << OffFSELog));  /* max not too large */
+        for (u=0; u<max; u++) {
+            info.maxNbAdditionalBits = MAX(info.maxNbAdditionalBits, table[u].nbAdditionalBits);
+            if (table[u].nbAdditionalBits > 22) info.longOffsetShare += 1;
+        }
+
+        assert(tableLog <= OffFSELog);
+        info.longOffsetShare <<= (OffFSELog - tableLog);  /* scale to OffFSELog */
    }

-    assert(tableLog <= OffFSELog);
-    total <<= (OffFSELog - tableLog);  /* scale to OffFSELog */
+    return info;
+}

-    return total;
+/**
+ * @returns The maximum offset we can decode in one read of our bitstream, without
+ * reloading more bits in the middle of the offset bits read. Any offsets larger
+ * than this must use the long offset decoder.
+ */
+static size_t ZSTD_maxShortOffset(void)
+{
+    if (MEM_64bits()) {
+        /* We can decode any offset without reloading bits.
+         * This might change if the max window size grows.
+         */
+        ZSTD_STATIC_ASSERT(ZSTD_WINDOWLOG_MAX <= 31);
+        return (size_t)-1;
+    } else {
+        /* The maximum offBase is (1 << (STREAM_ACCUMULATOR_MIN + 1)) - 1.
+         * This offBase would require STREAM_ACCUMULATOR_MIN extra bits.
+         * Then we have to subtract ZSTD_REP_NUM to get the maximum possible offset.
+         */
+        size_t const maxOffbase = ((size_t)1 << (STREAM_ACCUMULATOR_MIN + 1)) - 1;
+        size_t const maxOffset = maxOffbase - ZSTD_REP_NUM;
+        assert(ZSTD_highbit32((U32)maxOffbase) == STREAM_ACCUMULATOR_MIN);
+        return maxOffset;
+    }
 }
-#endif

 size_t
 ZSTD_decompressBlock_internal(ZSTD_DCtx* dctx,
@@ -2012,13 +2054,6 @@ ZSTD_decompressBlock_internal(ZSTD_DCtx* dctx,
                        const void* src, size_t srcSize, const int frame, const streaming_operation streaming)
 {   /* blockType == blockCompressed */
    const BYTE* ip = (const BYTE*)src;
-    /* isLongOffset must be true if there are long offsets.
-     * Offsets are long if they are larger than 2^STREAM_ACCUMULATOR_MIN.
-     * We don't expect that to be the case in 64-bit mode.
-     * In block mode, window size is not known, so we have to be conservative.
-     * (note: it could possibly be evaluated from current-lowLimit)
-     */
-    ZSTD_longOffset_e const isLongOffset = (ZSTD_longOffset_e)(MEM_32bits() && (!frame || (dctx->fParams.windowSize > (1ULL << STREAM_ACCUMULATOR_MIN))));
    DEBUGLOG(5, "ZSTD_decompressBlock_internal (size : %u)", (U32)srcSize);

    /* Note : the wording of the specification
@@ -2041,6 +2076,23 @@ ZSTD_decompressBlock_internal(ZSTD_DCtx* dctx,

    /* Build Decoding Tables */
    {
+        /* Compute the maximum block size, which must also work when !frame and fParams are unset.
+         * Additionally, take the min with dstCapacity to ensure that the totalHistorySize fits in a size_t.
+         */
+        size_t const blockSizeMax = MIN(dstCapacity, (frame ? dctx->fParams.blockSizeMax : ZSTD_BLOCKSIZE_MAX));
+        size_t const totalHistorySize = ZSTD_totalHistorySize((BYTE*)dst + blockSizeMax, (BYTE const*)dctx->virtualStart);
+        /* isLongOffset must be true if there are long offsets.
+         * Offsets are long if they are larger than ZSTD_maxShortOffset().
+         * We don't expect that to be the case in 64-bit mode.
+         *
+         * We check here to see if our history is large enough to allow long offsets.
+         * If it isn't, then we can't possible have (valid) long offsets. If the offset
+         * is invalid, then it is okay to read it incorrectly.
+         *
+         * If isLongOffsets is true, then we will later check our decoding table to see
+         * if it is even possible to generate long offsets.
+         */
+        ZSTD_longOffset_e isLongOffset = (ZSTD_longOffset_e)(MEM_32bits() && (totalHistorySize > ZSTD_maxShortOffset()));
        /* These macros control at build-time which decompressor implementation
         * we use. If neither is defined, we do some inspection and dispatch at
         * runtime.
@@ -2048,6 +2100,11 @@ ZSTD_decompressBlock_internal(ZSTD_DCtx* dctx,
 #if !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT) && \
    !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG)
        int usePrefetchDecoder = dctx->ddictIsCold;
+#else
+        /* Set to 1 to avoid computing offset info if we don't need to.
+         * Otherwise this value is ignored.
+         */
+        int usePrefetchDecoder = 1;
 #endif
        int nbSeq;
        size_t const seqHSize = ZSTD_decodeSeqHeaders(dctx, &nbSeq, ip, srcSize);
@@ -2055,28 +2112,42 @@ ZSTD_decompressBlock_internal(ZSTD_DCtx* dctx,
        ip += seqHSize;
        srcSize -= seqHSize;

-        RETURN_ERROR_IF(dst == NULL && nbSeq > 0, dstSize_tooSmall, "NULL not handled");
+        RETURN_ERROR_IF((dst == NULL || dstCapacity == 0) && nbSeq > 0, dstSize_tooSmall, "NULL not handled");
+        RETURN_ERROR_IF(MEM_64bits() && sizeof(size_t) == sizeof(void*) && (size_t)(-1) - (size_t)dst < (size_t)(1 << 20), dstSize_tooSmall,
+                "invalid dst");

-#if !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT) && \
-    !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG)
-        if ( !usePrefetchDecoder
-          && (!frame || (dctx->fParams.windowSize > (1<<24)))
-          && (nbSeq>ADVANCED_SEQS) ) {  /* could probably use a larger nbSeq limit */
-            U32 const shareLongOffsets = ZSTD_getLongOffsetsShare(dctx->OFTptr);
-            U32 const minShare = MEM_64bits() ? 7 : 20; /* heuristic values, correspond to 2.73% and 7.81% */
-            usePrefetchDecoder = (shareLongOffsets >= minShare);
+        /* If we could potentially have long offsets, or we might want to use the prefetch decoder,
+         * compute information about the share of long offsets, and the maximum nbAdditionalBits.
+         * NOTE: could probably use a larger nbSeq limit
+         */
+        if (isLongOffset || (!usePrefetchDecoder && (totalHistorySize > (1u << 24)) && (nbSeq > 8))) {
+            ZSTD_OffsetInfo const info = ZSTD_getOffsetInfo(dctx->OFTptr, nbSeq);
+            if (isLongOffset && info.maxNbAdditionalBits <= STREAM_ACCUMULATOR_MIN) {
+                /* If isLongOffset, but the maximum number of additional bits that we see in our table is small
+                 * enough, then we know it is impossible to have too long an offset in this block, so we can
+                 * use the regular offset decoder.
+                 */
+                isLongOffset = ZSTD_lo_isRegularOffset;
+            }
+            if (!usePrefetchDecoder) {
+                U32 const minShare = MEM_64bits() ? 7 : 20; /* heuristic values, correspond to 2.73% and 7.81% */
+                usePrefetchDecoder = (info.longOffsetShare >= minShare);
+            }
        }
-#endif

        dctx->ddictIsCold = 0;

 #if !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT) && \
    !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG)
-        if (usePrefetchDecoder)
+        if (usePrefetchDecoder) {
+#else
+        (void)usePrefetchDecoder;
+        {
 #endif
 #ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT
            return ZSTD_decompressSequencesLong(dctx, dst, dstCapacity, ip, srcSize, nbSeq, isLongOffset, frame);
 #endif
+        }

 #ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG
        /* else */
@@ -2100,9 +2171,9 @@ void ZSTD_checkContinuity(ZSTD_DCtx* dctx, const void* dst, size_t dstSize)
 }


-size_t ZSTD_decompressBlock(ZSTD_DCtx* dctx,
-                            void* dst, size_t dstCapacity,
-                      const void* src, size_t srcSize)
+size_t ZSTD_decompressBlock_deprecated(ZSTD_DCtx* dctx,
+                                       void* dst, size_t dstCapacity,
+                                 const void* src, size_t srcSize)
 {
    size_t dSize;
    ZSTD_checkContinuity(dctx, dst, dstCapacity);
@@ -2110,3 +2181,12 @@ size_t ZSTD_decompressBlock(ZSTD_DCtx* dctx,
    dctx->previousDstEnd = (char*)dst + dSize;
    return dSize;
 }
+
+
+/* NOTE: Must just wrap ZSTD_decompressBlock_deprecated() */
+size_t ZSTD_decompressBlock(ZSTD_DCtx* dctx,
+                            void* dst, size_t dstCapacity,
+                      const void* src, size_t srcSize)
+{
+    return ZSTD_decompressBlock_deprecated(dctx, dst, dstCapacity, src, srcSize);
+}
@@ -64,5 +64,10 @@ void ZSTD_buildFSETable(ZSTD_seqSymbol* dt,
                   unsigned tableLog, void* wksp, size_t wkspSize,
                   int bmi2);

+/* Internal definition of ZSTD_decompressBlock() to avoid deprecation warnings. */
+size_t ZSTD_decompressBlock_deprecated(ZSTD_DCtx* dctx,
+                            void* dst, size_t dstCapacity,
+                      const void* src, size_t srcSize);
+

 #endif /* ZSTD_DEC_BLOCK_H */
@@ -165,6 +165,7 @@ struct ZSTD_DCtx_s
    ZSTD_dictUses_e dictUses;
    ZSTD_DDictHashSet* ddictSet;                    /* Hash set for multiple ddicts */
    ZSTD_refMultipleDDicts_e refMultipleDDicts;     /* User specified: if == 1, will allow references to multiple DDicts. Default == 0 (disabled) */
+    int disableHufAsm;

    /* streaming */
    ZSTD_dStreamStage streamStage;
@@ -13,6 +13,8 @@
 /* *************************************
 *  Dependencies
 ***************************************/
+#define ZSTD_DISABLE_DEPRECATE_WARNINGS  /* suppress warning on ZSTD_initDStream_usingDict */
+#include "../zstd.h"        /* ZSTD_CStream, ZSTD_DStream, ZSTDLIB_API */
 #define ZBUFF_STATIC_LINKING_ONLY
 #include "zbuff.h"

@@ -951,9 +951,17 @@ void COVER_best_finish(COVER_best_t *best, ZDICT_cover_params_t parameters,
  }
 }

+static COVER_dictSelection_t setDictSelection(BYTE* buf, size_t s, size_t csz)
+{
+    COVER_dictSelection_t ds;
+    ds.dictContent = buf;
+    ds.dictSize = s;
+    ds.totalCompressedSize = csz;
+    return ds;
+}
+
 COVER_dictSelection_t COVER_dictSelectionError(size_t error) {
-    COVER_dictSelection_t selection = { NULL, 0, error };
-    return selection;
+    return setDictSelection(NULL, 0, error);
 }

 unsigned COVER_dictSelectionIsError(COVER_dictSelection_t selection) {
@@ -1006,9 +1014,8 @@ COVER_dictSelection_t COVER_selectDict(BYTE* customDictContent, size_t dictBuffe
  }

  if (params.shrinkDict == 0) {
-    COVER_dictSelection_t selection = { largestDictbuffer, dictContentSize, totalCompressedSize };
    free(candidateDictBuffer);
-    return selection;
+    return setDictSelection(largestDictbuffer, dictContentSize, totalCompressedSize);
  }

  largestDict = dictContentSize;
@@ -1041,19 +1048,15 @@ COVER_dictSelection_t COVER_selectDict(BYTE* customDictContent, size_t dictBuffe
    }

    if ((double)totalCompressedSize <= (double)largestCompressed * regressionTolerance) {
-      COVER_dictSelection_t selection = { candidateDictBuffer, dictContentSize, totalCompressedSize };
      free(largestDictbuffer);
-      return selection;
+      return setDictSelection( candidateDictBuffer, dictContentSize, totalCompressedSize );
    }
    dictContentSize *= 2;
  }
  dictContentSize = largestDict;
  totalCompressedSize = largestCompressed;
-  {
-    COVER_dictSelection_t selection = { largestDictbuffer, dictContentSize, totalCompressedSize };
-    free(candidateDictBuffer);
-    return selection;
-  }
+  free(candidateDictBuffer);
+  return setDictSelection( largestDictbuffer, dictContentSize, totalCompressedSize );
 }

 /**
@@ -44,7 +44,6 @@
 #ifndef ZDICT_STATIC_LINKING_ONLY
 #  define ZDICT_STATIC_LINKING_ONLY
 #endif
-#define HUF_STATIC_LINKING_ONLY

 #include "../common/mem.h"           /* read */
 #include "../common/fse.h"           /* FSE_normalizeCount, FSE_writeNCount */
@@ -567,11 +566,11 @@ static void ZDICT_countEStats(EStats_ress_t esr, const ZSTD_parameters* params,
    size_t cSize;

    if (srcSize > blockSizeMax) srcSize = blockSizeMax;   /* protection vs large samples */
-    {   size_t const errorCode = ZSTD_compressBegin_usingCDict(esr.zc, esr.dict);
+    {   size_t const errorCode = ZSTD_compressBegin_usingCDict_deprecated(esr.zc, esr.dict);
        if (ZSTD_isError(errorCode)) { DISPLAYLEVEL(1, "warning : ZSTD_compressBegin_usingCDict failed \n"); return; }

    }
-    cSize = ZSTD_compressBlock(esr.zc, esr.workPlace, ZSTD_BLOCKSIZE_MAX, src, srcSize);
+    cSize = ZSTD_compressBlock_deprecated(esr.zc, esr.workPlace, ZSTD_BLOCKSIZE_MAX, src, srcSize);
    if (ZSTD_isError(cSize)) { DISPLAYLEVEL(3, "warning : could not compress sample size %u \n", (unsigned)srcSize); return; }

    if (cSize) {  /* if == 0; block is not compressible */
@@ -676,6 +675,7 @@ static size_t ZDICT_analyzeEntropy(void*  dstBuffer, size_t maxDstSize,
    size_t const totalSrcSize = ZDICT_totalSampleSize(fileSizes, nbFiles);
    size_t const averageSampleSize = totalSrcSize / (nbFiles + !nbFiles);
    BYTE* dstPtr = (BYTE*)dstBuffer;
+    U32 wksp[HUF_CTABLE_WORKSPACE_SIZE_U32];

    /* init */
    DEBUGLOG(4, "ZDICT_analyzeEntropy");
@@ -716,7 +716,7 @@ static size_t ZDICT_analyzeEntropy(void*  dstBuffer, size_t maxDstSize,
    }   }

    /* analyze, build stats, starting with literals */
-    {   size_t maxNbBits = HUF_buildCTable (hufTable, countLit, 255, huffLog);
+    {   size_t maxNbBits = HUF_buildCTable_wksp(hufTable, countLit, 255, huffLog, wksp, sizeof(wksp));
        if (HUF_isError(maxNbBits)) {
            eSize = maxNbBits;
            DISPLAYLEVEL(1, " HUF_buildCTable error \n");
@@ -725,7 +725,7 @@ static size_t ZDICT_analyzeEntropy(void*  dstBuffer, size_t maxDstSize,
        if (maxNbBits==8) {  /* not compressible : will fail on HUF_writeCTable() */
            DISPLAYLEVEL(2, "warning : pathological dataset : literals are not compressible : samples are noisy or too regular \n");
            ZDICT_flatLit(countLit);  /* replace distribution by a fake "mostly flat but still compressible" distribution, that HUF_writeCTable() can encode */
-            maxNbBits = HUF_buildCTable (hufTable, countLit, 255, huffLog);
+            maxNbBits = HUF_buildCTable_wksp(hufTable, countLit, 255, huffLog, wksp, sizeof(wksp));
            assert(maxNbBits==9);
        }
        huffLog = (U32)maxNbBits;
@@ -766,7 +766,7 @@ static size_t ZDICT_analyzeEntropy(void*  dstBuffer, size_t maxDstSize,
    llLog = (U32)errorCode;

    /* write result to buffer */
-    {   size_t const hhSize = HUF_writeCTable(dstPtr, maxDstSize, hufTable, 255, huffLog);
+    {   size_t const hhSize = HUF_writeCTable_wksp(dstPtr, maxDstSize, hufTable, 255, huffLog, wksp, sizeof(wksp));
        if (HUF_isError(hhSize)) {
            eSize = hhSize;
            DISPLAYLEVEL(1, "HUF_writeCTable error \n");
@@ -242,6 +242,13 @@ MEM_STATIC ZSTD_frameSizeInfo ZSTD_findFrameSizeInfoLegacy(const void *src, size
        frameSizeInfo.compressedSize = ERROR(srcSize_wrong);
        frameSizeInfo.decompressedBound = ZSTD_CONTENTSIZE_ERROR;
    }
+    /* In all cases, decompressedBound == nbBlocks * ZSTD_BLOCKSIZE_MAX.
+     * So we can compute nbBlocks without having to change every function.
+     */
+    if (frameSizeInfo.decompressedBound != ZSTD_CONTENTSIZE_ERROR) {
+        assert((frameSizeInfo.decompressedBound & (ZSTD_BLOCKSIZE_MAX - 1)) == 0);
+        frameSizeInfo.nbBlocks = (size_t)(frameSizeInfo.decompressedBound / ZSTD_BLOCKSIZE_MAX);
+    }
    return frameSizeInfo;
 }

@@ -1720,20 +1720,26 @@ static size_t ZSTD_execSequence(BYTE* op,
    static const int dec32table[] = {0, 1, 2, 1, 4, 4, 4, 4};   /* added */
    static const int dec64table[] = {8, 8, 8, 7, 8, 9,10,11};   /* subtracted */
    const BYTE* const ostart = op;
+    BYTE* const oLitEnd = op + sequence.litLength;
    const size_t litLength = sequence.litLength;
    BYTE* const endMatch = op + litLength + sequence.matchLength;    /* risk : address space overflow (32-bits) */
    const BYTE* const litEnd = *litPtr + litLength;

-    /* check */
+    /* checks */
+    size_t const seqLength = sequence.litLength + sequence.matchLength;
+
+    if (seqLength > (size_t)(oend - op)) return ERROR(dstSize_tooSmall);
+    if (sequence.litLength > (size_t)(litLimit - *litPtr)) return ERROR(corruption_detected);
+    /* Now we know there are no overflow in literal nor match lengths, can use pointer checks */
+    if (sequence.offset > (U32)(oLitEnd - base)) return ERROR(corruption_detected);
+
    if (endMatch > oend) return ERROR(dstSize_tooSmall);   /* overwrite beyond dst buffer */
-    if (litEnd > litLimit) return ERROR(corruption_detected);
-    if (sequence.matchLength > (size_t)(*litPtr-op))  return ERROR(dstSize_tooSmall);    /* overwrite literal segment */
+    if (litEnd > litLimit) return ERROR(corruption_detected);   /* overRead beyond lit buffer */
+    if (sequence.matchLength > (size_t)(*litPtr-op)) return ERROR(dstSize_tooSmall);  /* overwrite literal segment */

    /* copy Literals */
-    if (((size_t)(*litPtr - op) < 8) || ((size_t)(oend-litEnd) < 8) || (op+litLength > oend-8))
-        memmove(op, *litPtr, litLength);   /* overwrite risk */
-    else
-        ZSTD_wildcopy(op, *litPtr, litLength);
+    ZSTD_memmove(op, *litPtr, sequence.litLength);   /* note : v0.1 seems to allow scenarios where output or input are close to end of buffer */
+
    op += litLength;
    *litPtr = litEnd;   /* update for next sequence */

@@ -3066,12 +3066,19 @@ static size_t ZSTD_execSequence(BYTE* op,
    const BYTE* const litEnd = *litPtr + sequence.litLength;

    /* checks */
-    if (oLitEnd > oend_8) return ERROR(dstSize_tooSmall);   /* last match must start at a minimum distance of 8 from oend */
+    size_t const seqLength = sequence.litLength + sequence.matchLength;
+
+    if (seqLength > (size_t)(oend - op)) return ERROR(dstSize_tooSmall);
+    if (sequence.litLength > (size_t)(litLimit - *litPtr)) return ERROR(corruption_detected);
+    /* Now we know there are no overflow in literal nor match lengths, can use the pointer check */
+    if (oLitEnd > oend_8) return ERROR(dstSize_tooSmall);
+    if (sequence.offset > (U32)(oLitEnd - base)) return ERROR(corruption_detected);
+
    if (oMatchEnd > oend) return ERROR(dstSize_tooSmall);   /* overwrite beyond dst buffer */
    if (litEnd > litLimit) return ERROR(corruption_detected);   /* overRead beyond lit buffer */

    /* copy Literals */
-    ZSTD_wildcopy(op, *litPtr, sequence.litLength);   /* note : oLitEnd <= oend-8 : no risk of overwrite beyond oend */
+    ZSTD_wildcopy(op, *litPtr, (ptrdiff_t)sequence.litLength);   /* note : oLitEnd <= oend-8 : no risk of overwrite beyond oend */
    op = oLitEnd;
    *litPtr = litEnd;   /* update for next sequence */

@@ -2706,18 +2706,24 @@ static size_t ZSTD_execSequence(BYTE* op,
    const BYTE* const litEnd = *litPtr + sequence.litLength;

    /* checks */
-    if (oLitEnd > oend_8) return ERROR(dstSize_tooSmall);   /* last match must start at a minimum distance of 8 from oend */
+    size_t const seqLength = sequence.litLength + sequence.matchLength;
+
+    if (seqLength > (size_t)(oend - op)) return ERROR(dstSize_tooSmall);
+    if (sequence.litLength > (size_t)(litLimit - *litPtr)) return ERROR(corruption_detected);
+    /* Now we know there are no overflow in literal nor match lengths, can use pointer checks */
+    if (oLitEnd > oend_8) return ERROR(dstSize_tooSmall);
+    if (sequence.offset > (U32)(oLitEnd - base)) return ERROR(corruption_detected);
+
    if (oMatchEnd > oend) return ERROR(dstSize_tooSmall);   /* overwrite beyond dst buffer */
    if (litEnd > litLimit) return ERROR(corruption_detected);   /* overRead beyond lit buffer */

    /* copy Literals */
-    ZSTD_wildcopy(op, *litPtr, sequence.litLength);   /* note : oLitEnd <= oend-8 : no risk of overwrite beyond oend */
+    ZSTD_wildcopy(op, *litPtr, (ptrdiff_t)sequence.litLength);   /* note : oLitEnd <= oend-8 : no risk of overwrite beyond oend */
    op = oLitEnd;
    *litPtr = litEnd;   /* update for next sequence */

    /* copy Match */
-    {
-        const BYTE* match = op - sequence.offset;
+    {   const BYTE* match = op - sequence.offset;

        /* check */
        if (sequence.offset > (size_t)op) return ERROR(corruption_detected);   /* address space overflow test (this test seems kept by clang optimizer) */
@@ -2828,13 +2828,19 @@ static size_t ZSTD_execSequence(BYTE* op,
    const BYTE* const litEnd = *litPtr + sequence.litLength;
    const BYTE* match = oLitEnd - sequence.offset;

-    /* check */
-    if (oLitEnd > oend_8) return ERROR(dstSize_tooSmall);   /* last match must start at a minimum distance of 8 from oend */
+    /* checks */
+    size_t const seqLength = sequence.litLength + sequence.matchLength;
+
+    if (seqLength > (size_t)(oend - op)) return ERROR(dstSize_tooSmall);
+    if (sequence.litLength > (size_t)(litLimit - *litPtr)) return ERROR(corruption_detected);
+    /* Now we know there are no overflow in literal nor match lengths, can use pointer checks */
+    if (oLitEnd > oend_8) return ERROR(dstSize_tooSmall);
+
    if (oMatchEnd > oend) return ERROR(dstSize_tooSmall);   /* overwrite beyond dst buffer */
-    if (litEnd > litLimit) return ERROR(corruption_detected);   /* risk read beyond lit buffer */
+    if (litEnd > litLimit) return ERROR(corruption_detected);   /* overRead beyond lit buffer */

    /* copy Literals */
-    ZSTD_wildcopy(op, *litPtr, sequence.litLength);   /* note : oLitEnd <= oend-8 : no risk of overwrite beyond oend */
+    ZSTD_wildcopy(op, *litPtr, (ptrdiff_t)sequence.litLength);   /* note : oLitEnd <= oend-8 : no risk of overwrite beyond oend */
    op = oLitEnd;
    *litPtr = litEnd;   /* update for next sequence */

@@ -3182,13 +3182,19 @@ static size_t ZSTDv05_execSequence(BYTE* op,
    const BYTE* const litEnd = *litPtr + sequence.litLength;
    const BYTE* match = oLitEnd - sequence.offset;

-    /* check */
-    if (oLitEnd > oend_8) return ERROR(dstSize_tooSmall);   /* last match must start at a minimum distance of 8 from oend */
+    /* checks */
+    size_t const seqLength = sequence.litLength + sequence.matchLength;
+
+    if (seqLength > (size_t)(oend - op)) return ERROR(dstSize_tooSmall);
+    if (sequence.litLength > (size_t)(litLimit - *litPtr)) return ERROR(corruption_detected);
+    /* Now we know there are no overflow in literal nor match lengths, can use pointer checks */
+    if (oLitEnd > oend_8) return ERROR(dstSize_tooSmall);
+
    if (oMatchEnd > oend) return ERROR(dstSize_tooSmall);   /* overwrite beyond dst buffer */
-    if (litEnd > litLimit) return ERROR(corruption_detected);   /* risk read beyond lit buffer */
+    if (litEnd > litLimit) return ERROR(corruption_detected);   /* overRead beyond lit buffer */

    /* copy Literals */
-    ZSTDv05_wildcopy(op, *litPtr, sequence.litLength);   /* note : oLitEnd <= oend-8 : no risk of overwrite beyond oend */
+    ZSTDv05_wildcopy(op, *litPtr, (ptrdiff_t)sequence.litLength);   /* note : oLitEnd <= oend-8 : no risk of overwrite beyond oend */
    op = oLitEnd;
    *litPtr = litEnd;   /* update for next sequence */

@@ -3322,13 +3322,19 @@ static size_t ZSTDv06_execSequence(BYTE* op,
    const BYTE* const iLitEnd = *litPtr + sequence.litLength;
    const BYTE* match = oLitEnd - sequence.offset;

-    /* check */
-    if (oLitEnd > oend_8) return ERROR(dstSize_tooSmall);   /* last match must start at a minimum distance of 8 from oend */
+    /* checks */
+    size_t const seqLength = sequence.litLength + sequence.matchLength;
+
+    if (seqLength > (size_t)(oend - op)) return ERROR(dstSize_tooSmall);
+    if (sequence.litLength > (size_t)(litLimit - *litPtr)) return ERROR(corruption_detected);
+    /* Now we know there are no overflow in literal nor match lengths, can use pointer checks */
+    if (oLitEnd > oend_8) return ERROR(dstSize_tooSmall);
+
    if (oMatchEnd > oend) return ERROR(dstSize_tooSmall);   /* overwrite beyond dst buffer */
-    if (iLitEnd > litLimit) return ERROR(corruption_detected);   /* over-read beyond lit buffer */
+    if (iLitEnd > litLimit) return ERROR(corruption_detected);   /* overRead beyond lit buffer */

    /* copy Literals */
-    ZSTDv06_wildcopy(op, *litPtr, sequence.litLength);   /* note : oLitEnd <= oend-8 : no risk of overwrite beyond oend */
+    ZSTDv06_wildcopy(op, *litPtr, (ptrdiff_t)sequence.litLength);   /* note : oLitEnd <= oend-8 : no risk of overwrite beyond oend */
    op = oLitEnd;
    *litPtr = iLitEnd;   /* update for next sequence */

@@ -3552,11 +3552,14 @@ size_t ZSTDv07_execSequence(BYTE* op,
    const BYTE* match = oLitEnd - sequence.offset;

    /* check */
-    if ((oLitEnd>oend_w) | (oMatchEnd>oend)) return ERROR(dstSize_tooSmall); /* last match must start at a minimum distance of WILDCOPY_OVERLENGTH from oend */
-    if (iLitEnd > litLimit) return ERROR(corruption_detected);   /* over-read beyond lit buffer */
+    assert(oend >= op);
+    if (sequence.litLength + WILDCOPY_OVERLENGTH > (size_t)(oend - op)) return ERROR(dstSize_tooSmall);
+    if (sequenceLength > (size_t)(oend - op)) return ERROR(dstSize_tooSmall);
+    assert(litLimit >= *litPtr);
+    if (sequence.litLength > (size_t)(litLimit - *litPtr)) return ERROR(corruption_detected);;

    /* copy Literals */
-    ZSTDv07_wildcopy(op, *litPtr, sequence.litLength);   /* note : since oLitEnd <= oend-WILDCOPY_OVERLENGTH, no risk of overwrite beyond oend */
+    ZSTDv07_wildcopy(op, *litPtr, (ptrdiff_t)sequence.litLength);   /* note : since oLitEnd <= oend-WILDCOPY_OVERLENGTH, no risk of overwrite beyond oend */
    op = oLitEnd;
    *litPtr = iLitEnd;   /* update for next sequence */

@@ -3570,7 +3573,7 @@ size_t ZSTDv07_execSequence(BYTE* op,
            return sequenceLength;
        }
        /* span extDict & currentPrefixSegment */
-        {   size_t const length1 = dictEnd - match;
+        {   size_t const length1 = (size_t)(dictEnd - match);
            memmove(oLitEnd, match, length1);
            op = oLitEnd + length1;
            sequence.matchLength -= length1;
@@ -14,7 +14,7 @@ extern "C" {
 #ifndef ZSTD_H_235446
 #define ZSTD_H_235446

-/* ======   Dependency   ======*/
+/* ======   Dependencies   ======*/
 #include <limits.h>   /* INT_MAX */
 #include <stddef.h>   /* size_t */

@@ -106,7 +106,7 @@ extern "C" {
 /*------   Version   ------*/
 #define ZSTD_VERSION_MAJOR    1
 #define ZSTD_VERSION_MINOR    5
-#define ZSTD_VERSION_RELEASE  3
+#define ZSTD_VERSION_RELEASE  5
 #define ZSTD_VERSION_NUMBER  (ZSTD_VERSION_MAJOR *100*100 + ZSTD_VERSION_MINOR *100 + ZSTD_VERSION_RELEASE)

 /*! ZSTD_versionNumber() :
@@ -148,7 +148,8 @@ ZSTDLIB_API const char* ZSTD_versionString(void);
 ***************************************/
 /*! ZSTD_compress() :
 *  Compresses `src` content as a single zstd compressed frame into already allocated `dst`.
- *  Hint : compression runs faster if `dstCapacity` >=  `ZSTD_compressBound(srcSize)`.
+ *  NOTE: Providing `dstCapacity >= ZSTD_compressBound(srcSize)` guarantees that zstd will have
+ *        enough space to successfully compress the data.
 *  @return : compressed size written into `dst` (<= `dstCapacity),
 *            or an error code if it fails (which can be tested using ZSTD_isError()). */
 ZSTDLIB_API size_t ZSTD_compress( void* dst, size_t dstCapacity,
@@ -478,7 +479,8 @@ typedef enum {
     * ZSTD_c_useBlockSplitter
     * ZSTD_c_useRowMatchFinder
     * ZSTD_c_prefetchCDictTables
-     * ZSTD_c_enableMatchFinderFallback
+     * ZSTD_c_enableSeqProducerFallback
+     * ZSTD_c_maxBlockSize
     * Because they are not stable, it's necessary to define ZSTD_STATIC_LINKING_ONLY to access them.
     * note : never ever use experimentalParam? names directly;
     *        also, the enums values themselves are unstable and can still change.
@@ -499,7 +501,9 @@ typedef enum {
     ZSTD_c_experimentalParam14=1011,
     ZSTD_c_experimentalParam15=1012,
     ZSTD_c_experimentalParam16=1013,
-     ZSTD_c_experimentalParam17=1014
+     ZSTD_c_experimentalParam17=1014,
+     ZSTD_c_experimentalParam18=1015,
+     ZSTD_c_experimentalParam19=1016
 } ZSTD_cParameter;

 typedef struct {
@@ -562,7 +566,7 @@ typedef enum {
 *                  They will be used to compress next frame.
 *                  Resetting session never fails.
 *  - The parameters : changes all parameters back to "default".
- *                  This also removes any reference to any dictionary or external matchfinder.
+ *                  This also removes any reference to any dictionary or external sequence producer.
 *                  Parameters can only be changed between 2 sessions (i.e. no compression is currently ongoing)
 *                  otherwise the reset fails, and function returns an error value (which can be tested using ZSTD_isError())
 *  - Both : similar to resetting the session, followed by resetting parameters.
@@ -575,7 +579,8 @@ ZSTDLIB_API size_t ZSTD_CCtx_reset(ZSTD_CCtx* cctx, ZSTD_ResetDirective reset);
 *  Should cctx hold data from a previously unfinished frame, everything about it is forgotten.
 *  - Compression parameters are pushed into CCtx before starting compression, using ZSTD_CCtx_set*()
 *  - The function is always blocking, returns when compression is completed.
- *  Hint : compression runs faster if `dstCapacity` >=  `ZSTD_compressBound(srcSize)`.
+ *  NOTE: Providing `dstCapacity >= ZSTD_compressBound(srcSize)` guarantees that zstd will have
+ *        enough space to successfully compress the data, though it is possible it fails for other reasons.
 * @return : compressed size written into `dst` (<= `dstCapacity),
 *           or an error code if it fails (which can be tested using ZSTD_isError()).
 */
@@ -612,13 +617,15 @@ typedef enum {
     * ZSTD_d_stableOutBuffer
     * ZSTD_d_forceIgnoreChecksum
     * ZSTD_d_refMultipleDDicts
+     * ZSTD_d_disableHuffmanAssembly
     * Because they are not stable, it's necessary to define ZSTD_STATIC_LINKING_ONLY to access them.
     * note : never ever use experimentalParam? names directly
     */
     ZSTD_d_experimentalParam1=1000,
     ZSTD_d_experimentalParam2=1001,
     ZSTD_d_experimentalParam3=1002,
-     ZSTD_d_experimentalParam4=1003
+     ZSTD_d_experimentalParam4=1003,
+     ZSTD_d_experimentalParam5=1004

 } ZSTD_dParameter;

@@ -797,8 +804,6 @@ ZSTDLIB_API size_t ZSTD_CStreamOutSize(void);   /**< recommended size for output
 * This following is a legacy streaming API, available since v1.0+ .
 * It can be replaced by ZSTD_CCtx_reset() and ZSTD_compressStream2().
 * It is redundant, but remains fully supported.
- * Streaming in combination with advanced parameters and dictionary compression
- * can only be used through the new API.
 ******************************************************************************/

 /*!
@@ -807,6 +812,9 @@ ZSTDLIB_API size_t ZSTD_CStreamOutSize(void);   /**< recommended size for output
 *     ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
 *     ZSTD_CCtx_refCDict(zcs, NULL); // clear the dictionary (if any)
 *     ZSTD_CCtx_setParameter(zcs, ZSTD_c_compressionLevel, compressionLevel);
+ *
+ * Note that ZSTD_initCStream() clears any previously set dictionary. Use the new API
+ * to compress with a dictionary.
 */
 ZSTDLIB_API size_t ZSTD_initCStream(ZSTD_CStream* zcs, int compressionLevel);
 /*!
@@ -1012,9 +1020,11 @@ ZSTDLIB_API unsigned ZSTD_getDictID_fromFrame(const void* src, size_t srcSize);
 * Advanced dictionary and prefix API (Requires v1.4.0+)
 *
 * This API allows dictionaries to be used with ZSTD_compress2(),
- * ZSTD_compressStream2(), and ZSTD_decompressDCtx(). Dictionaries are sticky, and
- * only reset with the context is reset with ZSTD_reset_parameters or
- * ZSTD_reset_session_and_parameters. Prefixes are single-use.
+ * ZSTD_compressStream2(), and ZSTD_decompressDCtx().
+ * Dictionaries are sticky, they remain valid when same context is re-used,
+ * they only reset when the context is reset
+ * with ZSTD_reset_parameters or ZSTD_reset_session_and_parameters.
+ * In contrast, Prefixes are single-use.
 ******************************************************************************/


@@ -1035,7 +1045,11 @@ ZSTDLIB_API unsigned ZSTD_getDictID_fromFrame(const void* src, size_t srcSize);
 *           Use experimental ZSTD_CCtx_loadDictionary_byReference() to reference content instead.
 *           In such a case, dictionary buffer must outlive its users.
 *  Note 4 : Use ZSTD_CCtx_loadDictionary_advanced()
- *           to precisely select how dictionary content must be interpreted. */
+ *           to precisely select how dictionary content must be interpreted.
+ *  Note 5 : This method does not benefit from LDM (long distance mode).
+ *           If you want to employ LDM on some large dictionary content,
+ *           prefer employing ZSTD_CCtx_refPrefix() described below.
+ */
 ZSTDLIB_API size_t ZSTD_CCtx_loadDictionary(ZSTD_CCtx* cctx, const void* dict, size_t dictSize);

 /*! ZSTD_CCtx_refCDict() : Requires v1.4.0+
@@ -1058,6 +1072,7 @@ ZSTDLIB_API size_t ZSTD_CCtx_refCDict(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict);
 *  Decompression will need same prefix to properly regenerate data.
 *  Compressing with a prefix is similar in outcome as performing a diff and compressing it,
 *  but performs much faster, especially during decompression (compression speed is tunable with compression level).
+ *  This method is compatible with LDM (long distance mode).
 * @result : 0, or an error code (which can be tested with ZSTD_isError()).
 *  Special: Adding any prefix (including NULL) invalidates any previous prefix or dictionary
 *  Note 1 : Prefix buffer is referenced. It **must** outlive compression.
@@ -1200,6 +1215,7 @@ ZSTDLIB_API size_t ZSTD_sizeof_DDict(const ZSTD_DDict* ddict);
 #define ZSTD_TARGETLENGTH_MIN     0   /* note : comparing this constant to an unsigned results in a tautological test */
 #define ZSTD_STRATEGY_MIN        ZSTD_fast
 #define ZSTD_STRATEGY_MAX        ZSTD_btultra2
+#define ZSTD_BLOCKSIZE_MAX_MIN (1 << 10) /* The minimum valid max blocksize. Maximum blocksizes smaller than this make compressBound() inaccurate. */


 #define ZSTD_OVERLAPLOG_MIN       0
@@ -1380,7 +1396,7 @@ typedef enum {
 } ZSTD_paramSwitch_e;

 /***************************************
-*  Frame size functions
+*  Frame header and size functions
 ***************************************/

 /*! ZSTD_findDecompressedSize() :
@@ -1427,6 +1443,75 @@ ZSTDLIB_STATIC_API unsigned long long ZSTD_decompressBound(const void* src, size
 *           or an error code (if srcSize is too small) */
 ZSTDLIB_STATIC_API size_t ZSTD_frameHeaderSize(const void* src, size_t srcSize);

+typedef enum { ZSTD_frame, ZSTD_skippableFrame } ZSTD_frameType_e;
+typedef struct {
+    unsigned long long frameContentSize; /* if == ZSTD_CONTENTSIZE_UNKNOWN, it means this field is not available. 0 means "empty" */
+    unsigned long long windowSize;       /* can be very large, up to <= frameContentSize */
+    unsigned blockSizeMax;
+    ZSTD_frameType_e frameType;          /* if == ZSTD_skippableFrame, frameContentSize is the size of skippable content */
+    unsigned headerSize;
+    unsigned dictID;
+    unsigned checksumFlag;
+    unsigned _reserved1;
+    unsigned _reserved2;
+} ZSTD_frameHeader;
+
+/*! ZSTD_getFrameHeader() :
+ *  decode Frame Header, or requires larger `srcSize`.
+ * @return : 0, `zfhPtr` is correctly filled,
+ *          >0, `srcSize` is too small, value is wanted `srcSize` amount,
+ *           or an error code, which can be tested using ZSTD_isError() */
+ZSTDLIB_STATIC_API size_t ZSTD_getFrameHeader(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize);   /**< doesn't consume input */
+/*! ZSTD_getFrameHeader_advanced() :
+ *  same as ZSTD_getFrameHeader(),
+ *  with added capability to select a format (like ZSTD_f_zstd1_magicless) */
+ZSTDLIB_STATIC_API size_t ZSTD_getFrameHeader_advanced(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize, ZSTD_format_e format);
+
+/*! ZSTD_decompressionMargin() :
+ * Zstd supports in-place decompression, where the input and output buffers overlap.
+ * In this case, the output buffer must be at least (Margin + Output_Size) bytes large,
+ * and the input buffer must be at the end of the output buffer.
+ *
+ *  _______________________ Output Buffer ________________________
+ * |                                                              |
+ * |                                        ____ Input Buffer ____|
+ * |                                       |                      |
+ * v                                       v                      v
+ * |---------------------------------------|-----------|----------|
+ * ^                                                   ^          ^
+ * |___________________ Output_Size ___________________|_ Margin _|
+ *
+ * NOTE: See also ZSTD_DECOMPRESSION_MARGIN().
+ * NOTE: This applies only to single-pass decompression through ZSTD_decompress() or
+ * ZSTD_decompressDCtx().
+ * NOTE: This function supports multi-frame input.
+ *
+ * @param src The compressed frame(s)
+ * @param srcSize The size of the compressed frame(s)
+ * @returns The decompression margin or an error that can be checked with ZSTD_isError().
+ */
+ZSTDLIB_STATIC_API size_t ZSTD_decompressionMargin(const void* src, size_t srcSize);
+
+/*! ZSTD_DECOMPRESS_MARGIN() :
+ * Similar to ZSTD_decompressionMargin(), but instead of computing the margin from
+ * the compressed frame, compute it from the original size and the blockSizeLog.
+ * See ZSTD_decompressionMargin() for details.
+ *
+ * WARNING: This macro does not support multi-frame input, the input must be a single
+ * zstd frame. If you need that support use the function, or implement it yourself.
+ *
+ * @param originalSize The original uncompressed size of the data.
+ * @param blockSize    The block size == MIN(windowSize, ZSTD_BLOCKSIZE_MAX).
+ *                     Unless you explicitly set the windowLog smaller than
+ *                     ZSTD_BLOCKSIZELOG_MAX you can just use ZSTD_BLOCKSIZE_MAX.
+ */
+#define ZSTD_DECOMPRESSION_MARGIN(originalSize, blockSize) ((size_t)(                                              \
+        ZSTD_FRAMEHEADERSIZE_MAX                                                              /* Frame header */ + \
+        4                                                                                         /* checksum */ + \
+        ((originalSize) == 0 ? 0 : 3 * (((originalSize) + (blockSize) - 1) / blockSize)) /* 3 bytes per block */ + \
+        (blockSize)                                                                    /* One block of margin */   \
+    ))
+
 typedef enum {
  ZSTD_sf_noBlockDelimiters = 0,         /* Representation of ZSTD_Sequence has no block delimiters, sequences only */
  ZSTD_sf_explicitBlockDelimiters = 1    /* Representation of ZSTD_Sequence contains explicit block delimiters */
@@ -1572,8 +1657,11 @@ ZSTDLIB_API unsigned ZSTD_isSkippableFrame(const void* buffer, size_t size);
 *  and ZSTD_estimateCCtxSize_usingCCtxParams(), which can be used in tandem with ZSTD_CCtxParams_setParameter().
 *  Both can be used to estimate memory using custom compression parameters and arbitrary srcSize limits.
 *
- *  Note 2 : only single-threaded compression is supported.
+ *  Note : only single-threaded compression is supported.
 *  ZSTD_estimateCCtxSize_usingCCtxParams() will return an error code if ZSTD_c_nbWorkers is >= 1.
+ *
+ *  Note 2 : ZSTD_estimateCCtxSize* functions are not compatible with the Block-Level Sequence Producer API at this time.
+ *  Size estimates assume that no external sequence producer is registered.
 */
 ZSTDLIB_STATIC_API size_t ZSTD_estimateCCtxSize(int compressionLevel);
 ZSTDLIB_STATIC_API size_t ZSTD_estimateCCtxSize_usingCParams(ZSTD_compressionParameters cParams);
@@ -1592,7 +1680,12 @@ ZSTDLIB_STATIC_API size_t ZSTD_estimateDCtxSize(void);
 *  or deducted from a valid frame Header, using ZSTD_estimateDStreamSize_fromFrame();
 *  Note : if streaming is init with function ZSTD_init?Stream_usingDict(),
 *         an internal ?Dict will be created, which additional size is not estimated here.
- *         In this case, get total size by adding ZSTD_estimate?DictSize */
+ *         In this case, get total size by adding ZSTD_estimate?DictSize
+ *  Note 2 : only single-threaded compression is supported.
+ *  ZSTD_estimateCStreamSize_usingCCtxParams() will return an error code if ZSTD_c_nbWorkers is >= 1.
+ *  Note 3 : ZSTD_estimateCStreamSize* functions are not compatible with the Block-Level Sequence Producer API at this time.
+ *  Size estimates assume that no external sequence producer is registered.
+ */
 ZSTDLIB_STATIC_API size_t ZSTD_estimateCStreamSize(int compressionLevel);
 ZSTDLIB_STATIC_API size_t ZSTD_estimateCStreamSize_usingCParams(ZSTD_compressionParameters cParams);
 ZSTDLIB_STATIC_API size_t ZSTD_estimateCStreamSize_usingCCtxParams(const ZSTD_CCtx_params* params);
@@ -1743,12 +1836,26 @@ ZSTDLIB_STATIC_API size_t ZSTD_checkCParams(ZSTD_compressionParameters params);
 ZSTDLIB_STATIC_API ZSTD_compressionParameters ZSTD_adjustCParams(ZSTD_compressionParameters cPar, unsigned long long srcSize, size_t dictSize);

 /*! ZSTD_CCtx_setCParams() :
- *  Set all parameters provided within @cparams into the working @cctx.
+ *  Set all parameters provided within @p cparams into the working @p cctx.
 *  Note : if modifying parameters during compression (MT mode only),
 *         note that changes to the .windowLog parameter will be ignored.
- * @return 0 on success, or an error code (can be checked with ZSTD_isError()) */
+ * @return 0 on success, or an error code (can be checked with ZSTD_isError()).
+ *         On failure, no parameters are updated.
+ */
 ZSTDLIB_STATIC_API size_t ZSTD_CCtx_setCParams(ZSTD_CCtx* cctx, ZSTD_compressionParameters cparams);

+/*! ZSTD_CCtx_setFParams() :
+ *  Set all parameters provided within @p fparams into the working @p cctx.
+ * @return 0 on success, or an error code (can be checked with ZSTD_isError()).
+ */
+ZSTDLIB_STATIC_API size_t ZSTD_CCtx_setFParams(ZSTD_CCtx* cctx, ZSTD_frameParameters fparams);
+
+/*! ZSTD_CCtx_setParams() :
+ *  Set all parameters provided within @p params into the working @p cctx.
+ * @return 0 on success, or an error code (can be checked with ZSTD_isError()).
+ */
+ZSTDLIB_STATIC_API size_t ZSTD_CCtx_setParams(ZSTD_CCtx* cctx, ZSTD_parameters params);
+
 /*! ZSTD_compress_advanced() :
 *  Note : this function is now DEPRECATED.
 *         It can be replaced by ZSTD_compress2(), in combination with ZSTD_CCtx_setParameter() and other parameter setters.
@@ -2053,19 +2160,55 @@ ZSTDLIB_STATIC_API size_t ZSTD_CCtx_refPrefix_advanced(ZSTD_CCtx* cctx, const vo
 */
 #define ZSTD_c_prefetchCDictTables ZSTD_c_experimentalParam16

-/* ZSTD_c_enableMatchFinderFallback
+/* ZSTD_c_enableSeqProducerFallback
 * Allowed values are 0 (disable) and 1 (enable). The default setting is 0.
 *
- * Controls whether zstd will fall back to an internal matchfinder if an
- * external matchfinder is registered and returns an error code. This fallback is
- * block-by-block: the internal matchfinder will only be called for blocks where
- * the external matchfinder returns an error code. Fallback compression will
+ * Controls whether zstd will fall back to an internal sequence producer if an
+ * external sequence producer is registered and returns an error code. This fallback
+ * is block-by-block: the internal sequence producer will only be called for blocks
+ * where the external sequence producer returns an error code. Fallback parsing will
 * follow any other cParam settings, such as compression level, the same as in a
 * normal (fully-internal) compression operation.
 *
- * The user is strongly encouraged to read the full external matchfinder API
+ * The user is strongly encouraged to read the full Block-Level Sequence Producer API
 * documentation (below) before setting this parameter. */
-#define ZSTD_c_enableMatchFinderFallback ZSTD_c_experimentalParam17
+#define ZSTD_c_enableSeqProducerFallback ZSTD_c_experimentalParam17
+
+/* ZSTD_c_maxBlockSize
+ * Allowed values are between 1KB and ZSTD_BLOCKSIZE_MAX (128KB).
+ * The default is ZSTD_BLOCKSIZE_MAX, and setting to 0 will set to the default.
+ *
+ * This parameter can be used to set an upper bound on the blocksize
+ * that overrides the default ZSTD_BLOCKSIZE_MAX. It cannot be used to set upper
+ * bounds greater than ZSTD_BLOCKSIZE_MAX or bounds lower than 1KB (will make
+ * compressBound() inaccurate). Only currently meant to be used for testing.
+ *
+ */
+#define ZSTD_c_maxBlockSize ZSTD_c_experimentalParam18
+
+/* ZSTD_c_searchForExternalRepcodes
+ * This parameter affects how zstd parses external sequences, such as sequences
+ * provided through the compressSequences() API or from an external block-level
+ * sequence producer.
+ *
+ * If set to ZSTD_ps_enable, the library will check for repeated offsets in
+ * external sequences, even if those repcodes are not explicitly indicated in
+ * the "rep" field. Note that this is the only way to exploit repcode matches
+ * while using compressSequences() or an external sequence producer, since zstd
+ * currently ignores the "rep" field of external sequences.
+ *
+ * If set to ZSTD_ps_disable, the library will not exploit repeated offsets in
+ * external sequences, regardless of whether the "rep" field has been set. This
+ * reduces sequence compression overhead by about 25% while sacrificing some
+ * compression ratio.
+ *
+ * The default value is ZSTD_ps_auto, for which the library will enable/disable
+ * based on compression level.
+ *
+ * Note: for now, this param only has an effect if ZSTD_c_blockDelimiters is
+ * set to ZSTD_sf_explicitBlockDelimiters. That may change in the future.
+ */
+#define ZSTD_c_searchForExternalRepcodes ZSTD_c_experimentalParam19

 /*! ZSTD_CCtx_getParameter() :
 *  Get the requested compression parameter value, selected by enum ZSTD_cParameter,
@@ -2276,6 +2419,17 @@ ZSTDLIB_STATIC_API size_t ZSTD_DCtx_getParameter(ZSTD_DCtx* dctx, ZSTD_dParamete
 */
 #define ZSTD_d_refMultipleDDicts ZSTD_d_experimentalParam4

+/* ZSTD_d_disableHuffmanAssembly
+ * Set to 1 to disable the Huffman assembly implementation.
+ * The default value is 0, which allows zstd to use the Huffman assembly
+ * implementation if available.
+ *
+ * This parameter can be used to disable Huffman assembly at runtime.
+ * If you want to disable it at compile time you can define the macro
+ * ZSTD_DISABLE_ASM.
+ */
+#define ZSTD_d_disableHuffmanAssembly ZSTD_d_experimentalParam5
+

 /*! ZSTD_DCtx_setFormat() :
 *  This function is REDUNDANT. Prefer ZSTD_DCtx_setParameter().
@@ -2345,12 +2499,9 @@ size_t ZSTD_initCStream_usingDict(ZSTD_CStream* zcs,
                           int compressionLevel);

 /*! ZSTD_initCStream_advanced() :
- * This function is DEPRECATED, and is approximately equivalent to:
+ * This function is DEPRECATED, and is equivalent to:
 *     ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
- *     // Pseudocode: Set each zstd parameter and leave the rest as-is.
- *     for ((param, value) : params) {
- *         ZSTD_CCtx_setParameter(zcs, param, value);
- *     }
+ *     ZSTD_CCtx_setParams(zcs, params);
 *     ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize);
 *     ZSTD_CCtx_loadDictionary(zcs, dict, dictSize);
 *
@@ -2379,12 +2530,9 @@ ZSTDLIB_STATIC_API
 size_t ZSTD_initCStream_usingCDict(ZSTD_CStream* zcs, const ZSTD_CDict* cdict);

 /*! ZSTD_initCStream_usingCDict_advanced() :
- *   This function is DEPRECATED, and is approximately equivalent to:
+ *   This function is DEPRECATED, and is equivalent to:
 *     ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
- *     // Pseudocode: Set each zstd frame parameter and leave the rest as-is.
- *     for ((fParam, value) : fParams) {
- *         ZSTD_CCtx_setParameter(zcs, fParam, value);
- *     }
+ *     ZSTD_CCtx_setFParams(zcs, fParams);
 *     ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize);
 *     ZSTD_CCtx_refCDict(zcs, cdict);
 *
@@ -2465,8 +2613,8 @@ ZSTDLIB_STATIC_API size_t ZSTD_toFlushNow(ZSTD_CCtx* cctx);
 *     ZSTD_DCtx_loadDictionary(zds, dict, dictSize);
 *
 * note: no dictionary will be used if dict == NULL or dictSize < 8
- * Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
 */
+ZSTD_DEPRECATED("use ZSTD_DCtx_reset + ZSTD_DCtx_loadDictionary, see zstd.h for detailed instructions")
 ZSTDLIB_STATIC_API size_t ZSTD_initDStream_usingDict(ZSTD_DStream* zds, const void* dict, size_t dictSize);

 /*!
@@ -2476,8 +2624,8 @@ ZSTDLIB_STATIC_API size_t ZSTD_initDStream_usingDict(ZSTD_DStream* zds, const vo
 *     ZSTD_DCtx_refDDict(zds, ddict);
 *
 * note : ddict is referenced, it must outlive decompression session
- * Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
 */
+ZSTD_DEPRECATED("use ZSTD_DCtx_reset + ZSTD_DCtx_refDDict, see zstd.h for detailed instructions")
 ZSTDLIB_STATIC_API size_t ZSTD_initDStream_usingDDict(ZSTD_DStream* zds, const ZSTD_DDict* ddict);

 /*!
@@ -2486,17 +2634,185 @@ ZSTDLIB_STATIC_API size_t ZSTD_initDStream_usingDDict(ZSTD_DStream* zds, const Z
 *     ZSTD_DCtx_reset(zds, ZSTD_reset_session_only);
 *
 * re-use decompression parameters from previous init; saves dictionary loading
- * Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
 */
+ZSTD_DEPRECATED("use ZSTD_DCtx_reset, see zstd.h for detailed instructions")
 ZSTDLIB_STATIC_API size_t ZSTD_resetDStream(ZSTD_DStream* zds);


+/* ********************* BLOCK-LEVEL SEQUENCE PRODUCER API *********************
+ *
+ * *** OVERVIEW ***
+ * The Block-Level Sequence Producer API allows users to provide their own custom
+ * sequence producer which libzstd invokes to process each block. The produced list
+ * of sequences (literals and matches) is then post-processed by libzstd to produce
+ * valid compressed blocks.
+ *
+ * This block-level offload API is a more granular complement of the existing
+ * frame-level offload API compressSequences() (introduced in v1.5.1). It offers
+ * an easier migration story for applications already integrated with libzstd: the
+ * user application continues to invoke the same compression functions
+ * ZSTD_compress2() or ZSTD_compressStream2() as usual, and transparently benefits
+ * from the specific advantages of the external sequence producer. For example,
+ * the sequence producer could be tuned to take advantage of known characteristics
+ * of the input, to offer better speed / ratio, or could leverage hardware
+ * acceleration not available within libzstd itself.
+ *
+ * See contrib/externalSequenceProducer for an example program employing the
+ * Block-Level Sequence Producer API.
+ *
+ * *** USAGE ***
+ * The user is responsible for implementing a function of type
+ * ZSTD_sequenceProducer_F. For each block, zstd will pass the following
+ * arguments to the user-provided function:
+ *
+ *   - sequenceProducerState: a pointer to a user-managed state for the sequence
+ *     producer.
+ *
+ *   - outSeqs, outSeqsCapacity: an output buffer for the sequence producer.
+ *     outSeqsCapacity is guaranteed >= ZSTD_sequenceBound(srcSize). The memory
+ *     backing outSeqs is managed by the CCtx.
+ *
+ *   - src, srcSize: an input buffer for the sequence producer to parse.
+ *     srcSize is guaranteed to be <= ZSTD_BLOCKSIZE_MAX.
+ *
+ *   - dict, dictSize: a history buffer, which may be empty, which the sequence
+ *     producer may reference as it parses the src buffer. Currently, zstd will
+ *     always pass dictSize == 0 into external sequence producers, but this will
+ *     change in the future.
+ *
+ *   - compressionLevel: a signed integer representing the zstd compression level
+ *     set by the user for the current operation. The sequence producer may choose
+ *     to use this information to change its compression strategy and speed/ratio
+ *     tradeoff. Note: the compression level does not reflect zstd parameters set
+ *     through the advanced API.
+ *
+ *   - windowSize: a size_t representing the maximum allowed offset for external
+ *     sequences. Note that sequence offsets are sometimes allowed to exceed the
+ *     windowSize if a dictionary is present, see doc/zstd_compression_format.md
+ *     for details.
+ *
+ * The user-provided function shall return a size_t representing the number of
+ * sequences written to outSeqs. This return value will be treated as an error
+ * code if it is greater than outSeqsCapacity. The return value must be non-zero
+ * if srcSize is non-zero. The ZSTD_SEQUENCE_PRODUCER_ERROR macro is provided
+ * for convenience, but any value greater than outSeqsCapacity will be treated as
+ * an error code.
+ *
+ * If the user-provided function does not return an error code, the sequences
+ * written to outSeqs must be a valid parse of the src buffer. Data corruption may
+ * occur if the parse is not valid. A parse is defined to be valid if the
+ * following conditions hold:
+ *   - The sum of matchLengths and literalLengths must equal srcSize.
+ *   - All sequences in the parse, except for the final sequence, must have
+ *     matchLength >= ZSTD_MINMATCH_MIN. The final sequence must have
+ *     matchLength >= ZSTD_MINMATCH_MIN or matchLength == 0.
+ *   - All offsets must respect the windowSize parameter as specified in
+ *     doc/zstd_compression_format.md.
+ *   - If the final sequence has matchLength == 0, it must also have offset == 0.
+ *
+ * zstd will only validate these conditions (and fail compression if they do not
+ * hold) if the ZSTD_c_validateSequences cParam is enabled. Note that sequence
+ * validation has a performance cost.
+ *
+ * If the user-provided function returns an error, zstd will either fall back
+ * to an internal sequence producer or fail the compression operation. The user can
+ * choose between the two behaviors by setting the ZSTD_c_enableSeqProducerFallback
+ * cParam. Fallback compression will follow any other cParam settings, such as
+ * compression level, the same as in a normal compression operation.
+ *
+ * The user shall instruct zstd to use a particular ZSTD_sequenceProducer_F
+ * function by calling
+ *         ZSTD_registerSequenceProducer(cctx,
+ *                                       sequenceProducerState,
+ *                                       sequenceProducer)
+ * This setting will persist until the next parameter reset of the CCtx.
+ *
+ * The sequenceProducerState must be initialized by the user before calling
+ * ZSTD_registerSequenceProducer(). The user is responsible for destroying the
+ * sequenceProducerState.
+ *
+ * *** LIMITATIONS ***
+ * This API is compatible with all zstd compression APIs which respect advanced parameters.
+ * However, there are three limitations:
+ *
+ * First, the ZSTD_c_enableLongDistanceMatching cParam is not currently supported.
+ * COMPRESSION WILL FAIL if it is enabled and the user tries to compress with a block-level
+ * external sequence producer.
+ *   - Note that ZSTD_c_enableLongDistanceMatching is auto-enabled by default in some
+ *     cases (see its documentation for details). Users must explicitly set
+ *     ZSTD_c_enableLongDistanceMatching to ZSTD_ps_disable in such cases if an external
+ *     sequence producer is registered.
+ *   - As of this writing, ZSTD_c_enableLongDistanceMatching is disabled by default
+ *     whenever ZSTD_c_windowLog < 128MB, but that's subject to change. Users should
+ *     check the docs on ZSTD_c_enableLongDistanceMatching whenever the Block-Level Sequence
+ *     Producer API is used in conjunction with advanced settings (like ZSTD_c_windowLog).
+ *
+ * Second, history buffers are not currently supported. Concretely, zstd will always pass
+ * dictSize == 0 to the external sequence producer (for now). This has two implications:
+ *   - Dictionaries are not currently supported. Compression will *not* fail if the user
+ *     references a dictionary, but the dictionary won't have any effect.
+ *   - Stream history is not currently supported. All advanced compression APIs, including
+ *     streaming APIs, work with external sequence producers, but each block is treated as
+ *     an independent chunk without history from previous blocks.
+ *
+ * Third, multi-threading within a single compression is not currently supported. In other words,
+ * COMPRESSION WILL FAIL if ZSTD_c_nbWorkers > 0 and an external sequence producer is registered.
+ * Multi-threading across compressions is fine: simply create one CCtx per thread.
+ *
+ * Long-term, we plan to overcome all three limitations. There is no technical blocker to
+ * overcoming them. It is purely a question of engineering effort.
+ */
+
+#define ZSTD_SEQUENCE_PRODUCER_ERROR ((size_t)(-1))
+
+typedef size_t ZSTD_sequenceProducer_F (
+  void* sequenceProducerState,
+  ZSTD_Sequence* outSeqs, size_t outSeqsCapacity,
+  const void* src, size_t srcSize,
+  const void* dict, size_t dictSize,
+  int compressionLevel,
+  size_t windowSize
+);
+
+/*! ZSTD_registerSequenceProducer() :
+ * Instruct zstd to use a block-level external sequence producer function.
+ *
+ * The sequenceProducerState must be initialized by the caller, and the caller is
+ * responsible for managing its lifetime. This parameter is sticky across
+ * compressions. It will remain set until the user explicitly resets compression
+ * parameters.
+ *
+ * Sequence producer registration is considered to be an "advanced parameter",
+ * part of the "advanced API". This means it will only have an effect on compression
+ * APIs which respect advanced parameters, such as compress2() and compressStream2().
+ * Older compression APIs such as compressCCtx(), which predate the introduction of
+ * "advanced parameters", will ignore any external sequence producer setting.
+ *
+ * The sequence producer can be "cleared" by registering a NULL function pointer. This
+ * removes all limitations described above in the "LIMITATIONS" section of the API docs.
+ *
+ * The user is strongly encouraged to read the full API documentation (above) before
+ * calling this function. */
+ZSTDLIB_STATIC_API void
+ZSTD_registerSequenceProducer(
+  ZSTD_CCtx* cctx,
+  void* sequenceProducerState,
+  ZSTD_sequenceProducer_F* sequenceProducer
+);
+
+
 /*********************************************************************
-*  Buffer-less and synchronous inner streaming functions
+*  Buffer-less and synchronous inner streaming functions (DEPRECATED)
 *
-*  This is an advanced API, giving full control over buffer management, for users which need direct control over memory.
-*  But it's also a complex one, with several restrictions, documented below.
-*  Prefer normal streaming API for an easier experience.
+*  This API is deprecated, and will be removed in a future version.
+*  It allows streaming (de)compression with user allocated buffers.
+*  However, it is hard to use, and not as well tested as the rest of
+*  our API.
+*
+*  Please use the normal streaming API instead: ZSTD_compressStream2,
+*  and ZSTD_decompressStream.
+*  If there is functionality that you need, but it doesn't provide,
+*  please open an issue on our GitHub.
 ********************************************************************* */

 /**
@@ -2529,15 +2845,20 @@ ZSTDLIB_STATIC_API size_t ZSTD_resetDStream(ZSTD_DStream* zds);
 */

 /*=====   Buffer-less streaming compression functions  =====*/
+ZSTD_DEPRECATED("The buffer-less API is deprecated in favor of the normal streaming API. See docs.")
 ZSTDLIB_STATIC_API size_t ZSTD_compressBegin(ZSTD_CCtx* cctx, int compressionLevel);
+ZSTD_DEPRECATED("The buffer-less API is deprecated in favor of the normal streaming API. See docs.")
 ZSTDLIB_STATIC_API size_t ZSTD_compressBegin_usingDict(ZSTD_CCtx* cctx, const void* dict, size_t dictSize, int compressionLevel);
+ZSTD_DEPRECATED("The buffer-less API is deprecated in favor of the normal streaming API. See docs.")
 ZSTDLIB_STATIC_API size_t ZSTD_compressBegin_usingCDict(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict); /**< note: fails if cdict==NULL */

 ZSTD_DEPRECATED("This function will likely be removed in a future release. It is misleading and has very limited utility.")
 ZSTDLIB_STATIC_API
 size_t ZSTD_copyCCtx(ZSTD_CCtx* cctx, const ZSTD_CCtx* preparedCCtx, unsigned long long pledgedSrcSize); /**<  note: if pledgedSrcSize is not known, use ZSTD_CONTENTSIZE_UNKNOWN */

+ZSTD_DEPRECATED("The buffer-less API is deprecated in favor of the normal streaming API. See docs.")
 ZSTDLIB_STATIC_API size_t ZSTD_compressContinue(ZSTD_CCtx* cctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
+ZSTD_DEPRECATED("The buffer-less API is deprecated in favor of the normal streaming API. See docs.")
 ZSTDLIB_STATIC_API size_t ZSTD_compressEnd(ZSTD_CCtx* cctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);

 /* The ZSTD_compressBegin_advanced() and ZSTD_compressBegin_usingCDict_advanced() are now DEPRECATED and will generate a compiler warning */
@@ -2621,29 +2942,7 @@ size_t ZSTD_compressBegin_usingCDict_advanced(ZSTD_CCtx* const cctx, const ZSTD_
 */

 /*=====   Buffer-less streaming decompression functions  =====*/
-typedef enum { ZSTD_frame, ZSTD_skippableFrame } ZSTD_frameType_e;
-typedef struct {
-    unsigned long long frameContentSize; /* if == ZSTD_CONTENTSIZE_UNKNOWN, it means this field is not available. 0 means "empty" */
-    unsigned long long windowSize;       /* can be very large, up to <= frameContentSize */
-    unsigned blockSizeMax;
-    ZSTD_frameType_e frameType;          /* if == ZSTD_skippableFrame, frameContentSize is the size of skippable content */
-    unsigned headerSize;
-    unsigned dictID;
-    unsigned checksumFlag;
-    unsigned _reserved1;
-    unsigned _reserved2;
-} ZSTD_frameHeader;

-/*! ZSTD_getFrameHeader() :
- *  decode Frame Header, or requires larger `srcSize`.
- * @return : 0, `zfhPtr` is correctly filled,
- *          >0, `srcSize` is too small, value is wanted `srcSize` amount,
- *           or an error code, which can be tested using ZSTD_isError() */
-ZSTDLIB_STATIC_API size_t ZSTD_getFrameHeader(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize);   /**< doesn't consume input */
-/*! ZSTD_getFrameHeader_advanced() :
- *  same as ZSTD_getFrameHeader(),
- *  with added capability to select a format (like ZSTD_f_zstd1_magicless) */
-ZSTDLIB_STATIC_API size_t ZSTD_getFrameHeader_advanced(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize, ZSTD_format_e format);
 ZSTDLIB_STATIC_API size_t ZSTD_decodingBufferSize_min(unsigned long long windowSize, unsigned long long frameContentSize);  /**< when frame content size is not known, pass in frameContentSize == ZSTD_CONTENTSIZE_UNKNOWN */

 ZSTDLIB_STATIC_API size_t ZSTD_decompressBegin(ZSTD_DCtx* dctx);
@@ -2662,11 +2961,23 @@ ZSTDLIB_STATIC_API ZSTD_nextInputType_e ZSTD_nextInputType(ZSTD_DCtx* dctx);



-/* ============================ */
-/**       Block level API       */
-/* ============================ */
+/* ========================================= */
+/**       Block level API (DEPRECATED)       */
+/* ========================================= */

 /*!
+
+    This API is deprecated in favor of the regular compression API.
+    You can get the frame header down to 2 bytes by setting:
+      - ZSTD_c_format = ZSTD_f_zstd1_magicless
+      - ZSTD_c_contentSizeFlag = 0
+      - ZSTD_c_checksumFlag = 0
+      - ZSTD_c_dictIDFlag = 0
+
+    This API is not as well tested as our normal API, so we recommend not using it.
+    We will be removing it in a future version. If the normal API doesn't provide
+    the functionality you need, please open a GitHub issue.
+
    Block functions produce and decode raw zstd blocks, without frame metadata.
    Frame metadata cost is typically ~12 bytes, which can be non-negligible for very small blocks (< 100 bytes).
    But users will have to take in charge needed metadata to regenerate data, such as compressed and content sizes.
@@ -2693,148 +3004,15 @@ ZSTDLIB_STATIC_API ZSTD_nextInputType_e ZSTD_nextInputType(ZSTD_DCtx* dctx);
 */

 /*=====   Raw zstd block functions  =====*/
+ZSTD_DEPRECATED("The block API is deprecated in favor of the normal compression API. See docs.")
 ZSTDLIB_STATIC_API size_t ZSTD_getBlockSize   (const ZSTD_CCtx* cctx);
+ZSTD_DEPRECATED("The block API is deprecated in favor of the normal compression API. See docs.")
 ZSTDLIB_STATIC_API size_t ZSTD_compressBlock  (ZSTD_CCtx* cctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
+ZSTD_DEPRECATED("The block API is deprecated in favor of the normal compression API. See docs.")
 ZSTDLIB_STATIC_API size_t ZSTD_decompressBlock(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
+ZSTD_DEPRECATED("The block API is deprecated in favor of the normal compression API. See docs.")
 ZSTDLIB_STATIC_API size_t ZSTD_insertBlock    (ZSTD_DCtx* dctx, const void* blockStart, size_t blockSize);  /**< insert uncompressed block into `dctx` history. Useful for multi-blocks decompression. */

-
-/* ********************** EXTERNAL MATCHFINDER API **********************
- *
- * *** OVERVIEW ***
- * This API allows users to replace the zstd internal block-level matchfinder
- * with an external matchfinder function. Potential applications of the API
- * include hardware-accelerated matchfinders and matchfinders specialized to
- * particular types of data.
- *
- * See contrib/externalMatchfinder for an example program employing the
- * external matchfinder API.
- *
- * *** USAGE ***
- * The user is responsible for implementing a function of type
- * ZSTD_externalMatchFinder_F. For each block, zstd will pass the following
- * arguments to the user-provided function:
- *
- *   - externalMatchState: a pointer to a user-managed state for the external
- *     matchfinder.
- *
- *   - outSeqs, outSeqsCapacity: an output buffer for sequences produced by the
- *     external matchfinder. outSeqsCapacity is guaranteed >=
- *     ZSTD_sequenceBound(srcSize). The memory backing outSeqs is managed by
- *     the CCtx.
- *
- *   - src, srcSize: an input buffer which the external matchfinder must parse
- *     into sequences. srcSize is guaranteed to be <= ZSTD_BLOCKSIZE_MAX.
- *
- *   - dict, dictSize: a history buffer, which may be empty, which the external
- *     matchfinder may reference as it produces sequences for the src buffer.
- *     Currently, zstd will always pass dictSize == 0 into external matchfinders,
- *     but this will change in the future.
- *
- *   - compressionLevel: a signed integer representing the zstd compression level
- *     set by the user for the current operation. The external matchfinder may
- *     choose to use this information to change its compression strategy and
- *     speed/ratio tradeoff. Note: The compression level does not reflect zstd
- *     parameters set through the advanced API.
- *
- *   - windowSize: a size_t representing the maximum allowed offset for external
- *     sequences. Note that sequence offsets are sometimes allowed to exceed the
- *     windowSize if a dictionary is present, see doc/zstd_compression_format.md
- *     for details.
- *
- * The user-provided function shall return a size_t representing the number of
- * sequences written to outSeqs. This return value will be treated as an error
- * code if it is greater than outSeqsCapacity. The return value must be non-zero
- * if srcSize is non-zero. The ZSTD_EXTERNAL_MATCHFINDER_ERROR macro is provided
- * for convenience, but any value greater than outSeqsCapacity will be treated as
- * an error code.
- *
- * If the user-provided function does not return an error code, the sequences
- * written to outSeqs must be a valid parse of the src buffer. Data corruption may
- * occur if the parse is not valid. A parse is defined to be valid if the
- * following conditions hold:
- *   - The sum of matchLengths and literalLengths is equal to srcSize.
- *   - All sequences in the parse have matchLength != 0, except for the final
- *     sequence. matchLength is not constrained for the final sequence.
- *   - All offsets respect the windowSize parameter as specified in
- *     doc/zstd_compression_format.md.
- *
- * zstd will only validate these conditions (and fail compression if they do not
- * hold) if the ZSTD_c_validateSequences cParam is enabled. Note that sequence
- * validation has a performance cost.
- *
- * If the user-provided function returns an error, zstd will either fall back
- * to an internal matchfinder or fail the compression operation. The user can
- * choose between the two behaviors by setting the
- * ZSTD_c_enableMatchFinderFallback cParam. Fallback compression will follow any
- * other cParam settings, such as compression level, the same as in a normal
- * compression operation.
- *
- * The user shall instruct zstd to use a particular ZSTD_externalMatchFinder_F
- * function by calling ZSTD_registerExternalMatchFinder(cctx, externalMatchState,
- * externalMatchFinder). This setting will persist until the next parameter reset
- * of the CCtx.
- *
- * The externalMatchState must be initialized by the user before calling
- * ZSTD_registerExternalMatchFinder. The user is responsible for destroying the
- * externalMatchState.
- *
- * *** LIMITATIONS ***
- * External matchfinders are compatible with all zstd compression APIs. There are
- * only two limitations.
- *
- * First, the ZSTD_c_enableLongDistanceMatching cParam is not supported.
- * COMPRESSION WILL FAIL if it is enabled and the user tries to compress with an
- * external matchfinder.
- *   - Note that ZSTD_c_enableLongDistanceMatching is auto-enabled by default in
- *     some cases (see its documentation for details). Users must explicitly set
- *     ZSTD_c_enableLongDistanceMatching to ZSTD_ps_disable in such cases if an
- *     external matchfinder is registered.
- *   - As of this writing, ZSTD_c_enableLongDistanceMatching is disabled by default
- *     whenever ZSTD_c_windowLog < 128MB, but that's subject to change. Users should
- *     check the docs on ZSTD_c_enableLongDistanceMatching whenever the external
- *     matchfinder API is used in conjunction with advanced settings (like windowLog).
- *
- * Second, history buffers are not supported. Concretely, zstd will always pass
- * dictSize == 0 to the external matchfinder (for now). This has two implications:
- *   - Dictionaries are not supported. Compression will *not* fail if the user
- *     references a dictionary, but the dictionary won't have any effect.
- *   - Stream history is not supported. All compression APIs, including streaming
- *     APIs, work with the external matchfinder, but the external matchfinder won't
- *     receive any history from the previous block. Each block is an independent chunk.
- *
- * Long-term, we plan to overcome both limitations. There is no technical blocker to
- * overcoming them. It is purely a question of engineering effort.
- */
-
-#define ZSTD_EXTERNAL_MATCHFINDER_ERROR ((size_t)(-1))
-
-typedef size_t ZSTD_externalMatchFinder_F (
-  void* externalMatchState,
-  ZSTD_Sequence* outSeqs, size_t outSeqsCapacity,
-  const void* src, size_t srcSize,
-  const void* dict, size_t dictSize,
-  int compressionLevel,
-  size_t windowSize
-);
-
-/*! ZSTD_registerExternalMatchFinder() :
- * Instruct zstd to use an external matchfinder function.
- *
- * The externalMatchState must be initialized by the caller, and the caller is
- * responsible for managing its lifetime. This parameter is sticky across
- * compressions. It will remain set until the user explicitly resets compression
- * parameters.
- *
- * The user is strongly encouraged to read the full API documentation (above)
- * before calling this function. */
-ZSTDLIB_STATIC_API void
-ZSTD_registerExternalMatchFinder(
-  ZSTD_CCtx* cctx,
-  void* externalMatchState,
-  ZSTD_externalMatchFinder_F* externalMatchFinder
-);
-
 #endif   /* ZSTD_H_ZSTD_STATIC_LINKING_ONLY */

 #if defined (__cplusplus)
@@ -88,12 +88,15 @@ typedef enum {
  ZSTD_error_dstSize_tooSmall = 70,
  ZSTD_error_srcSize_wrong    = 72,
  ZSTD_error_dstBuffer_null   = 74,
+  ZSTD_error_noForwardProgress_destFull = 80,
+  ZSTD_error_noForwardProgress_inputEmpty = 82,
  /* following error codes are __NOT STABLE__, they can be removed or changed in future versions */
  ZSTD_error_frameIndex_tooLarge = 100,
  ZSTD_error_seekableIO          = 102,
  ZSTD_error_dstBuffer_wrong     = 104,
  ZSTD_error_srcBuffer_wrong     = 105,
-  ZSTD_error_externalMatchFinder_failed = 106,
+  ZSTD_error_sequenceProducer_failed = 106,
+  ZSTD_error_externalSequences_invalid = 107,
  ZSTD_error_maxCode = 120  /* never EVER use this value directly, it can change in future versions! Use ZSTD_isError() instead */
 } ZSTD_ErrorCode;

@@ -225,15 +225,18 @@ zstd-noxz : zstd
 .PHONY: zstd-dll
 zstd-dll : LDFLAGS+= -L$(LIBZSTD)
 zstd-dll : LDLIBS += -lzstd
-zstd-dll : ZSTDLIB_LOCAL_SRC = xxhash.c
+zstd-dll : ZSTDLIB_LOCAL_SRC = xxhash.c pool.c threading.c
 zstd-dll : zstd


 ## zstd-pgo: zstd executable optimized with PGO.
 .PHONY: zstd-pgo
+zstd-pgo : LLVM_PROFDATA?=llvm-profdata
+zstd-pgo : PROF_GENERATE_FLAGS=-fprofile-generate $(if $(findstring gcc,$(CC)),-fprofile-dir=.)
+zstd-pgo : PROF_USE_FLAGS=-fprofile-use $(if $(findstring gcc,$(CC)),-fprofile-dir=. -Werror=missing-profile -Wno-error=coverage-mismatch)
 zstd-pgo :
 	$(MAKE) clean HASH_DIR=$(HASH_DIR)
-	$(MAKE) zstd HASH_DIR=$(HASH_DIR) MOREFLAGS=-fprofile-generate
+	$(MAKE) zstd HASH_DIR=$(HASH_DIR) MOREFLAGS="$(PROF_GENERATE_FLAGS)"
 	./zstd -b19i1 $(PROFILE_WITH)
 	./zstd -b16i1 $(PROFILE_WITH)
 	./zstd -b9i2 $(PROFILE_WITH)
@@ -245,8 +248,8 @@ ifndef BUILD_DIR
 else
 	$(RM) zstd $(BUILD_DIR)/zstd $(BUILD_DIR)/*.o
 endif
-	case $(CC) in *clang*) if ! [ -e default.profdata ]; then llvm-profdata merge -output=default.profdata default*.profraw; fi ;; esac
-	$(MAKE) zstd HASH_DIR=$(HASH_DIR) MOREFLAGS=-fprofile-use
+	case $(CC) in *clang*) if ! [ -e default.profdata ]; then $(LLVM_PROFDATA) merge -output=default.profdata default*.profraw; fi ;; esac
+	$(MAKE) zstd HASH_DIR=$(HASH_DIR) MOREFLAGS="$(PROF_USE_FLAGS)"

 ## zstd-small: minimal target, supporting only zstd compression and decompression. no bench. no legacy. no other format.
 CLEAN += zstd-small zstd-frugal
@@ -327,26 +327,31 @@ BMK_benchMemAdvancedNoAlloc(
    /* init */
    memset(&benchResult, 0, sizeof(benchResult));
    if (strlen(displayName)>17) displayName += strlen(displayName) - 17;   /* display last 17 characters */
-    if (adv->mode == BMK_decodeOnly) {  /* benchmark only decompression : source must be already compressed */
+    if (adv->mode == BMK_decodeOnly) {
+        /* benchmark only decompression : source must be already compressed */
        const char* srcPtr = (const char*)srcBuffer;
        U64 totalDSize64 = 0;
        U32 fileNb;
        for (fileNb=0; fileNb<nbFiles; fileNb++) {
            U64 const fSize64 = ZSTD_findDecompressedSize(srcPtr, fileSizes[fileNb]);
-            if (fSize64==0) RETURN_ERROR(32, BMK_benchOutcome_t, "Impossible to determine original size ");
+            if (fSize64 == ZSTD_CONTENTSIZE_UNKNOWN) {
+                RETURN_ERROR(32, BMK_benchOutcome_t, "Decompressed size cannot be determined: cannot benchmark");
+            }
+            if (fSize64 == ZSTD_CONTENTSIZE_ERROR) {
+                RETURN_ERROR(32, BMK_benchOutcome_t, "Error while trying to assess decompressed size: data may be invalid");
+            }
            totalDSize64 += fSize64;
            srcPtr += fileSizes[fileNb];
        }
        {   size_t const decodedSize = (size_t)totalDSize64;
            assert((U64)decodedSize == totalDSize64);   /* check overflow */
            free(*resultBufferPtr);
+            if (totalDSize64 > decodedSize) {  /* size_t overflow */
+                RETURN_ERROR(32, BMK_benchOutcome_t, "decompressed size is too large for local system");
+            }
            *resultBufferPtr = malloc(decodedSize);
            if (!(*resultBufferPtr)) {
-                RETURN_ERROR(33, BMK_benchOutcome_t, "not enough memory");
-            }
-            if (totalDSize64 > decodedSize) {  /* size_t overflow */
-                free(*resultBufferPtr);
-                RETURN_ERROR(32, BMK_benchOutcome_t, "original size is too large");
+                RETURN_ERROR(33, BMK_benchOutcome_t, "allocation error: not enough memory");
            }
            cSize = srcSize;
            srcSize = decodedSize;
@@ -387,12 +392,9 @@ BMK_benchMemAdvancedNoAlloc(
        RDG_genBuffer(compressedBuffer, maxCompressedSize, 0.10, 0.50, 1);
    }

-#if defined(UTIL_TIME_USES_C90_CLOCK)
-    if (adv->nbWorkers > 1) {
-        OUTPUTLEVEL(2, "Warning : time measurements restricted to C90 clock_t. \n")
-        OUTPUTLEVEL(2, "Warning : using C90 clock_t leads to incorrect measurements in multithreading mode. \n")
+    if (!UTIL_support_MT_measurements() && adv->nbWorkers > 1) {
+        OUTPUTLEVEL(2, "Warning : time measurements may be incorrect in multithreading mode... \n")
    }
-#endif

    /* Bench */
    {   U64 const crcOrig = (adv->mode == BMK_decodeOnly) ? 0 : XXH64(srcBuffer, srcSize, 0);
@@ -449,7 +451,7 @@ BMK_benchMemAdvancedNoAlloc(
                BMK_runOutcome_t const cOutcome = BMK_benchTimedFn( timeStateCompress, cbp);

                if (!BMK_isSuccessful_runOutcome(cOutcome)) {
-                    return BMK_benchOutcome_error();
+                    RETURN_ERROR(30, BMK_benchOutcome_t, "compression error");
                }

                {   BMK_runTime_t const cResult = BMK_extract_runTime(cOutcome);
@@ -477,7 +479,7 @@ BMK_benchMemAdvancedNoAlloc(
                BMK_runOutcome_t const dOutcome = BMK_benchTimedFn(timeStateDecompress, dbp);

                if(!BMK_isSuccessful_runOutcome(dOutcome)) {
-                    return BMK_benchOutcome_error();
+                    RETURN_ERROR(30, BMK_benchOutcome_t, "decompression error");
                }

                {   BMK_runTime_t const dResult = BMK_extract_runTime(dOutcome);
@@ -601,7 +603,7 @@ BMK_benchOutcome_t BMK_benchMemAdvanced(const void* srcBuffer, size_t srcSize,

    void* resultBuffer = srcSize ? malloc(srcSize) : NULL;

-    int allocationincomplete = !srcPtrs || !srcSizes || !cPtrs ||
+    int const allocationincomplete = !srcPtrs || !srcSizes || !cPtrs ||
        !cSizes || !cCapacities || !resPtrs || !resSizes ||
        !timeStateCompress || !timeStateDecompress ||
        !cctx || !dctx ||
@@ -695,9 +697,9 @@ static BMK_benchOutcome_t BMK_benchCLevel(const void* srcBuffer, size_t benchedS
                                displayLevel, displayName, adv);
 }

-BMK_benchOutcome_t BMK_syntheticTest(int cLevel, double compressibility,
-                          const ZSTD_compressionParameters* compressionParams,
-                          int displayLevel, const BMK_advancedParams_t* adv)
+int BMK_syntheticTest(int cLevel, double compressibility,
+                      const ZSTD_compressionParameters* compressionParams,
+                      int displayLevel, const BMK_advancedParams_t* adv)
 {
    char name[20] = {0};
    size_t const benchedSize = 10000000;
@@ -705,12 +707,16 @@ BMK_benchOutcome_t BMK_syntheticTest(int cLevel, double compressibility,
    BMK_benchOutcome_t res;

    if (cLevel > ZSTD_maxCLevel()) {
-        RETURN_ERROR(15, BMK_benchOutcome_t, "Invalid Compression Level");
+        DISPLAYLEVEL(1, "Invalid Compression Level");
+        return 15;
    }

    /* Memory allocation */
    srcBuffer = malloc(benchedSize);
-    if (!srcBuffer) RETURN_ERROR(21, BMK_benchOutcome_t, "not enough memory");
+    if (!srcBuffer) {
+        DISPLAYLEVEL(1, "allocation error : not enough memory");
+        return 16;
+    }

    /* Fill input buffer */
    RDG_genBuffer(srcBuffer, benchedSize, compressibility, 0.0, 0);
@@ -726,7 +732,7 @@ BMK_benchOutcome_t BMK_syntheticTest(int cLevel, double compressibility,
    /* clean up */
    free(srcBuffer);

-    return res;
+    return !BMK_isSuccessful_benchOutcome(res);
 }


@@ -788,7 +794,7 @@ static int BMK_loadFiles(void* buffer, size_t bufferSize,
    return 0;
 }

-BMK_benchOutcome_t BMK_benchFilesAdvanced(
+int BMK_benchFilesAdvanced(
                        const char* const * fileNamesTable, unsigned nbFiles,
                        const char* dictFileName, int cLevel,
                        const ZSTD_compressionParameters* compressionParams,
@@ -803,19 +809,25 @@ BMK_benchOutcome_t BMK_benchFilesAdvanced(
    U64 const totalSizeToLoad = UTIL_getTotalFileSize(fileNamesTable, nbFiles);

    if (!nbFiles) {
-        RETURN_ERROR(14, BMK_benchOutcome_t, "No Files to Benchmark");
+        DISPLAYLEVEL(1, "No Files to Benchmark");
+        return 13;
    }

    if (cLevel > ZSTD_maxCLevel()) {
-        RETURN_ERROR(15, BMK_benchOutcome_t, "Invalid Compression Level");
+        DISPLAYLEVEL(1, "Invalid Compression Level");
+        return 14;
    }

    if (totalSizeToLoad == UTIL_FILESIZE_UNKNOWN) {
-        RETURN_ERROR(9, BMK_benchOutcome_t, "Error loading files");
+        DISPLAYLEVEL(1, "Error loading files");
+        return 15;
    }

    fileSizes = (size_t*)calloc(nbFiles, sizeof(size_t));
-    if (!fileSizes) RETURN_ERROR(12, BMK_benchOutcome_t, "not enough memory for fileSizes");
+    if (!fileSizes) {
+        DISPLAYLEVEL(1, "not enough memory for fileSizes");
+        return 16;
+    }

    /* Load dictionary */
    if (dictFileName != NULL) {
@@ -823,18 +835,21 @@ BMK_benchOutcome_t BMK_benchFilesAdvanced(
        if (dictFileSize == UTIL_FILESIZE_UNKNOWN) {
            DISPLAYLEVEL(1, "error loading %s : %s \n", dictFileName, strerror(errno));
            free(fileSizes);
-            RETURN_ERROR(9, BMK_benchOutcome_t, "benchmark aborted");
+            DISPLAYLEVEL(1, "benchmark aborted");
+            return 17;
        }
        if (dictFileSize > 64 MB) {
            free(fileSizes);
-            RETURN_ERROR(10, BMK_benchOutcome_t, "dictionary file %s too large", dictFileName);
+            DISPLAYLEVEL(1, "dictionary file %s too large", dictFileName);
+            return 18;
        }
        dictBufferSize = (size_t)dictFileSize;
        dictBuffer = malloc(dictBufferSize);
        if (dictBuffer==NULL) {
            free(fileSizes);
-            RETURN_ERROR(11, BMK_benchOutcome_t, "not enough memory for dictionary (%u bytes)",
+            DISPLAYLEVEL(1, "not enough memory for dictionary (%u bytes)",
                            (unsigned)dictBufferSize);
+            return 19;
        }

        {   int const errorCode = BMK_loadFiles(dictBuffer, dictBufferSize,
@@ -856,7 +871,8 @@ BMK_benchOutcome_t BMK_benchFilesAdvanced(
    if (!srcBuffer) {
        free(dictBuffer);
        free(fileSizes);
-        RETURN_ERROR(12, BMK_benchOutcome_t, "not enough memory");
+        DISPLAYLEVEL(1, "not enough memory for srcBuffer");
+        return 20;
    }

    /* Load input buffer */
@@ -884,12 +900,11 @@ _cleanUp:
    free(srcBuffer);
    free(dictBuffer);
    free(fileSizes);
-    return res;
+    return !BMK_isSuccessful_benchOutcome(res);
 }


-BMK_benchOutcome_t BMK_benchFiles(
-                    const char* const * fileNamesTable, unsigned nbFiles,
+int BMK_benchFiles(const char* const * fileNamesTable, unsigned nbFiles,
                    const char* dictFileName,
                    int cLevel, const ZSTD_compressionParameters* compressionParams,
                    int displayLevel)
@@ -81,21 +81,13 @@ BMK_benchResult_t BMK_extract_benchResult(BMK_benchOutcome_t outcome);
 *      2 : + result + interaction + warnings;
 *      3 : + information;
 *      4 : + debug
- * @return:
- *      a variant, which expresses either an error, or a valid result.
- *      Use BMK_isSuccessful_benchOutcome() to check if function was successful.
- *      If yes, extract the valid result with BMK_extract_benchResult(),
- *      it will contain :
- *          .cSpeed: compression speed in bytes per second,
- *          .dSpeed: decompression speed in bytes per second,
- *          .cSize : compressed size, in bytes
- *          .cMem  : memory budget required for the compression context
+ * @return: 0 on success, !0 on error
 */
-BMK_benchOutcome_t BMK_benchFiles(
-                   const char* const * fileNamesTable, unsigned nbFiles,
-                   const char* dictFileName,
-                   int cLevel, const ZSTD_compressionParameters* compressionParams,
-                   int displayLevel);
+int BMK_benchFiles(
+            const char* const * fileNamesTable, unsigned nbFiles,
+            const char* dictFileName,
+            int cLevel, const ZSTD_compressionParameters* compressionParams,
+            int displayLevel);


 typedef enum {
@@ -126,11 +118,11 @@ BMK_advancedParams_t BMK_initAdvancedParams(void);
 /*! BMK_benchFilesAdvanced():
 *  Same as BMK_benchFiles(),
 *  with more controls, provided through advancedParams_t structure */
-BMK_benchOutcome_t BMK_benchFilesAdvanced(
-                   const char* const * fileNamesTable, unsigned nbFiles,
-                   const char* dictFileName,
-                   int cLevel, const ZSTD_compressionParameters* compressionParams,
-                   int displayLevel, const BMK_advancedParams_t* adv);
+int BMK_benchFilesAdvanced(
+               const char* const * fileNamesTable, unsigned nbFiles,
+               const char* dictFileName,
+               int cLevel, const ZSTD_compressionParameters* compressionParams,
+               int displayLevel, const BMK_advancedParams_t* adv);

 /*! BMK_syntheticTest() -- called from zstdcli */
 /*  Generates a sample with datagen, using compressibility argument */
@@ -139,20 +131,11 @@ BMK_benchOutcome_t BMK_benchFilesAdvanced(
 *  compressionParams - basic compression Parameters
 *  displayLevel - see benchFiles
 *  adv - see advanced_Params_t
- * @return:
- *      a variant, which expresses either an error, or a valid result.
- *      Use BMK_isSuccessful_benchOutcome() to check if function was successful.
- *      If yes, extract the valid result with BMK_extract_benchResult(),
- *      it will contain :
- *          .cSpeed: compression speed in bytes per second,
- *          .dSpeed: decompression speed in bytes per second,
- *          .cSize : compressed size, in bytes
- *          .cMem  : memory budget required for the compression context
+ * @return: 0 on success, !0 on error
 */
-BMK_benchOutcome_t BMK_syntheticTest(
-                          int cLevel, double compressibility,
-                          const ZSTD_compressionParameters* compressionParams,
-                          int displayLevel, const BMK_advancedParams_t* adv);
+int BMK_syntheticTest(int cLevel, double compressibility,
+                      const ZSTD_compressionParameters* compressionParams,
+                      int displayLevel, const BMK_advancedParams_t* adv);



@@ -190,8 +173,8 @@ BMK_benchOutcome_t BMK_benchMem(const void* srcBuffer, size_t srcSize,
                        int displayLevel, const char* displayName);


-/* BMK_benchMemAdvanced() : same as BMK_benchMem()
- * with following additional options :
+/* BMK_benchMemAdvanced() : used by Paramgrill
+ * same as BMK_benchMem() with following additional options :
 * dstBuffer - destination buffer to write compressed output in, NULL if none provided.
 * dstCapacity - capacity of destination buffer, give 0 if dstBuffer = NULL
 * adv = see advancedParams_t
@@ -274,21 +274,20 @@ static fileStats DiB_fileStats(const char** fileNamesTable, int nbFiles, size_t
    int n;
    memset(&fs, 0, sizeof(fs));

-    // We assume that if chunking is requested, the chunk size is < SAMPLESIZE_MAX
+    /* We assume that if chunking is requested, the chunk size is < SAMPLESIZE_MAX */
    assert( chunkSize <= SAMPLESIZE_MAX );

    for (n=0; n<nbFiles; n++) {
      S64 const fileSize = DiB_getFileSize(fileNamesTable[n]);
-      // TODO: is there a minimum sample size? What if the file is 1-byte?
+      /* TODO: is there a minimum sample size? What if the file is 1-byte? */
      if (fileSize == 0) {
        DISPLAYLEVEL(3, "Sample file '%s' has zero size, skipping...\n", fileNamesTable[n]);
        continue;
      }

      /* the case where we are breaking up files in sample chunks */
-      if (chunkSize > 0)
-      {
-        // TODO: is there a minimum sample size? Can we have a 1-byte sample?
+      if (chunkSize > 0) {
+        /* TODO: is there a minimum sample size? Can we have a 1-byte sample? */
        fs.nbSamples += (int)((fileSize + chunkSize-1) / chunkSize);
        fs.totalSizeToLoad += fileSize;
      }
@@ -28,6 +28,7 @@
 #include <stdio.h>      /* fprintf, open, fdopen, fread, _fileno, stdin, stdout */
 #include <stdlib.h>     /* malloc, free */
 #include <string.h>     /* strcmp, strlen */
+#include <time.h>       /* clock_t, to measure process time */
 #include <fcntl.h>      /* O_WRONLY */
 #include <assert.h>
 #include <errno.h>      /* errno */
@@ -113,11 +114,15 @@ char const* FIO_lzmaVersion(void)
 #define FNSPACE 30

 /* Default file permissions 0666 (modulated by umask) */
+/* Temporary restricted file permissions are used when we're going to
+ * chmod/chown at the end of the operation. */
 #if !defined(_WIN32)
 /* These macros aren't defined on windows. */
 #define DEFAULT_FILE_PERMISSIONS (S_IRUSR|S_IWUSR|S_IRGRP|S_IWGRP|S_IROTH|S_IWOTH)
+#define TEMPORARY_FILE_PERMISSIONS (S_IRUSR|S_IWUSR)
 #else
 #define DEFAULT_FILE_PERMISSIONS (0666)
+#define TEMPORARY_FILE_PERMISSIONS (0600)
 #endif

 /*-************************************
@@ -357,7 +362,7 @@ void FIO_setDictIDFlag(FIO_prefs_t* const prefs, int dictIDFlag) { prefs->dictID

 void FIO_setChecksumFlag(FIO_prefs_t* const prefs, int checksumFlag) { prefs->checksumFlag = checksumFlag; }

-void FIO_setRemoveSrcFile(FIO_prefs_t* const prefs, unsigned flag) { prefs->removeSrcFile = (flag>0); }
+void FIO_setRemoveSrcFile(FIO_prefs_t* const prefs, int flag) { prefs->removeSrcFile = (flag!=0); }

 void FIO_setMemLimit(FIO_prefs_t* const prefs, unsigned memLimit) { prefs->memLimit = memLimit; }

@@ -480,6 +485,11 @@ void FIO_setPassThroughFlag(FIO_prefs_t* const prefs, int value) {
    prefs->passThrough = (value != 0);
 }

+void FIO_setMMapDict(FIO_prefs_t* const prefs, ZSTD_paramSwitch_e value)
+{
+    prefs->mmapDict = value;
+}
+
 /* FIO_ctx_t functions */

 void FIO_setHasStdoutOutput(FIO_ctx_t* const fCtx, int value) {
@@ -530,26 +540,26 @@ static int FIO_removeFile(const char* path)
 /** FIO_openSrcFile() :
 *  condition : `srcFileName` must be non-NULL. `prefs` may be NULL.
 * @result : FILE* to `srcFileName`, or NULL if it fails */
-static FILE* FIO_openSrcFile(const FIO_prefs_t* const prefs, const char* srcFileName)
+static FILE* FIO_openSrcFile(const FIO_prefs_t* const prefs, const char* srcFileName, stat_t* statbuf)
 {
-    stat_t statbuf;
    int allowBlockDevices = prefs != NULL ? prefs->allowBlockDevices : 0;
    assert(srcFileName != NULL);
+    assert(statbuf != NULL);
    if (!strcmp (srcFileName, stdinmark)) {
        DISPLAYLEVEL(4,"Using stdin for input \n");
        SET_BINARY_MODE(stdin);
        return stdin;
    }

-    if (!UTIL_stat(srcFileName, &statbuf)) {
+    if (!UTIL_stat(srcFileName, statbuf)) {
        DISPLAYLEVEL(1, "zstd: can't stat %s : %s -- ignored \n",
                        srcFileName, strerror(errno));
        return NULL;
    }

-    if (!UTIL_isRegularFileStat(&statbuf)
-     && !UTIL_isFIFOStat(&statbuf)
-     && !(allowBlockDevices && UTIL_isBlockDevStat(&statbuf))
+    if (!UTIL_isRegularFileStat(statbuf)
+     && !UTIL_isFIFOStat(statbuf)
+     && !(allowBlockDevices && UTIL_isBlockDevStat(statbuf))
    ) {
        DISPLAYLEVEL(1, "zstd: %s is not a regular file -- ignored \n",
                        srcFileName);
@@ -571,6 +581,8 @@ FIO_openDstFile(FIO_ctx_t* fCtx, FIO_prefs_t* const prefs,
                const char* srcFileName, const char* dstFileName,
                const int mode)
 {
+    int isDstRegFile;
+
    if (prefs->testMode) return NULL;  /* do not open file in test mode */

    assert(dstFileName != NULL);
@@ -590,11 +602,16 @@ FIO_openDstFile(FIO_ctx_t* fCtx, FIO_prefs_t* const prefs,
        return NULL;
    }

+    isDstRegFile = UTIL_isRegularFile(dstFileName);  /* invoke once */
    if (prefs->sparseFileSupport == 1) {
        prefs->sparseFileSupport = ZSTD_SPARSE_DEFAULT;
+        if (!isDstRegFile) {
+            prefs->sparseFileSupport = 0;
+            DISPLAYLEVEL(4, "Sparse File Support is disabled when output is not a file \n");
+        }
    }

-    if (UTIL_isRegularFile(dstFileName)) {
+    if (isDstRegFile) {
        /* Check if destination file already exists */
 #if !defined(_WIN32)
        /* this test does not work on Windows :
@@ -607,7 +624,7 @@ FIO_openDstFile(FIO_ctx_t* fCtx, FIO_prefs_t* const prefs,
        if (!prefs->overwrite) {
            if (g_display_prefs.displayLevel <= 1) {
                /* No interaction possible */
-                DISPLAY("zstd: %s already exists; not overwritten  \n",
+                DISPLAYLEVEL(1, "zstd: %s already exists; not overwritten  \n",
                        dstFileName);
                return NULL;
            }
@@ -639,55 +656,70 @@ FIO_openDstFile(FIO_ctx_t* fCtx, FIO_prefs_t* const prefs,
 #endif
        if (f == NULL) {
            DISPLAYLEVEL(1, "zstd: %s: %s\n", dstFileName, strerror(errno));
+        } else {
+            /* An increased buffer size can provide a significant performance
+             * boost on some platforms. Note that providing a NULL buf with a
+             * size that's not 0 is not defined in ANSI C, but is defined in an
+             * extension. There are three possibilities here:
+             * 1. Libc supports the extended version and everything is good.
+             * 2. Libc ignores the size when buf is NULL, in which case
+             *    everything will continue as if we didn't call `setvbuf()`.
+             * 3. We fail the call and execution continues but a warning
+             *    message might be shown.
+             * In all cases due execution continues. For now, I believe that
+             * this is a more cost-effective solution than managing the buffers
+             * allocations ourselves (will require an API change).
+             */
+            if (setvbuf(f, NULL, _IOFBF, 1 MB)) {
+                DISPLAYLEVEL(2, "Warning: setvbuf failed for %s\n", dstFileName);
+            }
        }
-        /* An increased buffer size can provide a significant performance boost on some platforms.
-         * Note that providing a NULL buf with a size that's not 0 is not defined in ANSI C, but is defined
-         * in an extension. There are three possibilities here -
-         * 1. Libc supports the extended version and everything is good.
-         * 2. Libc ignores the size when buf is NULL, in which case everything will continue as if we didn't
-         *    call `setvbuf`.
-         * 3. We fail the call and execution continues but a warning message might be shown.
-         * In all cases due execution continues. For now, I believe that this is a more cost-effective
-         * solution than managing the buffers allocations ourselves (will require an API change). */
-        if(setvbuf(f, NULL, _IOFBF, 1 MB))
-            DISPLAYLEVEL(2, "Warning: setvbuf failed for %s\n", dstFileName);
        return f;
    }
 }

-/*! FIO_createDictBuffer() :
- *  creates a buffer, pointed by `*bufferPtr`,
+
+/* FIO_getDictFileStat() :
+ */
+static void FIO_getDictFileStat(const char* fileName, stat_t* dictFileStat) {
+    assert(dictFileStat != NULL);
+    if (fileName == NULL) return;
+
+    if (!UTIL_stat(fileName, dictFileStat)) {
+        EXM_THROW(31, "Stat failed on dictionary file %s: %s", fileName, strerror(errno));
+    }
+
+    if (!UTIL_isRegularFileStat(dictFileStat)) {
+        EXM_THROW(32, "Dictionary %s must be a regular file.", fileName);
+    }
+}
+
+/*  FIO_setDictBufferMalloc() :
+ *  allocates a buffer, pointed by `dict->dictBuffer`,
 *  loads `filename` content into it, up to DICTSIZE_MAX bytes.
 * @return : loaded size
 *  if fileName==NULL, returns 0 and a NULL pointer
 */
-static size_t FIO_createDictBuffer(void** bufferPtr, const char* fileName, FIO_prefs_t* const prefs)
+static size_t FIO_setDictBufferMalloc(FIO_Dict_t* dict, const char* fileName, FIO_prefs_t* const prefs, stat_t* dictFileStat)
 {
    FILE* fileHandle;
    U64 fileSize;
-    stat_t statbuf;
+    void** bufferPtr = &dict->dictBuffer;

    assert(bufferPtr != NULL);
+    assert(dictFileStat != NULL);
    *bufferPtr = NULL;
    if (fileName == NULL) return 0;

    DISPLAYLEVEL(4,"Loading %s as dictionary \n", fileName);

-    if (!UTIL_stat(fileName, &statbuf)) {
-        EXM_THROW(31, "Stat failed on dictionary file %s: %s", fileName, strerror(errno));
-    }
-
-    if (!UTIL_isRegularFileStat(&statbuf)) {
-        EXM_THROW(32, "Dictionary %s must be a regular file.", fileName);
-    }
-
    fileHandle = fopen(fileName, "rb");

    if (fileHandle == NULL) {
        EXM_THROW(33, "Couldn't open dictionary %s: %s", fileName, strerror(errno));
    }

-    fileSize = UTIL_getFileSizeStat(&statbuf);
+    fileSize = UTIL_getFileSizeStat(dictFileStat);
    {
        size_t const dictSizeMax = prefs->patchFromMode ? prefs->memLimit : DICTSIZE_MAX;
        if (fileSize >  dictSizeMax) {
@@ -707,6 +739,130 @@ static size_t FIO_createDictBuffer(void** bufferPtr, const char* fileName, FIO_p
    return (size_t)fileSize;
 }

+#if (PLATFORM_POSIX_VERSION > 0)
+#include <sys/mman.h>
+static void FIO_munmap(FIO_Dict_t* dict)
+{
+    munmap(dict->dictBuffer, dict->dictBufferSize);
+    dict->dictBuffer = NULL;
+    dict->dictBufferSize = 0;
+}
+static size_t FIO_setDictBufferMMap(FIO_Dict_t* dict, const char* fileName, FIO_prefs_t* const prefs, stat_t* dictFileStat)
+{
+    int fileHandle;
+    U64 fileSize;
+    void** bufferPtr = &dict->dictBuffer;
+
+    assert(bufferPtr != NULL);
+    assert(dictFileStat != NULL);
+    *bufferPtr = NULL;
+    if (fileName == NULL) return 0;
+
+    DISPLAYLEVEL(4,"Loading %s as dictionary \n", fileName);
+
+    fileHandle = open(fileName, O_RDONLY);
+
+    if (fileHandle == -1) {
+        EXM_THROW(33, "Couldn't open dictionary %s: %s", fileName, strerror(errno));
+    }
+
+    fileSize = UTIL_getFileSizeStat(dictFileStat);
+    {
+        size_t const dictSizeMax = prefs->patchFromMode ? prefs->memLimit : DICTSIZE_MAX;
+        if (fileSize >  dictSizeMax) {
+            EXM_THROW(34, "Dictionary file %s is too large (> %u bytes)",
+                            fileName,  (unsigned)dictSizeMax);   /* avoid extreme cases */
+        }
+    }
+
+    *bufferPtr = mmap(NULL, (size_t)fileSize, PROT_READ, MAP_PRIVATE, fileHandle, 0);
+    if (*bufferPtr==NULL) EXM_THROW(34, "%s", strerror(errno));
+
+    close(fileHandle);
+    return (size_t)fileSize;
+}
+#elif defined(_MSC_VER) || defined(_WIN32)
+#include <windows.h>
+static void FIO_munmap(FIO_Dict_t* dict)
+{
+    UnmapViewOfFile(dict->dictBuffer);
+    CloseHandle(dict->dictHandle);
+    dict->dictBuffer = NULL;
+    dict->dictBufferSize = 0;
+}
+static size_t FIO_setDictBufferMMap(FIO_Dict_t* dict, const char* fileName, FIO_prefs_t* const prefs, stat_t* dictFileStat)
+{
+    HANDLE fileHandle, mapping;
+    U64 fileSize;
+    void** bufferPtr = &dict->dictBuffer;
+
+    assert(bufferPtr != NULL);
+    assert(dictFileStat != NULL);
+    *bufferPtr = NULL;
+    if (fileName == NULL) return 0;
+
+    DISPLAYLEVEL(4,"Loading %s as dictionary \n", fileName);
+
+    fileHandle = CreateFileA(fileName, GENERIC_READ, FILE_SHARE_READ, NULL, OPEN_EXISTING, FILE_ATTRIBUTE_READONLY, NULL);
+
+    if (fileHandle == INVALID_HANDLE_VALUE) {
+        EXM_THROW(33, "Couldn't open dictionary %s: %s", fileName, strerror(errno));
+    }
+
+    fileSize = UTIL_getFileSizeStat(dictFileStat);
+    {
+        size_t const dictSizeMax = prefs->patchFromMode ? prefs->memLimit : DICTSIZE_MAX;
+        if (fileSize >  dictSizeMax) {
+            EXM_THROW(34, "Dictionary file %s is too large (> %u bytes)",
+                            fileName,  (unsigned)dictSizeMax);   /* avoid extreme cases */
+        }
+    }
+
+    mapping = CreateFileMapping(fileHandle, NULL, PAGE_READONLY, 0, 0, NULL);
+	if (mapping == NULL) {
+        EXM_THROW(35, "Couldn't map dictionary %s: %s", fileName, strerror(errno));
+    }
+
+	*bufferPtr = MapViewOfFile(mapping, FILE_MAP_READ, 0, 0, (DWORD)fileSize); /* we can only cast to DWORD here because dictSize <= 2GB */
+	if (*bufferPtr==NULL) EXM_THROW(36, "%s", strerror(errno));
+
+    dict->dictHandle = fileHandle;
+    return (size_t)fileSize;
+}
+#else
+static size_t FIO_setDictBufferMMap(FIO_Dict_t* dict, const char* fileName, FIO_prefs_t* const prefs, stat_t* dictFileStat)
+{
+   return FIO_setDictBufferMalloc(dict, fileName, prefs, dictFileStat);
+}
+static void FIO_munmap(FIO_Dict_t* dict) {
+   free(dict->dictBuffer);
+   dict->dictBuffer = NULL;
+   dict->dictBufferSize = 0;
+}
+#endif
+
+static void FIO_freeDict(FIO_Dict_t* dict) {
+    if (dict->dictBufferType == FIO_mallocDict) {
+        free(dict->dictBuffer);
+        dict->dictBuffer = NULL;
+        dict->dictBufferSize = 0;
+    } else if (dict->dictBufferType == FIO_mmapDict)  {
+        FIO_munmap(dict);
+    } else {
+        assert(0); /* Should not reach this case */
+    }
+}
+
+static void FIO_initDict(FIO_Dict_t* dict, const char* fileName, FIO_prefs_t* const prefs, stat_t* dictFileStat, FIO_dictBufferType_t dictBufferType) {
+    dict->dictBufferType = dictBufferType;
+    if (dict->dictBufferType == FIO_mallocDict) {
+        dict->dictBufferSize = FIO_setDictBufferMalloc(dict, fileName, prefs, dictFileStat);
+    } else if (dict->dictBufferType == FIO_mmapDict)  {
+        dict->dictBufferSize = FIO_setDictBufferMMap(dict, fileName, prefs, dictFileStat);
+    } else {
+        assert(0); /* Should not reach this case */
+    }
+}


 /* FIO_checkFilenameCollisions() :
@@ -718,7 +874,7 @@ int FIO_checkFilenameCollisions(const char** filenameTable, unsigned nbFiles) {

    filenameTableSorted = (const char**) malloc(sizeof(char*) * nbFiles);
    if (!filenameTableSorted) {
-        DISPLAY("Unable to malloc new str array, not checking for name collisions\n");
+        DISPLAYLEVEL(1, "Allocation error during filename collision checking \n");
        return 1;
    }

@@ -735,7 +891,7 @@ int FIO_checkFilenameCollisions(const char** filenameTable, unsigned nbFiles) {
    prevElem = filenameTableSorted[0];
    for (u = 1; u < nbFiles; ++u) {
        if (strcmp(prevElem, filenameTableSorted[u]) == 0) {
-            DISPLAY("WARNING: Two files have same filename: %s\n", prevElem);
+            DISPLAYLEVEL(2, "WARNING: Two files have same filename: %s\n", prevElem);
        }
        prevElem = filenameTableSorted[u];
    }
@@ -818,45 +974,89 @@ static void FIO_adjustMemLimitForPatchFromMode(FIO_prefs_t* const prefs,
    FIO_setMemLimit(prefs, (unsigned)maxSize);
 }

-/* FIO_removeMultiFilesWarning() :
+/* FIO_multiFilesConcatWarning() :
+ * This function handles logic when processing multiple files with -o or -c, displaying the appropriate warnings/prompts.
 * Returns 1 if the console should abort, 0 if console should proceed.
- * This function handles logic when processing multiple files with -o, displaying the appropriate warnings/prompts.
 *
- * If -f is specified, or there is just 1 file, zstd will always proceed as usual.
- * If --rm is specified, there will be a prompt asking for user confirmation.
- *         If -f is specified with --rm, zstd will proceed as usual
- *         If -q is specified with --rm, zstd will abort pre-emptively
- *         If neither flag is specified, zstd will prompt the user for confirmation to proceed.
- * If --rm is not specified, then zstd will print a warning to the user (which can be silenced with -q).
- * However, if the output is stdout, we will always abort rather than displaying the warning prompt.
+ * If output is stdout or test mode is active, check that `--rm` disabled.
+ *
+ * If there is just 1 file to process, zstd will proceed as usual.
+ * If each file get processed into its own separate destination file, proceed as usual.
+ *
+ * When multiple files are processed into a single output,
+ * display a warning message, then disable --rm if it's set.
+ *
+ * If -f is specified or if output is stdout, just proceed.
+ * If output is set with -o, prompt for confirmation.
 */
-static int FIO_removeMultiFilesWarning(FIO_ctx_t* const fCtx, const FIO_prefs_t* const prefs, const char* outFileName, int displayLevelCutoff)
+static int FIO_multiFilesConcatWarning(const FIO_ctx_t* fCtx, FIO_prefs_t* prefs, const char* outFileName, int displayLevelCutoff)
 {
-    int error = 0;
-    if (fCtx->nbFilesTotal > 1 && !prefs->overwrite) {
-        if (g_display_prefs.displayLevel <= displayLevelCutoff) {
-            if (prefs->removeSrcFile) {
-                DISPLAYLEVEL(1, "zstd: Aborting... not deleting files and processing into dst: %s\n", outFileName);
-                error =  1;
-            }
-        } else {
-            if (!strcmp(outFileName, stdoutmark)) {
-                DISPLAYLEVEL(2, "zstd: WARNING: all input files will be processed and concatenated into stdout. \n");
-            } else {
-                DISPLAYLEVEL(2, "zstd: WARNING: all input files will be processed and concatenated into a single output file: %s \n", outFileName);
-            }
-            DISPLAYLEVEL(2, "The concatenated output CANNOT regenerate the original directory tree. \n")
-            if (prefs->removeSrcFile) {
-                if (fCtx->hasStdoutOutput) {
-                    DISPLAYLEVEL(1, "Aborting. Use -f if you really want to delete the files and output to stdout\n");
-                    error = 1;
-                } else {
-                    error = g_display_prefs.displayLevel > displayLevelCutoff && UTIL_requireUserConfirmation("This is a destructive operation. Proceed? (y/n): ", "Aborting...", "yY", fCtx->hasStdinInput);
-                }
-            }
-        }
+    if (fCtx->hasStdoutOutput) {
+        if (prefs->removeSrcFile)
+            /* this should not happen ; hard fail, to protect user's data
+             * note: this should rather be an assert(), but we want to be certain that user's data will not be wiped out in case it nonetheless happen */
+            EXM_THROW(43, "It's not allowed to remove input files when processed output is piped to stdout. "
+                "This scenario is not supposed to be possible. "
+                "This is a programming error. File an issue for it to be fixed.");
    }
-    return error;
+    if (prefs->testMode) {
+        if (prefs->removeSrcFile)
+            /* this should not happen ; hard fail, to protect user's data
+             * note: this should rather be an assert(), but we want to be certain that user's data will not be wiped out in case it nonetheless happen */
+            EXM_THROW(43, "Test mode shall not remove input files! "
+                 "This scenario is not supposed to be possible. "
+                 "This is a programming error. File an issue for it to be fixed.");
+        return 0;
+    }
+
+    if (fCtx->nbFilesTotal == 1) return 0;
+    assert(fCtx->nbFilesTotal > 1);
+
+    if (!outFileName) return 0;
+
+    if (fCtx->hasStdoutOutput) {
+        DISPLAYLEVEL(2, "zstd: WARNING: all input files will be processed and concatenated into stdout. \n");
+    } else {
+        DISPLAYLEVEL(2, "zstd: WARNING: all input files will be processed and concatenated into a single output file: %s \n", outFileName);
+    }
+    DISPLAYLEVEL(2, "The concatenated output CANNOT regenerate original file names nor directory structure. \n")
+
+    /* multi-input into single output : --rm is not allowed */
+    if (prefs->removeSrcFile) {
+        DISPLAYLEVEL(2, "Since it's a destructive operation, input files will not be removed. \n");
+        prefs->removeSrcFile = 0;
+    }
+
+    if (fCtx->hasStdoutOutput) return 0;
+    if (prefs->overwrite) return 0;
+
+    /* multiple files concatenated into single destination file using -o without -f */
+    if (g_display_prefs.displayLevel <= displayLevelCutoff) {
+        /* quiet mode => no prompt => fail automatically */
+        DISPLAYLEVEL(1, "Concatenating multiple processed inputs into a single output loses file metadata. \n");
+        DISPLAYLEVEL(1, "Aborting. \n");
+        return 1;
+    }
+    /* normal mode => prompt */
+    return UTIL_requireUserConfirmation("Proceed? (y/n): ", "Aborting...", "yY", fCtx->hasStdinInput);
+}
+
+static ZSTD_inBuffer setInBuffer(const void* buf, size_t s, size_t pos)
+{
+    ZSTD_inBuffer i;
+    i.src = buf;
+    i.size = s;
+    i.pos = pos;
+    return i;
+}
+
+static ZSTD_outBuffer setOutBuffer(void* buf, size_t s, size_t pos)
+{
+    ZSTD_outBuffer o;
+    o.dst = buf;
+    o.size = s;
+    o.pos = pos;
+    return o;
 }

 #ifndef ZSTD_NOCOMPRESS
@@ -865,9 +1065,9 @@ static int FIO_removeMultiFilesWarning(FIO_ctx_t* const fCtx, const FIO_prefs_t*
 *  Compression
 ************************************************************************/
 typedef struct {
-    void* dictBuffer;
-    size_t dictBufferSize;
+    FIO_Dict_t dict;
    const char* dictFileName;
+    stat_t dictFileStat;
    ZSTD_CStream* cctx;
    WritePoolCtx_t *writeCtx;
    ReadPoolCtx_t *readCtx;
@@ -911,6 +1111,9 @@ static void FIO_adjustParamsForPatchFromMode(FIO_prefs_t* const prefs,
 static cRess_t FIO_createCResources(FIO_prefs_t* const prefs,
                                    const char* dictFileName, unsigned long long const maxSrcFileSize,
                                    int cLevel, ZSTD_compressionParameters comprParams) {
+    int useMMap = prefs->mmapDict == ZSTD_ps_enable;
+    int forceNoUseMMap = prefs->mmapDict == ZSTD_ps_disable;
+    FIO_dictBufferType_t dictBufferType;
    cRess_t ress;
    memset(&ress, 0, sizeof(ress));

@@ -920,19 +1123,25 @@ static cRess_t FIO_createCResources(FIO_prefs_t* const prefs,
        EXM_THROW(30, "allocation error (%s): can't create ZSTD_CCtx",
                    strerror(errno));

+    FIO_getDictFileStat(dictFileName, &ress.dictFileStat);
+
    /* need to update memLimit before calling createDictBuffer
     * because of memLimit check inside it */
    if (prefs->patchFromMode) {
+        U64 const dictSize = UTIL_getFileSizeStat(&ress.dictFileStat);
        unsigned long long const ssSize = (unsigned long long)prefs->streamSrcSize;
-        FIO_adjustParamsForPatchFromMode(prefs, &comprParams, UTIL_getFileSize(dictFileName), ssSize > 0 ? ssSize : maxSrcFileSize, cLevel);
+        useMMap |= dictSize > prefs->memLimit;
+        FIO_adjustParamsForPatchFromMode(prefs, &comprParams, dictSize, ssSize > 0 ? ssSize : maxSrcFileSize, cLevel);
    }
-    ress.dictBufferSize = FIO_createDictBuffer(&ress.dictBuffer, dictFileName, prefs);   /* works with dictFileName==NULL */
+
+    dictBufferType = (useMMap && !forceNoUseMMap) ? FIO_mmapDict : FIO_mallocDict;
+    FIO_initDict(&ress.dict, dictFileName, prefs, &ress.dictFileStat, dictBufferType);   /* works with dictFileName==NULL */

    ress.writeCtx = AIO_WritePool_create(prefs, ZSTD_CStreamOutSize());
    ress.readCtx = AIO_ReadPool_create(prefs, ZSTD_CStreamInSize());

    /* Advanced parameters, including dictionary */
-    if (dictFileName && (ress.dictBuffer==NULL))
+    if (dictFileName && (ress.dict.dictBuffer==NULL))
        EXM_THROW(32, "allocation error : can't create dictBuffer");
    ress.dictFileName = dictFileName;

@@ -982,17 +1191,17 @@ static cRess_t FIO_createCResources(FIO_prefs_t* const prefs,
 #endif
    /* dictionary */
    if (prefs->patchFromMode) {
-        CHECK( ZSTD_CCtx_refPrefix(ress.cctx, ress.dictBuffer, ress.dictBufferSize) );
+        CHECK( ZSTD_CCtx_refPrefix(ress.cctx, ress.dict.dictBuffer, ress.dict.dictBufferSize) );
    } else {
-        CHECK( ZSTD_CCtx_loadDictionary(ress.cctx, ress.dictBuffer, ress.dictBufferSize) );
+        CHECK( ZSTD_CCtx_loadDictionary_byReference(ress.cctx, ress.dict.dictBuffer, ress.dict.dictBufferSize) );
    }

    return ress;
 }

-static void FIO_freeCResources(const cRess_t* const ress)
+static void FIO_freeCResources(cRess_t* const ress)
 {
-    free(ress->dictBuffer);
+    FIO_freeDict(&(ress->dict));
    AIO_WritePool_free(ress->writeCtx);
    AIO_ReadPool_free(ress->readCtx);
    ZSTD_freeCStream(ress->cctx);   /* never fails */
@@ -1123,8 +1332,8 @@ FIO_compressLzmaFrame(cRess_t* ress,
    }

    writeJob =AIO_WritePool_acquireJob(ress->writeCtx);
-    strm.next_out = (Bytef*)writeJob->buffer;
-    strm.avail_out = (uInt)writeJob->bufferSize;
+    strm.next_out = (BYTE*)writeJob->buffer;
+    strm.avail_out = writeJob->bufferSize;
    strm.next_in = 0;
    strm.avail_in = 0;

@@ -1151,7 +1360,7 @@ FIO_compressLzmaFrame(cRess_t* ress,
                writeJob->usedBufferSize = compBytes;
                AIO_WritePool_enqueueAndReacquireWriteJob(&writeJob);
                outFileSize += compBytes;
-                strm.next_out = (Bytef*)writeJob->buffer;
+                strm.next_out = (BYTE*)writeJob->buffer;
                strm.avail_out = writeJob->bufferSize;
        }   }
        if (srcFileSize == UTIL_FILESIZE_UNKNOWN)
@@ -1277,7 +1486,6 @@ FIO_compressLz4Frame(cRess_t* ress,
 }
 #endif

-
 static unsigned long long
 FIO_compressZstdFrame(FIO_ctx_t* const fCtx,
                      FIO_prefs_t* const prefs,
@@ -1341,7 +1549,7 @@ FIO_compressZstdFrame(FIO_ctx_t* const fCtx,
        size_t stillToFlush;
        /* Fill input Buffer */
        size_t const inSize = AIO_ReadPool_fillBuffer(ress.readCtx, ZSTD_CStreamInSize());
-        ZSTD_inBuffer inBuff = { ress.readCtx->srcBuffer, ress.readCtx->srcBufferLoaded, 0 };
+        ZSTD_inBuffer inBuff = setInBuffer( ress.readCtx->srcBuffer, ress.readCtx->srcBufferLoaded, 0 );
        DISPLAYLEVEL(6, "fread %u bytes from source \n", (unsigned)inSize);
        *readsize += inSize;

@@ -1353,7 +1561,7 @@ FIO_compressZstdFrame(FIO_ctx_t* const fCtx,
            || (directive == ZSTD_e_end && stillToFlush != 0) ) {

            size_t const oldIPos = inBuff.pos;
-            ZSTD_outBuffer outBuff= { writeJob->buffer, writeJob->bufferSize, 0 };
+            ZSTD_outBuffer outBuff = setOutBuffer( writeJob->buffer, writeJob->bufferSize, 0 );
            size_t const toFlushNow = ZSTD_toFlushNow(ress.cctx);
            CHECK_V(stillToFlush, ZSTD_compressStream2(ress.cctx, &outBuff, &inBuff, directive));
            AIO_ReadPool_consumeBytes(ress.readCtx, inBuff.pos - oldIPos);
@@ -1625,28 +1833,30 @@ static int FIO_compressFilename_dstFile(FIO_ctx_t* const fCtx,
                                        cRess_t ress,
                                        const char* dstFileName,
                                        const char* srcFileName,
+                                        const stat_t* srcFileStat,
                                        int compressionLevel)
 {
    int closeDstFile = 0;
    int result;
-    stat_t statbuf;
-    int transferMTime = 0;
+    int transferStat = 0;
    FILE *dstFile;
+    int dstFd = -1;
+
    assert(AIO_ReadPool_getFile(ress.readCtx) != NULL);
    if (AIO_WritePool_getFile(ress.writeCtx) == NULL) {
-        int dstFilePermissions = DEFAULT_FILE_PERMISSIONS;
+        int dstFileInitialPermissions = DEFAULT_FILE_PERMISSIONS;
        if ( strcmp (srcFileName, stdinmark)
          && strcmp (dstFileName, stdoutmark)
-          && UTIL_stat(srcFileName, &statbuf)
-          && UTIL_isRegularFileStat(&statbuf) ) {
-            dstFilePermissions = statbuf.st_mode;
-            transferMTime = 1;
+          && UTIL_isRegularFileStat(srcFileStat) ) {
+            transferStat = 1;
+            dstFileInitialPermissions = TEMPORARY_FILE_PERMISSIONS;
        }

        closeDstFile = 1;
        DISPLAYLEVEL(6, "FIO_compressFilename_dstFile: opening dst: %s \n", dstFileName);
-        dstFile = FIO_openDstFile(fCtx, prefs, srcFileName, dstFileName, dstFilePermissions);
+        dstFile = FIO_openDstFile(fCtx, prefs, srcFileName, dstFileName, dstFileInitialPermissions);
        if (dstFile==NULL) return 1;  /* could not open dstFileName */
+        dstFd = fileno(dstFile);
        AIO_WritePool_setFile(ress.writeCtx, dstFile);
        /* Must only be added after FIO_openDstFile() succeeds.
         * Otherwise we may delete the destination file if it already exists,
@@ -1660,14 +1870,20 @@ static int FIO_compressFilename_dstFile(FIO_ctx_t* const fCtx,
    if (closeDstFile) {
        clearHandler();

+        if (transferStat) {
+            UTIL_setFDStat(dstFd, dstFileName, srcFileStat);
+        }
+
        DISPLAYLEVEL(6, "FIO_compressFilename_dstFile: closing dst: %s \n", dstFileName);
        if (AIO_WritePool_closeFile(ress.writeCtx)) { /* error closing file */
            DISPLAYLEVEL(1, "zstd: %s: %s \n", dstFileName, strerror(errno));
            result=1;
        }
-        if (transferMTime) {
-            UTIL_utime(dstFileName, &statbuf);
+
+        if (transferStat) {
+            UTIL_utime(dstFileName, srcFileStat);
        }
+
        if ( (result != 0)  /* operation failure */
          && strcmp(dstFileName, stdoutmark)  /* special case : don't remove() stdout */
          ) {
@@ -1708,18 +1924,26 @@ FIO_compressFilename_srcFile(FIO_ctx_t* const fCtx,
 {
    int result;
    FILE* srcFile;
+    stat_t srcFileStat;
+    U64 fileSize = UTIL_FILESIZE_UNKNOWN;
    DISPLAYLEVEL(6, "FIO_compressFilename_srcFile: %s \n", srcFileName);

-    /* ensure src is not a directory */
-    if (UTIL_isDirectory(srcFileName)) {
-        DISPLAYLEVEL(1, "zstd: %s is a directory -- ignored \n", srcFileName);
-        return 1;
-    }
+    if (strcmp(srcFileName, stdinmark)) {
+        if (UTIL_stat(srcFileName, &srcFileStat)) {
+            /* failure to stat at all is handled during opening */

-    /* ensure src is not the same as dict (if present) */
-    if (ress.dictFileName != NULL && UTIL_isSameFile(srcFileName, ress.dictFileName)) {
-        DISPLAYLEVEL(1, "zstd: cannot use %s as an input file and dictionary \n", srcFileName);
-        return 1;
+            /* ensure src is not a directory */
+            if (UTIL_isDirectoryStat(&srcFileStat)) {
+                DISPLAYLEVEL(1, "zstd: %s is a directory -- ignored \n", srcFileName);
+                return 1;
+            }
+
+            /* ensure src is not the same as dict (if present) */
+            if (ress.dictFileName != NULL && UTIL_isSameFileStat(srcFileName, ress.dictFileName, &srcFileStat, &ress.dictFileStat)) {
+                DISPLAYLEVEL(1, "zstd: cannot use %s as an input file and dictionary \n", srcFileName);
+                return 1;
+            }
+        }
    }

    /* Check if "srcFile" is compressed. Only done if --exclude-compressed flag is used
@@ -1731,16 +1955,30 @@ FIO_compressFilename_srcFile(FIO_ctx_t* const fCtx,
        return 0;
    }

-    srcFile = FIO_openSrcFile(prefs, srcFileName);
+    srcFile = FIO_openSrcFile(prefs, srcFileName, &srcFileStat);
    if (srcFile == NULL) return 1;   /* srcFile could not be opened */

+    /* Don't use AsyncIO for small files */
+    if (strcmp(srcFileName, stdinmark)) /* Stdin doesn't have stats */
+        fileSize = UTIL_getFileSizeStat(&srcFileStat);
+    if(fileSize != UTIL_FILESIZE_UNKNOWN && fileSize < ZSTD_BLOCKSIZE_MAX * 3) {
+        AIO_ReadPool_setAsync(ress.readCtx, 0);
+        AIO_WritePool_setAsync(ress.writeCtx, 0);
+    } else {
+        AIO_ReadPool_setAsync(ress.readCtx, 1);
+        AIO_WritePool_setAsync(ress.writeCtx, 1);
+    }
+
    AIO_ReadPool_setFile(ress.readCtx, srcFile);
-    result = FIO_compressFilename_dstFile(fCtx, prefs, ress, dstFileName, srcFileName, compressionLevel);
+    result = FIO_compressFilename_dstFile(
+            fCtx, prefs, ress,
+            dstFileName, srcFileName,
+            &srcFileStat, compressionLevel);
    AIO_ReadPool_closeFile(ress.readCtx);

-    if ( prefs->removeSrcFile   /* --rm */
-      && result == 0       /* success */
-      && strcmp(srcFileName, stdinmark)   /* exception : don't erase stdin */
+    if ( prefs->removeSrcFile  /* --rm */
+      && result == 0           /* success */
+      && strcmp(srcFileName, stdinmark)  /* exception : don't erase stdin */
      ) {
        /* We must clear the handler, since after this point calling it would
         * delete both the source and destination files.
@@ -1762,7 +2000,8 @@ checked_index(const char* options[], size_t length, size_t index) {

 #define INDEX(options, index) checked_index((options), sizeof(options)  / sizeof(char*), (size_t)(index))

-void FIO_displayCompressionParameters(const FIO_prefs_t* prefs) {
+void FIO_displayCompressionParameters(const FIO_prefs_t* prefs)
+{
    static const char* formatOptions[5] = {ZSTD_EXTENSION, GZ_EXTENSION, XZ_EXTENSION,
                                           LZMA_EXTENSION, LZ4_EXTENSION};
    static const char* sparseOptions[3] = {" --no-sparse", "", " --sparse"};
@@ -1801,7 +2040,7 @@ int FIO_compressFilename(FIO_ctx_t* const fCtx, FIO_prefs_t* const prefs, const
                         const char* srcFileName, const char* dictFileName,
                         int compressionLevel, ZSTD_compressionParameters comprParams)
 {
-    cRess_t const ress = FIO_createCResources(prefs, dictFileName, UTIL_getFileSize(srcFileName), compressionLevel, comprParams);
+    cRess_t ress = FIO_createCResources(prefs, dictFileName, UTIL_getFileSize(srcFileName), compressionLevel, comprParams);
    int const result = FIO_compressFilename_srcFile(fCtx, prefs, ress, dstFileName, srcFileName, compressionLevel);

 #define DISPLAY_LEVEL_DEFAULT 2
@@ -1891,7 +2130,7 @@ int FIO_compressMultipleFilenames(FIO_ctx_t* const fCtx,
    assert(outFileName != NULL || suffix != NULL);
    if (outFileName != NULL) {   /* output into a single destination (stdout typically) */
        FILE *dstFile;
-        if (FIO_removeMultiFilesWarning(fCtx, prefs, outFileName, 1 /* displayLevelCutoff */)) {
+        if (FIO_multiFilesConcatWarning(fCtx, prefs, outFileName, 1 /* displayLevelCutoff */)) {
            FIO_freeCResources(&ress);
            return 1;
        }
@@ -1971,6 +2210,7 @@ int FIO_compressMultipleFilenames(FIO_ctx_t* const fCtx,
 *  Decompression
 ***************************************************************************/
 typedef struct {
+    FIO_Dict_t dict;
    ZSTD_DStream* dctx;
    WritePoolCtx_t *writeCtx;
    ReadPoolCtx_t *readCtx;
@@ -1978,11 +2218,19 @@ typedef struct {

 static dRess_t FIO_createDResources(FIO_prefs_t* const prefs, const char* dictFileName)
 {
+    int useMMap = prefs->mmapDict == ZSTD_ps_enable;
+    int forceNoUseMMap = prefs->mmapDict == ZSTD_ps_disable;
+    stat_t statbuf;
    dRess_t ress;
    memset(&ress, 0, sizeof(ress));

-    if (prefs->patchFromMode)
-        FIO_adjustMemLimitForPatchFromMode(prefs, UTIL_getFileSize(dictFileName), 0 /* just use the dict size */);
+    FIO_getDictFileStat(dictFileName, &statbuf);
+
+    if (prefs->patchFromMode){
+        U64 const dictSize = UTIL_getFileSizeStat(&statbuf);
+        useMMap |= dictSize > prefs->memLimit;
+        FIO_adjustMemLimitForPatchFromMode(prefs, dictSize, 0 /* just use the dict size */);
+    }

    /* Allocation */
    ress.dctx = ZSTD_createDStream();
@@ -1992,27 +2240,34 @@ static dRess_t FIO_createDResources(FIO_prefs_t* const prefs, const char* dictFi
    CHECK( ZSTD_DCtx_setParameter(ress.dctx, ZSTD_d_forceIgnoreChecksum, !prefs->checksumFlag));

    /* dictionary */
-    {   void* dictBuffer;
-        size_t const dictBufferSize = FIO_createDictBuffer(&dictBuffer, dictFileName, prefs);
-        CHECK( ZSTD_initDStream_usingDict(ress.dctx, dictBuffer, dictBufferSize) );
-        free(dictBuffer);
+    {
+        FIO_dictBufferType_t dictBufferType = (useMMap && !forceNoUseMMap) ? FIO_mmapDict : FIO_mallocDict;
+        FIO_initDict(&ress.dict, dictFileName, prefs, &statbuf, dictBufferType);
+
+        CHECK(ZSTD_DCtx_reset(ress.dctx, ZSTD_reset_session_only) );
+
+        if (prefs->patchFromMode){
+            CHECK(ZSTD_DCtx_refPrefix(ress.dctx, ress.dict.dictBuffer, ress.dict.dictBufferSize));
+        } else {
+            CHECK(ZSTD_DCtx_loadDictionary_byReference(ress.dctx, ress.dict.dictBuffer, ress.dict.dictBufferSize));
+        }
    }

    ress.writeCtx = AIO_WritePool_create(prefs, ZSTD_DStreamOutSize());
    ress.readCtx = AIO_ReadPool_create(prefs, ZSTD_DStreamInSize());
-
    return ress;
 }

 static void FIO_freeDResources(dRess_t ress)
 {
+    FIO_freeDict(&(ress.dict));
    CHECK( ZSTD_freeDStream(ress.dctx) );
    AIO_WritePool_free(ress.writeCtx);
    AIO_ReadPool_free(ress.readCtx);
 }

-/** FIO_passThrough() : just copy input into output, for compatibility with gzip -df mode
-    @return : 0 (no error) */
+/* FIO_passThrough() : just copy input into output, for compatibility with gzip -df mode
+ * @return : 0 (no error) */
 static int FIO_passThrough(dRess_t *ress)
 {
    size_t const blockSize = MIN(MIN(64 KB, ZSTD_DStreamInSize()), ZSTD_DStreamOutSize());
@@ -2040,7 +2295,8 @@ static int FIO_passThrough(dRess_t *ress)
 static void
 FIO_zstdErrorHelp(const FIO_prefs_t* const prefs,
                  const dRess_t* ress,
-                  size_t err, const char* srcFileName)
+                  size_t err,
+                  const char* srcFileName)
 {
    ZSTD_frameHeader header;

@@ -2092,8 +2348,8 @@ FIO_decompressZstdFrame(FIO_ctx_t* const fCtx, dRess_t* ress,

    /* Main decompression Loop */
    while (1) {
-        ZSTD_inBuffer  inBuff = { ress->readCtx->srcBuffer, ress->readCtx->srcBufferLoaded, 0 };
-        ZSTD_outBuffer outBuff= { writeJob->buffer, writeJob->bufferSize, 0 };
+        ZSTD_inBuffer  inBuff = setInBuffer( ress->readCtx->srcBuffer, ress->readCtx->srcBufferLoaded, 0 );
+        ZSTD_outBuffer outBuff= setOutBuffer( writeJob->buffer, writeJob->bufferSize, 0 );
        size_t const readSizeHint = ZSTD_decompressStream(ress->dctx, &outBuff, &inBuff);
        UTIL_HumanReadableSize_t const hrs = UTIL_makeHumanReadableSize(alreadyDecoded+frameSize);
        if (ZSTD_isError(readSizeHint)) {
@@ -2242,8 +2498,8 @@ FIO_decompressLzmaFrame(dRess_t* ress,
    }

    writeJob = AIO_WritePool_acquireJob(ress->writeCtx);
-    strm.next_out = (Bytef*)writeJob->buffer;
-    strm.avail_out = (uInt)writeJob->bufferSize;
+    strm.next_out = (BYTE*)writeJob->buffer;
+    strm.avail_out = writeJob->bufferSize;
    strm.next_in = (BYTE const*)ress->readCtx->srcBuffer;
    strm.avail_in = ress->readCtx->srcBufferLoaded;

@@ -2271,7 +2527,7 @@ FIO_decompressLzmaFrame(dRess_t* ress,
                writeJob->usedBufferSize = decompBytes;
                AIO_WritePool_enqueueAndReacquireWriteJob(&writeJob);
                outFileSize += decompBytes;
-                strm.next_out = (Bytef*)writeJob->buffer;
+                strm.next_out = (BYTE*)writeJob->buffer;
                strm.avail_out = writeJob->bufferSize;
        }   }
        if (ret == LZMA_STREAM_END) break;
@@ -2459,28 +2715,30 @@ static int FIO_decompressFrames(FIO_ctx_t* const fCtx,
 static int FIO_decompressDstFile(FIO_ctx_t* const fCtx,
                                 FIO_prefs_t* const prefs,
                                 dRess_t ress,
-                                 const char* dstFileName, const char* srcFileName)
+                                 const char* dstFileName,
+                                 const char* srcFileName,
+                                 const stat_t* srcFileStat)
 {
    int result;
-    stat_t statbuf;
    int releaseDstFile = 0;
-    int transferMTime = 0;
+    int transferStat = 0;
+    int dstFd = 0;

    if ((AIO_WritePool_getFile(ress.writeCtx) == NULL) && (prefs->testMode == 0)) {
        FILE *dstFile;
        int dstFilePermissions = DEFAULT_FILE_PERMISSIONS;
        if ( strcmp(srcFileName, stdinmark)   /* special case : don't transfer permissions from stdin */
          && strcmp(dstFileName, stdoutmark)
-          && UTIL_stat(srcFileName, &statbuf)
-          && UTIL_isRegularFileStat(&statbuf) ) {
-            dstFilePermissions = statbuf.st_mode;
-            transferMTime = 1;
+          && UTIL_isRegularFileStat(srcFileStat) ) {
+            transferStat = 1;
+            dstFilePermissions = TEMPORARY_FILE_PERMISSIONS;
        }

        releaseDstFile = 1;

        dstFile = FIO_openDstFile(fCtx, prefs, srcFileName, dstFileName, dstFilePermissions);
        if (dstFile==NULL) return 1;
+        dstFd = fileno(dstFile);
        AIO_WritePool_setFile(ress.writeCtx, dstFile);

        /* Must only be added after FIO_openDstFile() succeeds.
@@ -2494,13 +2752,18 @@ static int FIO_decompressDstFile(FIO_ctx_t* const fCtx,

    if (releaseDstFile) {
        clearHandler();
+
+        if (transferStat) {
+            UTIL_setFDStat(dstFd, dstFileName, srcFileStat);
+        }
+
        if (AIO_WritePool_closeFile(ress.writeCtx)) {
            DISPLAYLEVEL(1, "zstd: %s: %s \n", dstFileName, strerror(errno));
            result = 1;
        }

-        if (transferMTime) {
-            UTIL_utime(dstFileName, &statbuf);
+        if (transferStat) {
+            UTIL_utime(dstFileName, srcFileStat);
        }

        if ( (result != 0)  /* operation failure */
@@ -2522,18 +2785,32 @@ static int FIO_decompressDstFile(FIO_ctx_t* const fCtx,
 static int FIO_decompressSrcFile(FIO_ctx_t* const fCtx, FIO_prefs_t* const prefs, dRess_t ress, const char* dstFileName, const char* srcFileName)
 {
    FILE* srcFile;
+    stat_t srcFileStat;
    int result;
+    U64 fileSize = UTIL_FILESIZE_UNKNOWN;

    if (UTIL_isDirectory(srcFileName)) {
        DISPLAYLEVEL(1, "zstd: %s is a directory -- ignored \n", srcFileName);
        return 1;
    }

-    srcFile = FIO_openSrcFile(prefs, srcFileName);
+    srcFile = FIO_openSrcFile(prefs, srcFileName, &srcFileStat);
    if (srcFile==NULL) return 1;
+
+    /* Don't use AsyncIO for small files */
+    if (strcmp(srcFileName, stdinmark)) /* Stdin doesn't have stats */
+        fileSize = UTIL_getFileSizeStat(&srcFileStat);
+    if(fileSize != UTIL_FILESIZE_UNKNOWN && fileSize < ZSTD_BLOCKSIZE_MAX * 3) {
+        AIO_ReadPool_setAsync(ress.readCtx, 0);
+        AIO_WritePool_setAsync(ress.writeCtx, 0);
+    } else {
+        AIO_ReadPool_setAsync(ress.readCtx, 1);
+        AIO_WritePool_setAsync(ress.writeCtx, 1);
+    }
+
    AIO_ReadPool_setFile(ress.readCtx, srcFile);

-    result = FIO_decompressDstFile(fCtx, prefs, ress, dstFileName, srcFileName);
+    result = FIO_decompressDstFile(fCtx, prefs, ress, dstFileName, srcFileName, &srcFileStat);

    AIO_ReadPool_setFile(ress.readCtx, NULL);

@@ -2567,6 +2844,8 @@ int FIO_decompressFilename(FIO_ctx_t* const fCtx, FIO_prefs_t* const prefs,

    int const decodingError = FIO_decompressSrcFile(fCtx, prefs, ress, dstFileName, srcFileName);

+
+
    FIO_freeDResources(ress);
    return decodingError;
 }
@@ -2710,7 +2989,7 @@ FIO_decompressMultipleFilenames(FIO_ctx_t* const fCtx,
    dRess_t ress = FIO_createDResources(prefs, dictFileName);

    if (outFileName) {
-        if (FIO_removeMultiFilesWarning(fCtx, prefs, outFileName, 1 /* displayLevelCutoff */)) {
+        if (FIO_multiFilesConcatWarning(fCtx, prefs, outFileName, 1 /* displayLevelCutoff */)) {
            FIO_freeDResources(ress);
            return 1;
        }
@@ -2756,7 +3035,8 @@ FIO_decompressMultipleFilenames(FIO_ctx_t* const fCtx,

    if (FIO_shouldDisplayMultipleFileSummary(fCtx)) {
        DISPLAY_PROGRESS("\r%79s\r", "");
-        DISPLAY_SUMMARY("%d files decompressed : %6zu bytes total \n", fCtx->nbFilesProcessed, fCtx->totalBytesOutput);
+        DISPLAY_SUMMARY("%d files decompressed : %6llu bytes total \n",
+            fCtx->nbFilesProcessed, (unsigned long long)fCtx->totalBytesOutput);
    }

    FIO_freeDResources(ress);
@@ -2785,7 +3065,7 @@ typedef enum {
  info_frame_error=1,
  info_not_zstd=2,
  info_file_error=3,
-  info_truncated_input=4,
+  info_truncated_input=4
 } InfoError;

 #define ERROR_IF(c,n,...) {             \
@@ -2897,10 +3177,11 @@ static InfoError
 getFileInfo_fileConfirmed(fileInfo_t* info, const char* inFileName)
 {
    InfoError status;
-    FILE* const srcFile = FIO_openSrcFile(NULL, inFileName);
+    stat_t srcFileStat;
+    FILE* const srcFile = FIO_openSrcFile(NULL, inFileName, &srcFileStat);
    ERROR_IF(srcFile == NULL, info_file_error, "Error: could not open source file %s", inFileName);

-    info->compressedSize = UTIL_getFileSize(inFileName);
+    info->compressedSize = UTIL_getFileSizeStat(&srcFileStat);
    status = FIO_analyzeFrames(info, srcFile);

    fclose(srcFile);
@@ -86,7 +86,7 @@ void FIO_setLdmMinMatch(FIO_prefs_t* const prefs, int ldmMinMatch);
 void FIO_setMemLimit(FIO_prefs_t* const prefs, unsigned memLimit);
 void FIO_setNbWorkers(FIO_prefs_t* const prefs, int nbWorkers);
 void FIO_setOverlapLog(FIO_prefs_t* const prefs, int overlapLog);
-void FIO_setRemoveSrcFile(FIO_prefs_t* const prefs, unsigned flag);
+void FIO_setRemoveSrcFile(FIO_prefs_t* const prefs, int flag);
 void FIO_setSparseWrite(FIO_prefs_t* const prefs, int sparse);  /**< 0: no sparse; 1: disable on stdout; 2: always enabled */
 void FIO_setRsyncable(FIO_prefs_t* const prefs, int rsyncable);
 void FIO_setStreamSrcSize(FIO_prefs_t* const prefs, size_t streamSrcSize);
@@ -106,6 +106,7 @@ void FIO_setContentSize(FIO_prefs_t* const prefs, int value);
 void FIO_displayCompressionParameters(const FIO_prefs_t* prefs);
 void FIO_setAsyncIOFlag(FIO_prefs_t* const prefs, int value);
 void FIO_setPassThroughFlag(FIO_prefs_t* const prefs, int value);
+void FIO_setMMapDict(FIO_prefs_t* const prefs, ZSTD_paramSwitch_e value);

 /* FIO_ctx_t functions */
 void FIO_setNbFilesTotal(FIO_ctx_t* const fCtx, int value);
@@ -140,7 +140,7 @@ int AIO_supported(void) {
 }

 /* ***********************************
- *  General IoPool implementation
+ *  Generic IoPool implementation
 *************************************/

 static IOJob_t *AIO_IOPool_createIoJob(IOPoolCtx_t *ctx, size_t bufferSize) {
@@ -163,20 +163,22 @@ static IOJob_t *AIO_IOPool_createIoJob(IOPoolCtx_t *ctx, size_t bufferSize) {
 * Displays warning if asyncio is requested but MT isn't available. */
 static void AIO_IOPool_createThreadPool(IOPoolCtx_t* ctx, const FIO_prefs_t* prefs) {
    ctx->threadPool = NULL;
+    ctx->threadPoolActive = 0;
    if(prefs->asyncIO) {
        if (ZSTD_pthread_mutex_init(&ctx->ioJobsMutex, NULL))
-            EXM_THROW(102,"Failed creating write availableJobs mutex");
+            EXM_THROW(102,"Failed creating ioJobsMutex mutex");
        /* We want MAX_IO_JOBS-2 queue items because we need to always have 1 free buffer to
         * decompress into and 1 buffer that's actively written to disk and owned by the writing thread. */
        assert(MAX_IO_JOBS >= 2);
        ctx->threadPool = POOL_create(1, MAX_IO_JOBS - 2);
+        ctx->threadPoolActive = 1;
        if (!ctx->threadPool)
-            EXM_THROW(104, "Failed creating writer thread pool");
+            EXM_THROW(104, "Failed creating I/O thread pool");
    }
 }

 /* AIO_IOPool_init:
- * Allocates and sets and a new write pool including its included availableJobs. */
+ * Allocates and sets and a new I/O thread pool including its included availableJobs. */
 static void AIO_IOPool_init(IOPoolCtx_t* ctx, const FIO_prefs_t* prefs, POOL_function poolFunction, size_t bufferSize) {
    int i;
    AIO_IOPool_createThreadPool(ctx, prefs);
@@ -192,27 +194,59 @@ static void AIO_IOPool_init(IOPoolCtx_t* ctx, const FIO_prefs_t* prefs, POOL_fun
 }


+/* AIO_IOPool_threadPoolActive:
+ * Check if current operation uses thread pool.
+ * Note that in some cases we have a thread pool initialized but choose not to use it. */
+static int AIO_IOPool_threadPoolActive(IOPoolCtx_t* ctx) {
+    return ctx->threadPool && ctx->threadPoolActive;
+}
+
+
+/* AIO_IOPool_lockJobsMutex:
+ * Locks the IO jobs mutex if threading is active */
+static void AIO_IOPool_lockJobsMutex(IOPoolCtx_t* ctx) {
+    if(AIO_IOPool_threadPoolActive(ctx))
+        ZSTD_pthread_mutex_lock(&ctx->ioJobsMutex);
+}
+
+/* AIO_IOPool_unlockJobsMutex:
+ * Unlocks the IO jobs mutex if threading is active */
+static void AIO_IOPool_unlockJobsMutex(IOPoolCtx_t* ctx) {
+    if(AIO_IOPool_threadPoolActive(ctx))
+        ZSTD_pthread_mutex_unlock(&ctx->ioJobsMutex);
+}
+
 /* AIO_IOPool_releaseIoJob:
 * Releases an acquired job back to the pool. Doesn't execute the job. */
 static void AIO_IOPool_releaseIoJob(IOJob_t* job) {
    IOPoolCtx_t* const ctx = (IOPoolCtx_t *) job->ctx;
-    if(ctx->threadPool)
-        ZSTD_pthread_mutex_lock(&ctx->ioJobsMutex);
+    AIO_IOPool_lockJobsMutex(ctx);
    assert(ctx->availableJobsCount < ctx->totalIoJobs);
    ctx->availableJobs[ctx->availableJobsCount++] = job;
-    if(ctx->threadPool)
-        ZSTD_pthread_mutex_unlock(&ctx->ioJobsMutex);
+    AIO_IOPool_unlockJobsMutex(ctx);
 }

 /* AIO_IOPool_join:
 * Waits for all tasks in the pool to finish executing. */
 static void AIO_IOPool_join(IOPoolCtx_t* ctx) {
-    if(ctx->threadPool)
+    if(AIO_IOPool_threadPoolActive(ctx))
        POOL_joinJobs(ctx->threadPool);
 }

+/* AIO_IOPool_setThreaded:
+ * Allows (de)activating threaded mode, to be used when the expected overhead
+ * of threading costs more than the expected gains. */
+static void AIO_IOPool_setThreaded(IOPoolCtx_t* ctx, int threaded) {
+    assert(threaded == 0 || threaded == 1);
+    assert(ctx != NULL);
+    if(ctx->threadPoolActive != threaded) {
+        AIO_IOPool_join(ctx);
+        ctx->threadPoolActive = threaded;
+    }
+}
+
 /* AIO_IOPool_free:
- * Release a previously allocated write thread pool. Makes sure all takss are done and released. */
+ * Release a previously allocated IO thread pool. Makes sure all tasks are done and released. */
 static void AIO_IOPool_destroy(IOPoolCtx_t* ctx) {
    int i;
    if(ctx->threadPool) {
@@ -236,12 +270,10 @@ static void AIO_IOPool_destroy(IOPoolCtx_t* ctx) {
 static IOJob_t* AIO_IOPool_acquireJob(IOPoolCtx_t* ctx) {
    IOJob_t *job;
    assert(ctx->file != NULL || ctx->prefs->testMode);
-    if(ctx->threadPool)
-        ZSTD_pthread_mutex_lock(&ctx->ioJobsMutex);
+    AIO_IOPool_lockJobsMutex(ctx);
    assert(ctx->availableJobsCount > 0);
    job = (IOJob_t*) ctx->availableJobs[--ctx->availableJobsCount];
-    if(ctx->threadPool)
-        ZSTD_pthread_mutex_unlock(&ctx->ioJobsMutex);
+    AIO_IOPool_unlockJobsMutex(ctx);
    job->usedBufferSize = 0;
    job->file = ctx->file;
    job->offset = 0;
@@ -251,8 +283,7 @@ static IOJob_t* AIO_IOPool_acquireJob(IOPoolCtx_t* ctx) {

 /* AIO_IOPool_setFile:
 * Sets the destination file for future files in the pool.
- * Requires completion of all queues write jobs and release of all otherwise acquired jobs.
- * Also requires ending of sparse write if a previous file was used in sparse mode. */
+ * Requires completion of all queued jobs and release of all otherwise acquired jobs. */
 static void AIO_IOPool_setFile(IOPoolCtx_t* ctx, FILE* file) {
    assert(ctx!=NULL);
    AIO_IOPool_join(ctx);
@@ -269,7 +300,7 @@ static FILE* AIO_IOPool_getFile(const IOPoolCtx_t* ctx) {
 * The queued job shouldn't be used directly after queueing it. */
 static void AIO_IOPool_enqueueJob(IOJob_t* job) {
    IOPoolCtx_t* const ctx = (IOPoolCtx_t *)job->ctx;
-    if(ctx->threadPool)
+    if(AIO_IOPool_threadPoolActive(ctx))
        POOL_add(ctx->threadPool, ctx->poolFunction, job);
    else
        ctx->poolFunction(job);
@@ -300,8 +331,7 @@ void AIO_WritePool_enqueueAndReacquireWriteJob(IOJob_t **job) {
 * Blocks on completion of all current write jobs before executing. */
 void AIO_WritePool_sparseWriteEnd(WritePoolCtx_t* ctx) {
    assert(ctx != NULL);
-    if(ctx->base.threadPool)
-        POOL_joinJobs(ctx->base.threadPool);
+    AIO_IOPool_join(&ctx->base);
    AIO_fwriteSparseEnd(ctx->base.prefs, ctx->base.file, ctx->storedSkips);
    ctx->storedSkips = 0;
 }
@@ -368,6 +398,13 @@ void AIO_WritePool_free(WritePoolCtx_t* ctx) {
    free(ctx);
 }

+/* AIO_WritePool_setAsync:
+ * Allows (de)activating async mode, to be used when the expected overhead
+ * of asyncio costs more than the expected gains. */
+void AIO_WritePool_setAsync(WritePoolCtx_t* ctx, int async) {
+    AIO_IOPool_setThreaded(&ctx->base, async);
+}
+

 /* ***********************************
 *  ReadPool implementation
@@ -383,14 +420,13 @@ static void AIO_ReadPool_releaseAllCompletedJobs(ReadPoolCtx_t* ctx) {

 static void AIO_ReadPool_addJobToCompleted(IOJob_t* job) {
    ReadPoolCtx_t* const ctx = (ReadPoolCtx_t *)job->ctx;
-    if(ctx->base.threadPool)
-        ZSTD_pthread_mutex_lock(&ctx->base.ioJobsMutex);
+    AIO_IOPool_lockJobsMutex(&ctx->base);
    assert(ctx->completedJobsCount < MAX_IO_JOBS);
    ctx->completedJobs[ctx->completedJobsCount++] = job;
-    if(ctx->base.threadPool) {
+    if(AIO_IOPool_threadPoolActive(&ctx->base)) {
        ZSTD_pthread_cond_signal(&ctx->jobCompletedCond);
-        ZSTD_pthread_mutex_unlock(&ctx->base.ioJobsMutex);
    }
+    AIO_IOPool_unlockJobsMutex(&ctx->base);
 }

 /* AIO_ReadPool_findNextWaitingOffsetCompletedJob_locked:
@@ -426,8 +462,7 @@ static size_t AIO_ReadPool_numReadsInFlight(ReadPoolCtx_t* ctx) {
 * Would block. */
 static IOJob_t* AIO_ReadPool_getNextCompletedJob(ReadPoolCtx_t* ctx) {
    IOJob_t *job = NULL;
-    if (ctx->base.threadPool)
-        ZSTD_pthread_mutex_lock(&ctx->base.ioJobsMutex);
+    AIO_IOPool_lockJobsMutex(&ctx->base);

    job = AIO_ReadPool_findNextWaitingOffsetCompletedJob_locked(ctx);

@@ -443,8 +478,7 @@ static IOJob_t* AIO_ReadPool_getNextCompletedJob(ReadPoolCtx_t* ctx) {
        ctx->waitingOnOffset += job->usedBufferSize;
    }

-    if (ctx->base.threadPool)
-        ZSTD_pthread_mutex_unlock(&ctx->base.ioJobsMutex);
+    AIO_IOPool_unlockJobsMutex(&ctx->base);
    return job;
 }

@@ -524,7 +558,7 @@ ReadPoolCtx_t* AIO_ReadPool_create(const FIO_prefs_t* prefs, size_t bufferSize)

    if(ctx->base.threadPool)
        if (ZSTD_pthread_cond_init(&ctx->jobCompletedCond, NULL))
-            EXM_THROW(103,"Failed creating write jobCompletedCond mutex");
+            EXM_THROW(103,"Failed creating jobCompletedCond cond");

    return ctx;
 }
@@ -620,3 +654,10 @@ int AIO_ReadPool_closeFile(ReadPoolCtx_t* ctx) {
    AIO_ReadPool_setFile(ctx, NULL);
    return fclose(file);
 }
+
+/* AIO_ReadPool_setAsync:
+ * Allows (de)activating async mode, to be used when the expected overhead
+ * of asyncio costs more than the expected gains. */
+void AIO_ReadPool_setAsync(ReadPoolCtx_t* ctx, int async) {
+    AIO_IOPool_setThreaded(&ctx->base, async);
+}
@@ -8,6 +8,17 @@
 * You may select, at your option, one of the above-listed licenses.
 */

+ /*
+  * FileIO AsyncIO exposes read/write IO pools that allow doing IO asynchronously.
+  * Current implementation relies on having one thread that reads and one that
+  * writes.
+  * Each IO pool supports up to `MAX_IO_JOBS` that can be enqueued for work, but
+  * are performed serially by the appropriate worker thread.
+  * Most systems exposes better primitives to perform asynchronous IO, such as
+  * io_uring on newer linux systems. The API is built in such a way that in the
+  * future we could replace the threads with better solutions when available.
+  */
+
 #ifndef ZSTD_FILEIO_ASYNCIO_H
 #define ZSTD_FILEIO_ASYNCIO_H

@@ -27,6 +38,7 @@ extern "C" {
 typedef struct {
    /* These struct fields should be set only on creation and not changed afterwards */
    POOL_ctx* threadPool;
+    int threadPoolActive;
    int totalIoJobs;
    const FIO_prefs_t* prefs;
    POOL_function poolFunction;
@@ -136,6 +148,11 @@ WritePoolCtx_t* AIO_WritePool_create(const FIO_prefs_t* prefs, size_t bufferSize
 * Frees and releases a writePool and its resources. Closes destination file. */
 void AIO_WritePool_free(WritePoolCtx_t* ctx);

+/* AIO_WritePool_setAsync:
+ * Allows (de)activating async mode, to be used when the expected overhead
+ * of asyncio costs more than the expected gains. */
+void AIO_WritePool_setAsync(WritePoolCtx_t* ctx, int async);
+
 /* AIO_ReadPool_create:
 * Allocates and sets and a new readPool including its included jobs.
 * bufferSize should be set to the maximal buffer we want to read at a time, will also be used
@@ -146,6 +163,11 @@ ReadPoolCtx_t* AIO_ReadPool_create(const FIO_prefs_t* prefs, size_t bufferSize);
 * Frees and releases a readPool and its resources. Closes source file. */
 void AIO_ReadPool_free(ReadPoolCtx_t* ctx);

+/* AIO_ReadPool_setAsync:
+ * Allows (de)activating async mode, to be used when the expected overhead
+ * of asyncio costs more than the expected gains. */
+void AIO_ReadPool_setAsync(ReadPoolCtx_t* ctx, int async);
+
 /* AIO_ReadPool_consumeBytes:
 * Consumes byes from srcBuffer's beginning and updates srcBufferLoaded accordingly. */
 void AIO_ReadPool_consumeBytes(ReadPoolCtx_t *ctx, size_t n);
@@ -122,4 +122,4 @@ extern UTIL_time_t g_displayClock;
 #if defined (__cplusplus)
 }
 #endif
-#endif //ZSTD_FILEIO_COMMON_H
+#endif /* ZSTD_FILEIO_COMMON_H */
@@ -69,6 +69,18 @@ typedef struct FIO_prefs_s {
    int contentSize;
    int allowBlockDevices;
    int passThrough;
+    ZSTD_paramSwitch_e mmapDict;
 } FIO_prefs_t;

+typedef enum {FIO_mallocDict, FIO_mmapDict} FIO_dictBufferType_t;
+
+typedef struct {
+    void* dictBuffer;
+    size_t dictBufferSize;
+    FIO_dictBufferType_t dictBufferType;
+#if defined(_MSC_VER) || defined(_WIN32)
+    HANDLE dictHandle;
+#endif
+} FIO_Dict_t;
+
 #endif /* FILEIO_TYPES_HEADER */
@@ -12,7 +12,8 @@
 /* ===  Dependencies  === */

 #include "timefn.h"
-
+#include "platform.h" /* set _POSIX_C_SOURCE */
+#include <time.h>     /* CLOCK_MONOTONIC, TIME_UTC */

 /*-****************************************
 *  Time functions
@@ -20,12 +21,11 @@

 #if defined(_WIN32)   /* Windows */

+#include <windows.h>  /* LARGE_INTEGER */
 #include <stdlib.h>   /* abort */
 #include <stdio.h>    /* perror */

-UTIL_time_t UTIL_getTime(void) { UTIL_time_t x; QueryPerformanceCounter(&x); return x; }
-
-PTime UTIL_getSpanTimeMicro(UTIL_time_t clockStart, UTIL_time_t clockEnd)
+UTIL_time_t UTIL_getTime(void)
 {
    static LARGE_INTEGER ticksPerSecond;
    static int init = 0;
@@ -36,30 +36,20 @@ PTime UTIL_getSpanTimeMicro(UTIL_time_t clockStart, UTIL_time_t clockEnd)
        }
        init = 1;
    }
-    return 1000000ULL*(clockEnd.QuadPart - clockStart.QuadPart)/ticksPerSecond.QuadPart;
-}
-
-PTime UTIL_getSpanTimeNano(UTIL_time_t clockStart, UTIL_time_t clockEnd)
-{
-    static LARGE_INTEGER ticksPerSecond;
-    static int init = 0;
-    if (!init) {
-        if (!QueryPerformanceFrequency(&ticksPerSecond)) {
-            perror("timefn::QueryPerformanceFrequency");
-            abort();
-        }
-        init = 1;
+    {   UTIL_time_t r;
+        LARGE_INTEGER x;
+        QueryPerformanceCounter(&x);
+        r.t = (PTime)(x.QuadPart * 1000000000ULL / ticksPerSecond.QuadPart);
+        return r;
    }
-    return 1000000000ULL*(clockEnd.QuadPart - clockStart.QuadPart)/ticksPerSecond.QuadPart;
 }


-
 #elif defined(__APPLE__) && defined(__MACH__)

-UTIL_time_t UTIL_getTime(void) { return mach_absolute_time(); }
+#include <mach/mach_time.h> /* mach_timebase_info_data_t, mach_timebase_info, mach_absolute_time */

-PTime UTIL_getSpanTimeMicro(UTIL_time_t clockStart, UTIL_time_t clockEnd)
+UTIL_time_t UTIL_getTime(void)
 {
    static mach_timebase_info_data_t rate;
    static int init = 0;
@@ -67,23 +57,39 @@ PTime UTIL_getSpanTimeMicro(UTIL_time_t clockStart, UTIL_time_t clockEnd)
        mach_timebase_info(&rate);
        init = 1;
    }
-    return (((clockEnd - clockStart) * (PTime)rate.numer) / ((PTime)rate.denom))/1000ULL;
+    {   UTIL_time_t r;
+        r.t = mach_absolute_time() * (PTime)rate.numer / (PTime)rate.denom;
+        return r;
+    }
 }

-PTime UTIL_getSpanTimeNano(UTIL_time_t clockStart, UTIL_time_t clockEnd)
+/* POSIX.1-2001 (optional) */
+#elif defined(CLOCK_MONOTONIC)
+
+#include <stdlib.h>   /* abort */
+#include <stdio.h>    /* perror */
+
+UTIL_time_t UTIL_getTime(void)
 {
-    static mach_timebase_info_data_t rate;
-    static int init = 0;
-    if (!init) {
-        mach_timebase_info(&rate);
-        init = 1;
+    /* time must be initialized, othersize it may fail msan test.
+     * No good reason, likely a limitation of timespec_get() for some target */
+    struct timespec time = { 0, 0 };
+    if (clock_gettime(CLOCK_MONOTONIC, &time) != 0) {
+        perror("timefn::clock_gettime(CLOCK_MONOTONIC)");
+        abort();
+    }
+    {   UTIL_time_t r;
+        r.t = (PTime)time.tv_sec * 1000000000ULL + (PTime)time.tv_nsec;
+        return r;
    }
-    return ((clockEnd - clockStart) * (PTime)rate.numer) / ((PTime)rate.denom);
 }


-/* C11 requires timespec_get, but FreeBSD 11 lacks it, while still claiming C11 compliance.
-   Android also lacks it but does define TIME_UTC. */
+/* C11 requires support of timespec_get().
+ * However, FreeBSD 11 claims C11 compliance while lacking timespec_get().
+ * Double confirm timespec_get() support by checking the definition of TIME_UTC.
+ * However, some versions of Android manage to simultaneously define TIME_UTC
+ * and lack timespec_get() support... */
 #elif (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 201112L) /* C11 */) \
    && defined(TIME_UTC) && !defined(__ANDROID__)

@@ -94,65 +100,49 @@ UTIL_time_t UTIL_getTime(void)
 {
    /* time must be initialized, othersize it may fail msan test.
     * No good reason, likely a limitation of timespec_get() for some target */
-    UTIL_time_t time = UTIL_TIME_INITIALIZER;
+    struct timespec time = { 0, 0 };
    if (timespec_get(&time, TIME_UTC) != TIME_UTC) {
-        perror("timefn::timespec_get");
+        perror("timefn::timespec_get(TIME_UTC)");
        abort();
    }
-    return time;
+    {   UTIL_time_t r;
+        r.t = (PTime)time.tv_sec * 1000000000ULL + (PTime)time.tv_nsec;
+        return r;
+    }
 }

-static UTIL_time_t UTIL_getSpanTime(UTIL_time_t begin, UTIL_time_t end)
+
+#else   /* relies on standard C90 (note : clock_t produces wrong measurements for multi-threaded workloads) */
+
+UTIL_time_t UTIL_getTime(void)
 {
-    UTIL_time_t diff;
-    if (end.tv_nsec < begin.tv_nsec) {
-        diff.tv_sec = (end.tv_sec - 1) - begin.tv_sec;
-        diff.tv_nsec = (end.tv_nsec + 1000000000ULL) - begin.tv_nsec;
-    } else {
-        diff.tv_sec = end.tv_sec - begin.tv_sec;
-        diff.tv_nsec = end.tv_nsec - begin.tv_nsec;
-    }
-    return diff;
+    UTIL_time_t r;
+    r.t = (PTime)clock() * 1000000000ULL / CLOCKS_PER_SEC;
+    return r;
+}
+
+#define TIME_MT_MEASUREMENTS_NOT_SUPPORTED
+
+#endif
+
+/* ==== Common functions, valid for all time API ==== */
+
+PTime UTIL_getSpanTimeNano(UTIL_time_t clockStart, UTIL_time_t clockEnd)
+{
+    return clockEnd.t - clockStart.t;
 }

 PTime UTIL_getSpanTimeMicro(UTIL_time_t begin, UTIL_time_t end)
 {
-    UTIL_time_t const diff = UTIL_getSpanTime(begin, end);
-    PTime micro = 0;
-    micro += 1000000ULL * diff.tv_sec;
-    micro += diff.tv_nsec / 1000ULL;
-    return micro;
+    return UTIL_getSpanTimeNano(begin, end) / 1000ULL;
 }

-PTime UTIL_getSpanTimeNano(UTIL_time_t begin, UTIL_time_t end)
-{
-    UTIL_time_t const diff = UTIL_getSpanTime(begin, end);
-    PTime nano = 0;
-    nano += 1000000000ULL * diff.tv_sec;
-    nano += diff.tv_nsec;
-    return nano;
-}
-
-
-
-#else   /* relies on standard C90 (note : clock_t measurements can be wrong when using multi-threading) */
-
-UTIL_time_t UTIL_getTime(void) { return clock(); }
-PTime UTIL_getSpanTimeMicro(UTIL_time_t clockStart, UTIL_time_t clockEnd) { return 1000000ULL * (clockEnd - clockStart) / CLOCKS_PER_SEC; }
-PTime UTIL_getSpanTimeNano(UTIL_time_t clockStart, UTIL_time_t clockEnd) { return 1000000000ULL * (clockEnd - clockStart) / CLOCKS_PER_SEC; }
-
-#endif
-
-
-
-/* returns time span in microseconds */
 PTime UTIL_clockSpanMicro(UTIL_time_t clockStart )
 {
    UTIL_time_t const clockEnd = UTIL_getTime();
    return UTIL_getSpanTimeMicro(clockStart, clockEnd);
 }

-/* returns time span in microseconds */
 PTime UTIL_clockSpanNano(UTIL_time_t clockStart )
 {
    UTIL_time_t const clockEnd = UTIL_getTime();
@@ -167,3 +157,12 @@ void UTIL_waitForNextTick(void)
        clockEnd = UTIL_getTime();
    } while (UTIL_getSpanTimeNano(clockStart, clockEnd) == 0);
 }
+
+int UTIL_support_MT_measurements(void)
+{
+# if defined(TIME_MT_MEASUREMENTS_NOT_SUPPORTED)
+    return 0;
+# else
+    return 1;
+# endif
+}
@@ -16,71 +16,51 @@ extern "C" {
 #endif


-/*-****************************************
-*  Dependencies
-******************************************/
-#include <time.h>         /* clock_t, clock, CLOCKS_PER_SEC */
-
-

 /*-****************************************
-*  Local Types
+*  Types
 ******************************************/

 #if !defined (__VMS) && (defined (__cplusplus) || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */) )
 # if defined(_AIX)
 #  include <inttypes.h>
 # else
-#  include <stdint.h> /* intptr_t */
+#  include <stdint.h> /* uint64_t */
 # endif
  typedef uint64_t           PTime;  /* Precise Time */
 #else
  typedef unsigned long long PTime;  /* does not support compilers without long long support */
 #endif

+/* UTIL_time_t contains a nanosecond time counter.
+ * The absolute value is not meaningful.
+ * It's only valid to compute the difference between 2 measurements. */
+typedef struct { PTime t; } UTIL_time_t;
+#define UTIL_TIME_INITIALIZER { 0 }


 /*-****************************************
 *  Time functions
 ******************************************/
-#if defined(_WIN32)   /* Windows */
-
-    #include <windows.h>   /* LARGE_INTEGER */
-    typedef LARGE_INTEGER UTIL_time_t;
-    #define UTIL_TIME_INITIALIZER { { 0, 0 } }
-
-#elif defined(__APPLE__) && defined(__MACH__)
-
-    #include <mach/mach_time.h>
-    typedef PTime UTIL_time_t;
-    #define UTIL_TIME_INITIALIZER 0
-
-/* C11 requires timespec_get, but FreeBSD 11 lacks it, while still claiming C11 compliance.
-   Android also lacks it but does define TIME_UTC. */
-#elif (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 201112L) /* C11 */) \
-    && defined(TIME_UTC) && !defined(__ANDROID__)
-
-    typedef struct timespec UTIL_time_t;
-    #define UTIL_TIME_INITIALIZER { 0, 0 }
-
-#else   /* relies on standard C90 (note : clock_t measurements can be wrong when using multi-threading) */
-
-    #define UTIL_TIME_USES_C90_CLOCK
-    typedef clock_t UTIL_time_t;
-    #define UTIL_TIME_INITIALIZER 0
-
-#endif
-

 UTIL_time_t UTIL_getTime(void);
-PTime UTIL_getSpanTimeMicro(UTIL_time_t clockStart, UTIL_time_t clockEnd);
-PTime UTIL_getSpanTimeNano(UTIL_time_t clockStart, UTIL_time_t clockEnd);

-#define SEC_TO_MICRO ((PTime)1000000)
-PTime UTIL_clockSpanMicro(UTIL_time_t clockStart);
+/* Timer resolution can be low on some platforms.
+ * To improve accuracy, it's recommended to wait for a new tick
+ * before starting benchmark measurements */
+void UTIL_waitForNextTick(void);
+/* tells if timefn will return correct time measurements
+ * in presence of multi-threaded workload.
+ * note : this is not the case if only C90 clock_t measurements are available */
+int UTIL_support_MT_measurements(void);
+
+PTime UTIL_getSpanTimeNano(UTIL_time_t clockStart, UTIL_time_t clockEnd);
 PTime UTIL_clockSpanNano(UTIL_time_t clockStart);

-void UTIL_waitForNextTick(void);
+PTime UTIL_getSpanTimeMicro(UTIL_time_t clockStart, UTIL_time_t clockEnd);
+PTime UTIL_clockSpanMicro(UTIL_time_t clockStart);
+
+#define SEC_TO_MICRO ((PTime)1000000)  /* nb of microseconds in a second */


 #if defined (__cplusplus)
@@ -102,6 +102,17 @@ UTIL_STATIC void* UTIL_realloc(void *ptr, size_t size)
    #define chmod _chmod
 #endif

+#ifndef ZSTD_HAVE_FCHMOD
+#if PLATFORM_POSIX_VERSION >= 199309L
+#define ZSTD_HAVE_FCHMOD
+#endif
+#endif
+
+#ifndef ZSTD_HAVE_FCHOWN
+#if PLATFORM_POSIX_VERSION >= 200809L
+#define ZSTD_HAVE_FCHOWN
+#endif
+#endif

 /*-****************************************
 *  Console log
@@ -121,7 +132,7 @@ int UTIL_requireUserConfirmation(const char* prompt, const char* abortMsg,
    ch = getchar();
    result = 0;
    if (strchr(acceptableLetters, ch) == NULL) {
-        UTIL_DISPLAY("%s", abortMsg);
+        UTIL_DISPLAY("%s \n", abortMsg);
        result = 1;
    }
    /* flush the rest */
@@ -147,21 +158,38 @@ void UTIL_traceFileStat(void)
    g_traceFileStat = 1;
 }

-int UTIL_stat(const char* filename, stat_t* statbuf)
+int UTIL_fstat(const int fd, const char* filename, stat_t* statbuf)
 {
    int ret;
-    UTIL_TRACE_CALL("UTIL_stat(%s)", filename);
+    UTIL_TRACE_CALL("UTIL_stat(%d, %s)", fd, filename);
 #if defined(_MSC_VER)
-    ret = !_stat64(filename, statbuf);
+    if (fd >= 0) {
+        ret = !_fstat64(fd, statbuf);
+    } else {
+        ret = !_stat64(filename, statbuf);
+    }
 #elif defined(__MINGW32__) && defined (__MSVCRT__)
-    ret = !_stati64(filename, statbuf);
+    if (fd >= 0) {
+        ret = !_fstati64(fd, statbuf);
+    } else {
+        ret = !_stati64(filename, statbuf);
+    }
 #else
-    ret = !stat(filename, statbuf);
+    if (fd >= 0) {
+        ret = !fstat(fd, statbuf);
+    } else {
+        ret = !stat(filename, statbuf);
+    }
 #endif
    UTIL_TRACE_RET(ret);
    return ret;
 }

+int UTIL_stat(const char* filename, stat_t* statbuf)
+{
+    return UTIL_fstat(-1, filename, statbuf);
+}
+
 int UTIL_isRegularFile(const char* infilename)
 {
    stat_t statbuf;
@@ -183,11 +211,16 @@ int UTIL_isRegularFileStat(const stat_t* statbuf)

 /* like chmod, but avoid changing permission of /dev/null */
 int UTIL_chmod(char const* filename, const stat_t* statbuf, mode_t permissions)
+{
+    return UTIL_fchmod(-1, filename, statbuf, permissions);
+}
+
+int UTIL_fchmod(const int fd, char const* filename, const stat_t* statbuf, mode_t permissions)
 {
    stat_t localStatBuf;
-    UTIL_TRACE_CALL("UTIL_chmod(%s, %u)", filename, (unsigned)permissions);
+    UTIL_TRACE_CALL("UTIL_chmod(%s, %#4o)", filename, (unsigned)permissions);
    if (statbuf == NULL) {
-        if (!UTIL_stat(filename, &localStatBuf)) {
+        if (!UTIL_fstat(fd, filename, &localStatBuf)) {
            UTIL_TRACE_RET(0);
            return 0;
        }
@@ -197,9 +230,20 @@ int UTIL_chmod(char const* filename, const stat_t* statbuf, mode_t permissions)
        UTIL_TRACE_RET(0);
        return 0; /* pretend success, but don't change anything */
    }
-    UTIL_TRACE_CALL("chmod");
+#ifdef ZSTD_HAVE_FCHMOD
+    if (fd >= 0) {
+        int ret;
+        UTIL_TRACE_CALL("fchmod");
+        ret = fchmod(fd, permissions);
+        UTIL_TRACE_RET(ret);
+        UTIL_TRACE_RET(ret);
+        return ret;
+    } else
+#endif
    {
-        int const ret = chmod(filename, permissions);
+        int ret;
+        UTIL_TRACE_CALL("chmod");
+        ret = chmod(filename, permissions);
        UTIL_TRACE_RET(ret);
        UTIL_TRACE_RET(ret);
        return ret;
@@ -236,24 +280,52 @@ int UTIL_utime(const char* filename, const stat_t *statbuf)
 }

 int UTIL_setFileStat(const char *filename, const stat_t *statbuf)
+{
+    return UTIL_setFDStat(-1, filename, statbuf);
+}
+
+int UTIL_setFDStat(const int fd, const char *filename, const stat_t *statbuf)
 {
    int res = 0;
    stat_t curStatBuf;
-    UTIL_TRACE_CALL("UTIL_setFileStat(%s)", filename);
+    UTIL_TRACE_CALL("UTIL_setFileStat(%d, %s)", fd, filename);

-    if (!UTIL_stat(filename, &curStatBuf) || !UTIL_isRegularFileStat(&curStatBuf)) {
+    if (!UTIL_fstat(fd, filename, &curStatBuf) || !UTIL_isRegularFileStat(&curStatBuf)) {
        UTIL_TRACE_RET(-1);
        return -1;
    }

-    /* set access and modification times */
-    res += UTIL_utime(filename, statbuf);
+    /* Mimic gzip's behavior:
+     *
+     * "Change the group first, then the permissions, then the owner.
+     * That way, the permissions will be correct on systems that allow
+     * users to give away files, without introducing a security hole.
+     * Security depends on permissions not containing the setuid or
+     * setgid bits." */

 #if !defined(_WIN32)
-    res += chown(filename, statbuf->st_uid, statbuf->st_gid);  /* Copy ownership */
+#ifdef ZSTD_HAVE_FCHOWN
+    if (fd >= 0) {
+        res += fchown(fd, -1, statbuf->st_gid);  /* Apply group ownership */
+    } else
+#endif
+    {
+        res += chown(filename, -1, statbuf->st_gid);  /* Apply group ownership */
+    }
 #endif

-    res += UTIL_chmod(filename, &curStatBuf, statbuf->st_mode & 07777);  /* Copy file permissions */
+    res += UTIL_fchmod(fd, filename, &curStatBuf, statbuf->st_mode & 0777);  /* Copy file permissions */
+
+#if !defined(_WIN32)
+#ifdef ZSTD_HAVE_FCHOWN
+    if (fd >= 0) {
+        res += fchown(fd, statbuf->st_uid, -1);  /* Apply user ownership */
+    } else
+#endif
+    {
+        res += chown(filename, statbuf->st_uid, -1);  /* Apply user ownership */
+    }
+#endif

    errno = 0;
    UTIL_TRACE_RET(-res);
@@ -272,11 +344,15 @@ int UTIL_isDirectory(const char* infilename)

 int UTIL_isDirectoryStat(const stat_t* statbuf)
 {
+    int ret;
+    UTIL_TRACE_CALL("UTIL_isDirectoryStat()");
 #if defined(_MSC_VER)
-    return (statbuf->st_mode & _S_IFDIR) != 0;
+    ret = (statbuf->st_mode & _S_IFDIR) != 0;
 #else
-    return S_ISDIR(statbuf->st_mode) != 0;
+    ret = S_ISDIR(statbuf->st_mode) != 0;
 #endif
+    UTIL_TRACE_RET(ret);
+    return ret;
 }

 int UTIL_compareStr(const void *p1, const void *p2) {
@@ -299,8 +375,32 @@ int UTIL_isSameFile(const char* fName1, const char* fName2)
        stat_t file2Stat;
        ret =  UTIL_stat(fName1, &file1Stat)
            && UTIL_stat(fName2, &file2Stat)
-            && (file1Stat.st_dev == file2Stat.st_dev)
-            && (file1Stat.st_ino == file2Stat.st_ino);
+            && UTIL_isSameFileStat(fName1, fName2, &file1Stat, &file2Stat);
+    }
+#endif
+    UTIL_TRACE_RET(ret);
+    return ret;
+}
+
+int UTIL_isSameFileStat(
+        const char* fName1, const char* fName2,
+        const stat_t* file1Stat, const stat_t* file2Stat)
+{
+    int ret;
+    assert(fName1 != NULL); assert(fName2 != NULL);
+    UTIL_TRACE_CALL("UTIL_isSameFileStat(%s, %s)", fName1, fName2);
+#if defined(_MSC_VER) || defined(_WIN32)
+    /* note : Visual does not support file identification by inode.
+     *        inode does not work on Windows, even with a posix layer, like msys2.
+     *        The following work-around is limited to detecting exact name repetition only,
+     *        aka `filename` is considered different from `subdir/../filename` */
+    (void)file1Stat;
+    (void)file2Stat;
+    ret = !strcmp(fName1, fName2);
+#else
+    {
+        ret =  (file1Stat->st_dev == file2Stat->st_dev)
+            && (file1Stat->st_ino == file2Stat->st_ino);
    }
 #endif
    UTIL_TRACE_RET(ret);
--- a/Show More
+++ b/Show More