compat/zlib/contrib/dotzlib/DotZLib/UnitTests.cs
compat/zlib/contrib/vstudio/readme.txt
compat/zlib/contrib/vstudio/*/zlib.rc
compat/zlib/contrib/vstudio/*/zlibvc.def
compat/zlib/contrib/vstudio/*/*.sln
compat/zlib/win32/zlib.def
compat/zlib/win32/*.txt
compat/zlib/win64/*.txt
compat/zlib/zlib.map
libtommath/*.dsp
libtommath/*.sln
libtommath/*.vcproj
tools/tcl.hpj.in
tools/tcl.wse.in
win/buildall.vc.bat
win/coffbase.txt
win/makefile.vc
win/rules.vc
win/rules-ext.vc
win/targets.vc
win/tcl.dsp
win/tcl.dsw
win/tcl.hpj.in

tools/tcl.hpj.in
tools/tcl.wse.in
win/buildall.vc.bat
win/coffbase.txt
win/makefile.vc
win/rules.vc
win/tcl.dsp
win/tcl.dsw
win/tcl.hpj.in

*/config.status
*/tclConfig.sh
*/tclsh*
*/tcltest*
*/versions.vc
*/version.vc
*/libtcl.vfs
*/libtcl_*.zip
*/libtcl*.zip
*/tclUuid.h
html
libtommath/bn.ilg
libtommath/bn.ind
libtommath/pretty.build
libtommath/tommath.src
libtommath/*.log
libtommath/*.pdf

Important Note
==========
Please do not file issues with Tcl on Github. They are unlikely to be noticed in a timely fashion. Tcl issues are hosted in the [tcl fossil repository on core.tcl.tk](https://core.tcl.tk/tcl/tktnew); please post them there.
Please do not file issues with Tcl on Github. They are unlikely to be noticed in a timely fashion. Tcl issues are hosted in the [tcl fossil repository on core.tcl-lang.org](https://core.tcl-lang.org/tcl/tktnew); please post them there.

Important Note
==========
Please do not file pull requests with Tcl on Github. They are unlikely to be noticed in a timely fashion. Tcl issues (including patches) are hosted in the [tcl fossil repository on core.tcl.tk](https://core.tcl.tk/tcl/tktnew); please post them there.
Please do not file pull requests with Tcl on Github. They are unlikely to be noticed in a timely fashion. Tcl issues (including patches) are hosted in the [tcl fossil repository on core.tcl-lang.org](https://core.tcl-lang.org/tcl/tktnew); please post them there.

version: 2
updates:
- package-ecosystem: "github-actions"
  directory: "/"
  schedule:
      interval: "weekly"

name: Linux
on: [push]
permissions:
  contents: read
jobs:
  gcc:
    runs-on: ubuntu-20.04
    strategy:
      matrix:
        cfgopt:
          - ""
          - "CFLAGS=-DTCL_UTF_MAX=3"
          - "CFLAGS=-DTCL_NO_DEPRECATED=1"
          - "--disable-shared"
          - "--enable-symbols"
          - "--enable-symbols=mem"
          - "--enable-symbols=all"
    defaults:
      run:
        shell: bash
        working-directory: unix
    steps:
      - name: Checkout
        uses: actions/checkout@v3
      - name: Prepare
        run: |
          touch tclStubInit.c tclOOStubInit.c tclOOScript.h
        working-directory: generic
      - name: Configure ${{ matrix.cfgopt }}
        run: |
          mkdir "${HOME}/install dir"
          ./configure ${CFGOPT} "--prefix=$HOME/install dir" || (cat config.log && exit 1)
        env:
          CFGOPT: ${{ matrix.cfgopt }}
      - name: Build
        run: |
          make all
      - name: Build Test Harness
        run: |
          make tcltest
      - name: Run Tests
        run: |
          make test
        env:
          ERROR_ON_FAILURES: 1
      - name: Test-Drive Installation
        run: |
          make install
      - name: Create Distribution Package
        run: |
          make dist
      - name: Convert Documentation to HTML
        run: |
          make html-tcl

name: macOS
on: [push]
permissions:
  contents: read
jobs:
  xcode:
    runs-on: macos-11
    defaults:
      run:
        shell: bash
        working-directory: macosx
    steps:
      - name: Checkout
        uses: actions/checkout@v3
      - name: Prepare
        run: |
          touch tclStubInit.c tclOOStubInit.c tclOOScript.h
        working-directory: generic
      - name: Build
        run: make all
        env:
          CFLAGS: -arch x86_64 -arch arm64e
      - name: Run Tests
        run: make test styles=develop
        env:
          ERROR_ON_FAILURES: 1
          MAC_CI: 1
  clang:
    runs-on: macos-11
    strategy:
      matrix:
        cfgopt:
          - ""
          - "--disable-shared"
          - "--enable-symbols"
          - "--enable-symbols=mem"
          - "--enable-symbols=all"
    defaults:
      run:
        shell: bash
        working-directory: unix
    steps:
      - name: Checkout
        uses: actions/checkout@v3
      - name: Prepare
        run: |
          touch tclStubInit.c tclOOStubInit.c tclOOScript.h
          mkdir "$HOME/install dir"
        working-directory: generic
      - name: Configure ${{ matrix.cfgopt }}
        # Note that macOS is always a 64 bit platform
        run: ./configure --enable-dtrace --enable-framework ${CFGOPT} "--prefix=$HOME/install" || (cat config.log && exit 1)
        env:
          CFLAGS: -arch x86_64 -arch arm64e
          CFGOPT: ${{ matrix.cfgopt }}
      - name: Build
        run: |
          make all tcltest
        env:
          CFLAGS: -arch x86_64 -arch arm64e
      - name: Run Tests
        run: |
          make test
        env:
          ERROR_ON_FAILURES: 1
          MAC_CI: 1

name: Build Binaries
on: [push]
permissions:
  contents: read
jobs:
  linux:
    name: Linux
    runs-on: ubuntu-18.04
    defaults:
      run:
        shell: bash
    steps:
      - name: Checkout
        uses: actions/checkout@v3
      - name: Prepare
        run: |
          touch generic/tclStubInit.c generic/tclOOStubInit.c
          mkdir 1dist
          echo "VER_PATH=$(cd tools; pwd)/addVerToFile.tcl" >> $GITHUB_ENV
        working-directory: .
      - name: Configure
        run: ./configure --disable-symbols --disable-shared --enable-zipfs
        working-directory: unix
      - name: Build
        run: |
          make tclsh
          make shell SCRIPT="$VER_PATH $GITHUB_ENV"
          echo "TCL_ZIP=`pwd`/`echo libtcl*.zip`" >> $GITHUB_ENV
        working-directory: unix
      - name: Package
        run: |
          cp ../unix/tclsh tclsh${TCL_PATCHLEVEL}_snapshot
          chmod +x tclsh${TCL_PATCHLEVEL}_snapshot
          tar -cf tclsh${TCL_PATCHLEVEL}_snapshot.tar tclsh${TCL_PATCHLEVEL}_snapshot
        working-directory: 1dist
      - name: Upload
        uses: actions/upload-artifact@v3
        with:
          name: Tclsh ${{ env.TCL_PATCHLEVEL }} Linux single-file build (snapshot)
          path: 1dist/*.tar
  macos:
    name: macOS
    runs-on: macos-11
    defaults:
      run:
        shell: bash
    steps:
      - name: Checkout
        uses: actions/checkout@v3
      - name: Checkout create-dmg
        uses: actions/checkout@v3
        with:
          repository: create-dmg/create-dmg
          ref: v1.0.8
          path: create-dmg
      - name: Prepare
        run: |
          mkdir 1dist
          touch generic/tclStubInit.c generic/tclOOStubInit.c || true
          wget https://github.com/culler/macher/releases/download/v1.3/macher
          sudo cp macher /usr/local/bin
          sudo chmod a+x /usr/local/bin/macher
          echo "VER_PATH=$(cd tools; pwd)/addVerToFile.tcl" >> $GITHUB_ENV
          echo "CREATE_DMG=$(cd create-dmg;pwd)/create-dmg" >> $GITHUB_ENV
          echo "CFLAGS=-arch x86_64 -arch arm64e" >> $GITHUB_ENV
      - name: Configure
        run: ./configure --disable-symbols --disable-shared --enable-zipfs
        working-directory: unix
      - name: Build
        run: |
          make tclsh
          make shell SCRIPT="$VER_PATH $GITHUB_ENV"
          echo "TCL_BIN=`pwd`/tclsh" >> $GITHUB_ENV
          echo "TCL_ZIP=`pwd`/`echo libtcl*.zip`" >> $GITHUB_ENV
        working-directory: unix
      - name: Package
        run: |
          mkdir contents
          cp $TCL_BIN contents/tclsh${TCL_PATCHLEVEL}_snapshot
          chmod +x contents/tclsh${TCL_PATCHLEVEL}_snapshot
          cat > contents/README.txt <<EOF
          This is a single-file executable developer preview of Tcl $TCL_PATCHLEVEL

          It is not intended as an official release at all, so it is unsigned and unnotarized.
          Use strictly at your own risk.

          To run it, you need to copy the executable out and run:
              xattr -d com.apple.quarantine tclsh${TCL_PATCHLEVEL}_snapshot
          to mark the executable as runnable on your machine.
          EOF
          $CREATE_DMG \
              --volname "Tcl $TCL_PATCHLEVEL (snapshot)" \
              --window-pos 200 120 \
              --window-size 800 400 \
              "Tcl-$TCL_PATCHLEVEL-(snapshot).dmg" \
              "contents/"
        working-directory: 1dist
      - name: Upload
        uses: actions/upload-artifact@v3
        with:
          name: Tclsh ${{ env.TCL_PATCHLEVEL }} macOS single-file build (snapshot)
          path: 1dist/*.dmg
  win:
    name: Windows
    runs-on: windows-2019
    defaults:
      run:
        shell: msys2 {0}
    env:
      CC: gcc
      CFGOPT: --disable-symbols --disable-shared
    steps:
      - name: Install MSYS2
        uses: msys2/setup-msys2@v2
        with:
          msystem: UCRT64
          install: git mingw-w64-ucrt-x86_64-toolchain make zip
      - name: Checkout
        uses: actions/checkout@v3
      - name: Prepare
        run: |
          touch generic/tclStubInit.c generic/tclOOStubInit.c
          echo "VER_PATH=$(cd tools; pwd)/addVerToFile.tcl" >> $GITHUB_ENV
          mkdir 1dist
        working-directory: .
      - name: Configure
        run: ./configure $CFGOPT
        working-directory: win
      - name: Build
        run: |
          make binaries libraries
          echo "TCL_ZIP=`pwd`/`echo libtcl*.zip`" >> $GITHUB_ENV
        working-directory: win
      - name: Get Exact Version
        run: |
          ./tclsh*.exe $VER_PATH $GITHUB_ENV
        working-directory: win
      - name: Set Executable Name
        run: |
          cp ../win/tclsh*.exe tclsh${TCL_PATCHLEVEL}_snapshot.exe
        working-directory: 1dist
      - name: Upload
        uses: actions/upload-artifact@v3
        with:
          name: Tclsh ${{ env.TCL_PATCHLEVEL }} Windows single-file build (snapshot)
          path: '1dist/*_snapshot.exe'

name: Windows
on: [push]
permissions:
  contents: read
env:
  ERROR_ON_FAILURES: 1
jobs:
  msvc:
    runs-on: windows-2022
    defaults:
      run:
        shell: powershell
        working-directory: win
    strategy:
      matrix:
        cfgopt:
          - ""
          - "OPTS=utf16"
          - "CHECKS=nodep"
          - "OPTS=static"
          - "OPTS=symbols"
          - "OPTS=symbols STATS=compdbg,memdbg"
    # Using powershell means we need to explicitly stop on failure
    steps:
      - name: Checkout
        uses: actions/checkout@v3
      - name: Init MSVC
        uses: ilammy/msvc-dev-cmd@v1
      - name: Build ${{ matrix.cfgopt }}
        run: |
          &nmake -f makefile.vc ${{ matrix.cfgopt }} all
          if ($lastexitcode -ne 0) {
             throw "nmake exit code: $lastexitcode"
          }
      - name: Build Test Harness ${{ matrix.cfgopt }}
        run: |
          &nmake -f makefile.vc ${{ matrix.cfgopt }} tcltest
          if ($lastexitcode -ne 0) {
             throw "nmake exit code: $lastexitcode"
          }
      - name: Run Tests ${{ matrix.cfgopt }}
        run: |
          &nmake -f makefile.vc ${{ matrix.cfgopt }} test
          if ($lastexitcode -ne 0) {
             throw "nmake exit code: $lastexitcode"
          }
  gcc:
    runs-on: windows-2022
    defaults:
      run:
        shell: msys2 {0}
        working-directory: win
    strategy:
      matrix:
        cfgopt:
          - ""
          - "CFLAGS=-DTCL_UTF_MAX=3"
          - "CFLAGS=-DTCL_NO_DEPRECATED=1"
          - "--disable-shared"
          - "--enable-symbols"
          - "--enable-symbols=mem"
          - "--enable-symbols=all"
    # Using powershell means we need to explicitly stop on failure
    steps:
      - name: Install MSYS2
        uses: msys2/setup-msys2@v2
        with:
          msystem: MINGW64
          install: git mingw-w64-x86_64-toolchain make
      - name: Checkout
        uses: actions/checkout@v3
      - name: Prepare
        run: |
          touch tclStubInit.c tclOOStubInit.c tclOOScript.h
          mkdir "${HOME}/install dir"
        working-directory: generic
      - name: Configure ${{ matrix.cfgopt }}
        run: |
          ./configure ${CFGOPT} "--prefix=$HOME/install dir" || (cat config.log && exit 1)
        env:
          CFGOPT: --enable-64bit ${{ matrix.cfgopt }}
      - name: Build
        run: make all
      - name: Build Test Harness
        run: make tcltest
      - name: Run Tests
        run: make test

# If you add builds with Wine, be sure to define the environment variable
# CI_USING_WINE when running them so that broken tests know not to run.

_FOSSIL_
*/tclConfig.sh
*/tclsh*
*/tcltest*
*/versions.vc
*/version.vc
*/libtcl.vfs
*/libtcl_*.zip
*/libtcl*.zip
*/tclUuid.h
libtommath/bn.ilg
libtommath/bn.ind
libtommath/pretty.build
libtommath/tommath.src
libtommath/*.log
libtommath/*.pdf
libtommath/*.pl

    - name: "Linux/GCC/Mem-Debug"
      os: linux
      dist: focal
      compiler: gcc
      env:
        - BUILD_DIR=unix
        - CFGOPT="--enable-symbols=mem"
# C++ build.
    - name: "Linux/G++/Shared"
      os: linux
      dist: focal
      compiler: g++
      env:
        - BUILD_DIR=unix
        - CFGOPT="CC=g++ CFLAGS=-Dregister=dont+use+register"
# Newer/Older versions of GCC
    - name: "Linux/GCC 10/Shared"
      os: linux
      dist: focal
      compiler: gcc-10
      addons:
        apt:

        - BUILD_DIR=macosx
      install: []
      script: *mactest
      addons:
        homebrew:
          packages:
            - libtommath
    - name: "macOS/Clang++/Xcode 12/Shared"
      os: osx
      osx_image: xcode12.2
      env:
        - BUILD_DIR=unix
        - CFGOPT="CC=clang++ --enable-framework --enable-dtrace CFLAGS=-Dregister=dont+use+register CPPFLAGS=-D__private_extern__=extern"
      script:
        - make all tcltest
# Newer MacOS versions
    - name: "macOS/Clang/Xcode 12/Universal Apps/Shared"
      os: osx
      osx_image: xcode12u
      env:
        - BUILD_DIR=macosx
      install: []

    - name: "Windows/MSVC/Shared"
      os: windows
      compiler: cl
      env: &vcenv
        - BUILD_DIR=win
        - VCDIR="/C/Program Files (x86)/Microsoft Visual Studio/2017/BuildTools/VC/Auxiliary/Build"
      before_install: &vcpreinst
        - rm -rf tests/safe-stock8*.test
        - touch generic/tclStubInit.c generic/tclOOStubInit.c generic/tclOOScript.h
        - PATH="$PATH:$VCDIR"
        - cd ${BUILD_DIR}
      install: []
      script:
        - cmd.exe //C vcvarsall.bat x64 '&&' nmake '-f' makefile.vc all tcltest
        - cmd.exe //C vcvarsall.bat x64 '&&' nmake '-f' makefile.vc test
    - name: "Windows/MSVC/Shared: NO_DEPRECATED"
      os: windows
      compiler: cl
      env: *vcenv
      before_install: *vcpreinst
      install: []
      script:
        - cmd.exe //C vcvarsall.bat x64 '&&' nmake 'OPTS=nodep' '-f' makefile.vc all tcltest
        - cmd.exe //C vcvarsall.bat x64 '&&' nmake 'OPTS=nodep' '-f' makefile.vc test
        - cmd.exe //C vcvarsall.bat x64 '&&' nmake 'CHECKS=nodep' '-f' makefile.vc all tcltest
        - cmd.exe //C vcvarsall.bat x64 '&&' nmake 'CHECKS=nodep' '-f' makefile.vc test
    - name: "Windows/MSVC/Static"
      os: windows
      compiler: cl
      env: *vcenv
      before_install: *vcpreinst
      install: []
      script:
        - cmd.exe //C vcvarsall.bat x64 '&&' nmake 'OPTS=static,msvcrt' '-f' makefile.vc all tcltest
        - cmd.exe //C vcvarsall.bat x64 '&&' nmake 'OPTS=static,msvcrt' '-f' makefile.vc test
        - cmd.exe //C vcvarsall.bat x64 '&&' nmake 'OPTS=static' '-f' makefile.vc all tcltest
        - cmd.exe //C vcvarsall.bat x64 '&&' nmake 'OPTS=static' '-f' makefile.vc test
    - name: "Windows/MSVC/Debug"
      os: windows
      compiler: cl
      env: *vcenv
      before_install: *vcpreinst
      install: []
      script:

    - name: "Windows/MSVC-x86/Shared: NO_DEPRECATED"
      os: windows
      compiler: cl
      env: *vcenv
      before_install: *vcpreinst
      install: []
      script:
        - cmd.exe //C vcvarsall.bat x86 '&&' nmake 'OPTS=nodep' '-f' makefile.vc all tcltest
        - cmd.exe //C vcvarsall.bat x86 '&&' nmake 'OPTS=nodep' '-f' makefile.vc test
        - cmd.exe //C vcvarsall.bat x86 '&&' nmake 'CHECKS=nodep' '-f' makefile.vc all tcltest
        - cmd.exe //C vcvarsall.bat x86 '&&' nmake 'CHECKS=nodep' '-f' makefile.vc test
    - name: "Windows/MSVC-x86/Static"
      os: windows
      compiler: cl
      env: *vcenv
      before_install: *vcpreinst
      install: []
      script:
        - cmd.exe //C vcvarsall.bat x86 '&&' nmake 'OPTS=static,msvcrt' '-f' makefile.vc all tcltest
        - cmd.exe //C vcvarsall.bat x86 '&&' nmake 'OPTS=static,msvcrt' '-f' makefile.vc test
        - cmd.exe //C vcvarsall.bat x86 '&&' nmake 'OPTS=static' '-f' makefile.vc all tcltest
        - cmd.exe //C vcvarsall.bat x86 '&&' nmake 'OPTS=static' '-f' makefile.vc test
    - name: "Windows/MSVC-x86/Debug"
      os: windows
      compiler: cl
      env: *vcenv
      before_install: *vcpreinst
      install: []
      script:

    - name: "Windows/GCC/Shared"
      os: windows
      compiler: gcc
      env:
        - BUILD_DIR=win
        - CFGOPT="--enable-64bit"
      before_install: &makepreinst
        - rm -rf tests/safe-stock8*.test
        - touch generic/tclStubInit.c generic/tclOOStubInit.c generic/tclOOScript.h
        - choco install -y make zip
        - cd ${BUILD_DIR}
    - name: "Windows/GCC/Shared: UTF_MAX=3"
      os: windows
      compiler: gcc
      env:
        - BUILD_DIR=win
        - CFGOPT="--enable-64bit CFLAGS=-DTCL_UTF_MAX=3"
      before_install: *makepreinst
    - name: "Windows/GCC/Shared: NO_DEPRECATED"
      os: windows
      compiler: gcc
      env:
        - BUILD_DIR=win
        - CFGOPT="--enable-64bit CFLAGS=-DTCL_NO_DEPRECATED=1"
      before_install: *makepreinst
    - name: "Windows/G++/Shared"
      os: windows
      compiler: g++
      env:
        - BUILD_DIR=win
        - CFGOPT="CC=g++ --enable-64bit"
      before_install: *makepreinst
      script:
        - make all tcltest
    - name: "Windows/GCC/Static"
      os: windows
      compiler: gcc
      env:
        - BUILD_DIR=win
        - CFGOPT="--enable-64bit --disable-shared"
      before_install: *makepreinst

    - name: "Windows/GCC-x86/Shared: NO_DEPRECATED"
      os: windows
      compiler: gcc
      env:
        - BUILD_DIR=win
        - CFGOPT="CFLAGS=-DTCL_NO_DEPRECATED=1"
      before_install: *makepreinst
    - name: "Windows/G++-x86/Shared"
      os: windows
      compiler: g++
      env:
        - BUILD_DIR=win
        - CFGOPT="CC=g++"
      before_install: *makepreinst
      script:
        - make all tcltest
    - name: "Windows/GCC-x86/Static"
      os: windows
      compiler: gcc
      env:
        - BUILD_DIR=win
        - CFGOPT="--disable-shared"
      before_install: *makepreinst

      dist: focal
      compiler: gcc
      env:
        - BUILD_DIR=unix
      script:
        - make dist
before_install:
  - rm -rf tests/safe-stock8*.test
  - touch generic/tclStubInit.c generic/tclOOStubInit.c generic/tclOOScript.h
  - cd ${BUILD_DIR}
install:
  - mkdir "$HOME/install dir"
  - ./configure ${CFGOPT} "--prefix=$HOME/install dir" || (cat config.log && exit 1)
before_script:
  - export ERROR_ON_FAILURES=1
script:
  - make all tcltest || echo "Something wrong, maybe a hickup, let's try again"
  - make test
  - make install

	* tests/fileName.test:	was computing the wrong results for both [file
	dirname] and [file tail] on "path" arguments with the PATHFLAGS != 0
	intrep and with an empty string for the "joined-on" part.

2009-03-25  Jan Nijtmans  <nijtmans@users.sf.net>

	* doc/tclsh.1:		 Bring doc and tools in line with
	* tools/installData.tcl: http://wiki.tcl.tk/812
	* tools/installData.tcl: https://wiki.tcl-lang.org/page/exec+magic
	* tools/str2c
	* tools/tcltk-man2html.tcl

2009-03-25  Donal K. Fellows  <dkf@users.sf.net>

	* doc/coroutine.n: [Bug 2152285]: Added basic documentation for the
	coroutine and yield commands.

	* tests/encoding.test:	Tcl(Get|Set)EncodingSearchPath.	 Updated tests.

2004-11-30  Kevin B. Kenny  <kennykb@acm.org>

	* library/clock.tcl: Corrected the regular expressions that match a
	time zone to allow for time zones specified as +HH or -HH.
	* tests/clock.test: Added regression test case for the above issue.
	Thanks to Rolf Ade for reporting this issue [http://wiki.tcl.tk/13094]
	Thanks to Rolf Ade for reporting this issue [https://wiki.tcl-lang.org/page/Parsing+ISO8601+dates+and+times]
	* win/tclWinDde.c (Tcl_DdeObjCmd): Corrected a typo that caused a
	compilation failure on VC++.

2004-11-29  Andreas Kupries <andreask@activestate.com>

	* win/Makefile.in (install-libraries): Brought entry '2004-10-26 Don
	Porter (Tcl Modules)' into the windows world, actually the

	* tests/scan.test:	decimal formatted integers. Fixed to match.

2006-04-19  Kevin B. Kenny  <kennykb@acm.org>

	* generic/tclStrToD.c: Added code to support the "middle endian"
	floating point format used in the Nokia N770's software-based floating
	point. Thanks to Bruce Johnson for reporting this bug, originally on
	http://wiki.tcl.tk/15408.
	https://wiki.tcl-lang.org/page/Nokia+770.
	* library/clock.tcl: Fixed a bug with Daylight Saving Time and Posix
	time zone specifiers reported by Martin Lemburg in
	http://groups.google.com/group/comp.lang.tcl/browse_thread/thread/9a8b15a4dfc0b7a0
	(and not at SourceForge).
	* tests/clock.test: Added test case for the above bug.

2006-04-18  Donal K. Fellows  <dkf@users.sf.net>

# README:  Tcl

This is the **Tcl 9.0a2** source distribution.
This is the **Tcl 9.0a4** source distribution.

You can get any source release of Tcl from [our distribution
site](https://sourceforge.net/projects/tcl/files/Tcl/).

8.6 (production release, daily build)
[![Build Status](https://github.com/tcltk/tcl/workflows/Linux/badge.svg?branch=core-8-6-branch)](https://github.com/tcltk/tcl/actions?query=workflow%3A%22Linux%22+branch%3Acore-8-6-branch)
[![Build Status](https://github.com/tcltk/tcl/workflows/Windows/badge.svg?branch=core-8-6-branch)](https://github.com/tcltk/tcl/actions?query=workflow%3A%22Windows%22+branch%3Acore-8-6-branch)
[![Build Status](https://github.com/tcltk/tcl/workflows/macOS/badge.svg?branch=core-8-6-branch)](https://github.com/tcltk/tcl/actions?query=workflow%3A%22macOS%22+branch%3Acore-8-6-branch)
<br>
8.7 (in development, daily build))
[![Build Status](https://github.com/tcltk/tcl/workflows/Linux/badge.svg?branch=core-8-branch)](https://github.com/tcltk/tcl/actions?query=workflow%3A%22Linux%22+branch%3Acore-8-branch)
[![Build Status](https://github.com/tcltk/tcl/workflows/Windows/badge.svg?branch=core-8-branch)](https://github.com/tcltk/tcl/actions?query=workflow%3A%22Windows%22+branch%3Acore-8-branch)
[![Build Status](https://github.com/tcltk/tcl/workflows/macOS/badge.svg?branch=core-8-branch)](https://github.com/tcltk/tcl/actions?query=workflow%3A%22macOS%22+branch%3Acore-8-branch)
<br>
9.0 (in development, daily build))
[![Build Status](https://travis-ci.org/tcltk/tcl.svg?branch=master)](https://travis-ci.org/tcltk/tcl)
[![Build Status](https://github.com/tcltk/tcl/workflows/Linux/badge.svg?branch=main)](https://github.com/tcltk/tcl/actions?query=workflow%3A%22Linux%22+branch%3Amain)
[![Build Status](https://github.com/tcltk/tcl/workflows/Windows/badge.svg?branch=main)](https://github.com/tcltk/tcl/actions?query=workflow%3A%22Windows%22+branch%3Amain)
[![Build Status](https://github.com/tcltk/tcl/workflows/macOS/badge.svg?branch=main)](https://github.com/tcltk/tcl/actions?query=workflow%3A%22macOS%22+branch%3Amain)

## Contents
 1. [Introduction](#intro)
 2. [Documentation](#doc)
 3. [Compiling and installing Tcl](#build)
 4. [Development tools](#devtools)
 5. [Tcl newsgroup](#complangtcl)

When paired with the Tk toolkit, Tcl provides the fastest and most powerful
way to create GUI applications that run on PCs, Unix, and Mac OS X.
Tcl can also be used for a variety of web-related tasks and for creating
powerful command languages for applications.

Tcl is maintained, enhanced, and distributed freely by the Tcl community.
Source code development and tracking of bug reports and feature requests
takes place at [core.tcl-lang.org](https://core.tcl-lang.org/).
take place at [core.tcl-lang.org](https://core.tcl-lang.org/).
Tcl/Tk release and mailing list services are [hosted by
SourceForge](https://sourceforge.net/projects/tcl/)
with the Tcl Developer Xchange hosted at
[www.tcl-lang.org](https://www.tcl-lang.org).

Tcl is a freely available open source package.  You can do virtually
Tcl is a freely available open-source package.  You can do virtually
anything you like with it, such as modifying it, redistributing it,
and selling it either in whole or in part.  See the file
`license.terms` for complete information.

## <a id="doc">2.</a> Documentation
Extensive documentation is available at our website.
Extensive documentation is available on our website.
The home page for this release, including new features, is
[here](https://www.tcl.tk/software/tcltk/9.0.html).
[here](https://www.tcl-lang.org/software/tcltk/9.0.html).
Detailed release notes can be found at the
[file distributions page](https://sourceforge.net/projects/tcl/files/Tcl/)
by clicking on the relevant version.

Information about Tcl itself can be found at the [Developer
Xchange](https://www.tcl-lang.org/about/).
There have been many Tcl books on the market.  Many are mentioned in

## <a id="build">3.</a> Compiling and installing Tcl
There are brief notes in the `unix/README`, `win/README`, and `macosx/README`
about compiling on these different platforms.  There is additional information
about building Tcl from sources
[online](https://www.tcl-lang.org/doc/howto/compile.html).

## <a id="devtools">4.</a> Development tools
ActiveState produces a high quality set of commercial quality development
ActiveState produces a high-quality set of commercial quality development
tools that is available to accelerate your Tcl application development.
Tcl Dev Kit builds on the earlier TclPro toolset and provides a debugger,
static code checker, single-file wrapping utility, bytecode compiler and
static code checker, single-file wrapping utility, bytecode compiler, and
more.  More information can be found at

	http://www.ActiveState.com/Tcl
	https://www.activestate.com/products/tcl/

## <a id="complangtcl">5.</a> Tcl newsgroup
There is a USENET news group, "`comp.lang.tcl`", intended for the exchange of
There is a USENET newsgroup, "`comp.lang.tcl`", intended for the exchange of
information about Tcl, Tk, and related applications.  The newsgroup is a
great place to ask general information questions.  For bug reports, please
see the "Support and bug fixes" section below.

## <a id="wiki">6.</a> Tcl'ers Wiki
There is a [wiki-based open community site](https://wiki.tcl-lang.org/)
covering all aspects of Tcl/Tk.

2017-06-23 (TIP 472) Support 0d as prefix of decimal numbers (iyer,griffin)

2017-08-31 (bug)[2a9465] http state 100 continue handling broken (oehlmann)

2017-09-02 (bug)[0e4d88] replace command, delete trace kills namespace (porter)

--- Released 8.7a1, September 8, 2017 --- http://core.tcl.tk/tcl/ for details
--- Released 8.7a1, September 8, 2017 --- https://core.tcl-lang.org/tcl/ for details

2017-08-10 [array names -regexp] supports backrefs (goth)

2017-08-10 Fix gcc build failures due to #pragma placement (cassoff,fellows)

2017-08-29 (bug)[b50fb2] exec redir append stdout and stderr to file (coulter)

2018-10-29 Update tcltest package for Travis support (fellows)
=> tcltest 2.5.0

2018-11-09 (bug)[35a8f1] overlong string length of some lists (owens)

2018-11-16 (bug)[00d04c] Repair [binary encode base64] (sebres)

- Released 8.6.9, November 16, 2018 - details at http://core.tcl-lang.org/tcl/ -
- Released 8.6.9, November 16, 2018 - details at https://core.tcl-lang.org/tcl/ -

2018-11-22 (bug)[7a9dc5] [file normalize ~/~foo] segfault (sebres)

2018-12-30 (bug)[3cf3a9] variable 'timezone' deprecated in vc2017 (nijtmans)

2019-01-09 (bug)[cc1e91] [list [list {*}[set a " "]]] regression (sebres)

2019-10-25 OSX: system Tcl deprecated. End default use of its packages. (walzer)

2019-10-28 (bug)[bcd100] bad fs cache when system encoding changes (coulter)

2019-11-15 (bug)[135804] segfault in [next] after destroy (coulter,sebres)

2019-11-18 (bug)[13657a] application/json us text, not binary (noe,nijtmans)
=> http 2.9.1

- Released 8.6.10, Nov 21, 2019 - details at http://core.tcl-lang.org/tcl/ -
- Released 8.6.10, Nov 21, 2019 - details at https://core.tcl-lang.org/tcl/ -

Changes to 8.7a3 include all changes to the 8.6 line through 8.6.10,
plus the following, which focuses on the high-level feature changes
in this changeset (new minor version) rather than bug fixes:

2017-11-01 (bug)[3c32a3] crash deleting class mixed into instance (coulter)

2019-04-14 [TIP 367] [lremove]

2019-04-14 [TIP 504] [string insert]

2019-04-16 [TIP 342] [dict getwithdefault]

2019-04-23 (bug)[67a5ea] make [chan postevent] asynchronous
        *** POTENTIAL INCOMPATIBILITY ***

2019-05-25 [TIP 431] [file tempdir]

2019-05-25 [TIP 383] [coroinject], [coroprobe]

2019-05-31 [TIP 544] Tcl_GetIntForIndex()

2019-06-12 Replace TclOffset() with offsetof()

2019-06-15 [TIP 461] string compare operators for [expr]

2019-06-16 [TIP 521] floating point classification functions for [expr]

2019-06-20 (bug)[6bdadf] crash multi-arg traced [lappend] (fellows)

2019-06-28 [TIP 547] New encodings utf-16, ucs-2

2019-09-14 [TIP 414] Tcl_InitSubsystems()

2019-09-14 [TIP 548] wchar_t conversion functions

- Released 8.7a3, Nov 21, 2019 --- http://core.tcl-lang.org/tcl/ for details -
- Released 8.7a3, Nov 21, 2019 --- https://core.tcl-lang.org/tcl/ for details -

Changes to 9.0a1 include all changes to the 8.7 line through 8.7a3,
plus the following, which focuses on the high-level feature changes
in this changeset (new minor version) rather than bug fixes:

2017-11-03 [TIP 114] Leading zero integer no longer means octal

2017-11-03 [TIP 278] Revise variable name resolution, solve "Creative Writing"

2017-11-03 [TIPs 330,336] Encapsulate struct Tcl_Interp

2017-11-17 [TIP 422] Remove all Tcl_*VA() routines

2017-12-15 [TIP 488] Disable magic $::tcl_precision

2018-10-08 [TIP 494] Increased support for size_t value ranges

2019-05-31 [TIP 537] 64-bit indices in regexp matching

- Released 9.0a1, Nov 25, 2019 --- http://core.tcl-lang.org/tcl/ for details -
- Released 9.0a1, Nov 25, 2019 --- https://core.tcl-lang.org/tcl/ for details -

2019-12-03 (bug)[3cd9be] Corner case in surrogate handling (nijtmans)

2019-12-09 (new) Add tcltest::(Setup|Eval|Cleanup|)Test (coulter,sebres)
=> tcltest 2.5.2

2019-12-12 (new) Add 3 libtommath functions to stub table (nijtmans)

2019-12-23 (bug)[ce3b9f] compilation errors with clang, windows msys2 (nijtmans)

2019-12-27 (bug)[1de6b0] [expr 1e2147483648] => 0.0 (kbk)

2020-01-04 (bug)[912886] tis-620 encoding fails to load (coulter)

2020-01-13 (bug)[0b9332] Win: support system encoding init to utf-8 (jedli�ka)

2020-01-17 (bug)[8cd2fe] [unload] corrupted list of loaded packages (berc)

2020-01-17 (bug)[5d989f] segfault in lsort for large list length (sebres)

2020-01-30 (bug) Reset WSAGetLastError()/errno in channel close (nijtmans)

2020-02-17 (bug) Win: avoid create of legacy error-vars on init phase (sebres)

2020-02-25 (bug) release refs when setting class's superclasses fails (dkf)

2020-02-26 (bug) C++ compiler compatibility for registry and dde (nijtmans)
=> registry 1.3.5
=> dde 1.4.3

2020-03-05 (new) Update to Unicode-13 (nijtmans)

2020-03-16 (bug)[8f89e2] Win: env var encoding, env-2.5 (sebres, nijtmans)

2020-03-27 (bug)[767e07] Tcl_Get(Range|UniChar) validate index inputs (nijtmans)

2020-03-28 (bug)[8edfce] [binary encode base64] & multi-byte wrapchars (dgp)

2020-03-28 (bug)[ffeb20] [binary decode base64] ignore invalid chars (dgp)
See RFC 2045
        *** POTENTIAL INCOMPATIBILITY ***

2020-03-31 (bug)[b8e82d] some -maxlen values break uuencode round trip (dgp)
        *** POTENTIAL INCOMPATIBILITY ***

2020-04-01 (bug)[f58371] Fileevent run in proper thread (bron,sebres)

2020-04-13 (bug)[afa4b2] TclNeedSpace bug; tests util-8.5 .. util-8.11 (dgp)

2020-04-13 (bug)[085913] Tcl_DStringAppendElement # quoting precision (dgp)
        *** POTENTIAL INCOMPATIBILITY ***

2020-04-13 (bug)[a7f685] test util-5.52 (dgp)

2020-04-13 (bug)[c61818] Tcl_UtfPrev regression (dgp)

2020-04-15 (bug)[8af92d] zlib transform issue, bad inflate (sebres)

2020-04-16 (bug)[5e6346] Tcl_UtfPrev handling of overlong sequences (dgp)

2020-04-27 (bug)[45ca23] [string tolower] inconsistency (dgp)

2020-04-30 (bug)[da2352] init [info hostname] with DNS, not NetBIOS (nadkarni)

2020-05-11 (bug)[d402ff] Win32 potential crash when using main() (werner)

2020-05-13 (bug)[81242a] revised documentation for Tcl_UtfAtIndex() (nijtmans)
        *** POTENTIAL INCOMPATIBILITY ***

2020-05-13 (bug)[ed2980] Tcl_UtfToUniChar reads > TCL_UTF_MAX bytes (nijtmans)
        *** POTENTIAL INCOMPATIBILITY ***

2020-06-02 (bug) prevent segfault in parser (sebres)

2020-06-21 (bug)[f81bec] http POST a binary file (alakendu,nash)
=> http 2.9.2

2020-06-23 (bug)[41c985] auto_path nonsense in Safe Base (nash)

2020-06-24 (bug)[f70ce1] zlib multi-stream inflate acts only on first (sebres)

2020-07-09 (bug)[a1bd37] [clock scan] new ISO format (clock-34.(19-24)) (sebres)
        *** POTENTIAL INCOMPATIBILITY ***

2020-07-10 (bug)[501974] [clock scan] +time zone (clock-34.(53-68)) (sebres)
        *** POTENTIAL INCOMPATIBILITY ***

2020-07-15 (bug)[3c6e47] compiled [lappend] performance, avoid copy (sebres)

2020-07-16 (bug)[5bbd04] Fix index underflow (schwab)

2020-07-27 (bug)[cb0373] http::geturl -keepalive fixes (nash)
=> http 2.9.3

2020-08-10 (bug)[29e884] cmd resolution cycle (namespace-57.0) (coulter,sebres)

2020-08-12 (bug)[e87000] Tcl_BadChannelOption tolerate NULL (werner,nijtmans)

2020-08-31 (TIP #581) disfavor Master/Slave terminology (nijtmans)
=> opt 0.4.8

2020-09-11 (bug)[3bc0f4] UBSan complains about body.chars[] usage (nijtmans)

2020-09-17 (bug)[835c93] Support TIP 525 exit code for -singleproc 1 (nijtmans)
=> tcltest 2.5.3

2020-09-25 (new) force -eofchar \032 when evaluating library scripts (nijtmans)
        *** POTENTIAL INCOMPATIBILITY ***

2020-09-29 (bug)[0063cb] http::geturl -headers must be dict (oehlmann,nijtmans)

2020-10-19 (bug)[cb4582] Update install-sh script (stu,nijtmans)

2020-10-22 (bug)[c97593] Usage of gnu_printf in latest mingw-w64 (nijtmans)

2020-10-26 (new)[48898a] improve error message consistency (stu)
        *** POTENTIAL INCOMPATIBILITY ***

2020-11-06 (new) revised case of module names (nijtmans)
        *** POTENTIAL INCOMPATIBILITY ***

2020-12-10 (bug)[ed5be7] Win: recognize "comx:" as serial port (oehlmann)

2020-12-11 (new) support for msys2, Big Sur (nijtmans)
=> platform 1.0.15

2020-12-23 tzdata updated to Olson's tzdata2020e (jima)

- Released 8.6.11, Dec 31, 2020 - details at https://core.tcl-lang.org/tcl/ -

Changes to 8.7a5 include all changes to the 8.6 line through 8.6.11,
plus the following, which focuses on the high-level feature changes
in this changeset (new minor version) rather than bug fixes:

2019-12-13 [TIP 538] Externalize libtommath

2020-01-20 [TIP 542] Support for switchable Full Unicode support

2020-01-21 [TIP 543] Eliminate `TCL_INTERP_DESTROYED` flag value

2020-01-24 [TIP 559] Eliminate public routine `Tcl_FreeResult

2020-01-31 (new) Implement 64-bit seek on Zip channels. (nijtmans)

2020-02-28 [TIP 557] C++ support for Tcl

2020-02-28 [TIP 562] Deprecate channel types 1-4

2020-03-11 (bug)[234d6c] Segfault in [set l {}; lpop l] (sebres)

2020-03-12 (bug) Crash in tests binary-79.[12] (porter)

2020-03-13 [TIP 569] Eliminate Comments That Serve Lint

2020-04-06 (bug)[dd010c] [string trim*] on astral characters (porter,nijtmans)

2020-05-30 [TIP 551] Permit underscore in numerical literals in source code

2020-07-03 [TIP 578] Death to TCL_DBGX

2020-08-11 (bug)[e87000] Win32 crash in [fconfigure stdout] (werner,nijtmans)

2020-09-06 (bug)[c1a376] deletion trace on imported ensemble (coulter)

2020-09-13 [TIP 585] Promote the INDEX_TEMP_TABLE flag of Tcl_GetIndexFromObj*() to the public interface

2020-09-15 (bug)[b5777d] crash in [string index abcd 0-0x10000000000000000]

2020-09-19 [b9ecf3] revised stork mgmt [uplevel [list $cmd ...]] (coulter)

2020-10-23 [TIP 587] Default utf-8 for source command

2020-10-27 (bug)[11229b] test string-31.26.* (porter)

2020-11-08 [TIP 582] Comments in Expressions

2020-11-16 [TIP 586] C String Parsing Support for binary scan

2020-12-07 [TIP 590] Recommend lowercase Package Names

2021-01-06 Bump to tcltest 2.5.4

2021-01-15 [TIP 481] `Tcl_GetStringFromObj()` with `size_t` length parameter

2021-01-15 [TIP 592] End support: Windows XP, Server 2003, Vista, Server 2008

2021-01-25 tzdata updated to Olson's tzdata2021a (nijtmans)

2021-01-29 (bug)[113be1] zipfs on mac

2021-03-15 [TIP 575] Switchable Tcl_UtfCharComplete()/Tcl_UtfNext()/Tcl_UtfPrev()

2021-03-19 (new)[0221b9] Drop TCL_WINDOW_EVENTS from Tcl's [update idletasks]

2021-03-30 (new)[4b4830] [chan truncate] for reflected channels

2021-04-30 [TIP 597] "string is unicode" and better utf-8/utf-16/cesu-8 encodings

2021-04-09 [TIP 598] export TclWinConvertError

2021-05-15 (bug)[463b7a] segfault from Tcl_Unload (coulter)

2021-05-15 (bug)[fb2a41] tclZipfs.c free all memory (coulter)

2021-05-18 (bug)[688fcc,28027d] namespace teardown reform (coulter)

- Released 8.7a5, Jun 18, 2021 --- https://core.tcl-lang.org/tcl/ for details -

Changes to 9.0a3 include all changes to the 8.7 line through 8.7a5,
plus the following, which focuses on the high-level feature changes
in this changeset (new major version) rather than bug fixes:

Many of the TIPs in Tcl 8.7 mentioned above are extended further in 9.0

2020-02-28 [TIP 497] Full support for Unicode planes 1-16

2020-08-21 (bug)[43b434] improper calls to stat64()

2021-04-08 [TIP 595] Unicode-aware loadable library handling.

2021-04-30 [TIP 596] Stubs support for embedding Tcl in apps

Many internal changes to broaden support for sizes beyond 32-bits.

- Released 9.0a3, Jun 23, 2021 --- https://core.tcl-lang.org/tcl/ for details -

2021-02-02 (new) support for MacOS Big Sur updates (nijtmans)
=> platform 1.0.17

2021-02-15 (bug)[d43f96] [string trim*] broken for Emoji (werner)

2021-02-16 (bug)[22324b] [string reverse] broken for Emoji (werner)

2021-02-19 (bug)[1dab71,7c64aa] BRE broken by uninitialized value use (lane)

2021-03-09 (bug)[8419c5] Unix tty channels tolerate EINTR (nijtmans)
        *** POTENTIAL INCOMPATIBILITY ***

2021-03-10 (bug)[4c591f] [string compare] EIAS violation (nijtmans)

2021-04-08 (new) dde package installation compatible with Tcl 9 (nijtmans)
=> dde 1.4.4

2021-04-14 (bug)[266494] [concat foo [list #]] EIAS violation (porter)

2021-05-03 (bug)[24b918] Save IO buffers from modern optimizers (rupprecht)

2021-05-06 (new) support for POSIX error EILSEQ (nijtmans)

2021-05-17 (bug)[688fcc] segfault during traced delete of alias (coulter)

2021-06-22 (bug)[bad6cc] More secure build tool. CVE-2021-35331 (nijtmans)

2021-07-17 (bug)[592a25] Win: segfault in Tcl_PutEnv() (danckaert,nijtmans)

2021-09-02 (bug)[ccc448] segfault in ensemble rewrite machinery (coulter)

2021-09-14 (new) Update to Unicode-14 (nijtmans)

2021-10-08 (bug)[a8579d] failed proc argument spec processing (russell,coulter)

2021-10-27 (new) support for MacOS Monterey (nijtmans)
=> platform 1.0.18

2021-10-27 tzdata updated to Olson's tzdata2021e (nijtmans)

- Released 8.6.12, Nov 5, 2021 - details at https://core.tcl-lang.org/tcl/ -

cmake_minimum_required(VERSION 2.4.4)
set(CMAKE_ALLOW_LOOSE_LOOP_CONSTRUCTS ON)

project(zlib C)

set(VERSION "1.2.11")
set(VERSION "1.2.12")

option(ASM686 "Enable building i686 assembly implementation")
option(AMD64 "Enable building amd64 assembly implementation")

set(INSTALL_BIN_DIR "${CMAKE_INSTALL_PREFIX}/bin" CACHE PATH "Installation directory for executables")
set(INSTALL_LIB_DIR "${CMAKE_INSTALL_PREFIX}/lib" CACHE PATH "Installation directory for libraries")
set(INSTALL_INC_DIR "${CMAKE_INSTALL_PREFIX}/include" CACHE PATH "Installation directory for headers")

                ChangeLog file for zlib

Changes in 1.2.12 (27 Mar 2022)
- Cygwin does not have _wopen(), so do not create gzopen_w() there
- Permit a deflateParams() parameter change as soon as possible
- Limit hash table inserts after switch from stored deflate
- Fix bug when window full in deflate_stored()
- Fix CLEAR_HASH macro to be usable as a single statement
- Avoid a conversion error in gzseek when off_t type too small
- Have Makefile return non-zero error code on test failure
- Avoid some conversion warnings in gzread.c and gzwrite.c
- Update use of errno for newer Windows CE versions
- Small speedup to inflate [psumbera]
- Return an error if the gzputs string length can't fit in an int
- Add address checking in clang to -w option of configure
- Don't compute check value for raw inflate if asked to validate
- Handle case where inflateSync used when header never processed
- Avoid the use of ptrdiff_t
- Avoid an undefined behavior of memcpy() in gzappend()
- Avoid undefined behaviors of memcpy() in gz*printf()
- Avoid an undefined behavior of memcpy() in _tr_stored_block()
- Make the names in functions declarations identical to definitions
- Remove old assembler code in which bugs have manifested
- Fix deflateEnd() to not report an error at start of raw deflate
- Add legal disclaimer to README
- Emphasize the need to continue decompressing gzip members
- Correct the initialization requirements for deflateInit2()
- Fix a bug that can crash deflate on some input when using Z_FIXED
- Assure that the number of bits for deflatePrime() is valid
- Use a structure to make globals in enough.c evident
- Use a macro for the printf format of big_t in enough.c
- Clean up code style in enough.c, update version
- Use inline function instead of macro for index in enough.c
- Clarify that prefix codes are counted in enough.c
- Show all the codes for the maximum tables size in enough.c
- Add gznorm.c example, which normalizes gzip files
- Fix the zran.c example to work on a multiple-member gzip file
- Add tables for crc32_combine(), to speed it up by a factor of 200
- Add crc32_combine_gen() and crc32_combine_op() for fast combines
- Speed up software CRC-32 computation by a factor of 1.5 to 3
- Use atomic test and set, if available, for dynamic CRC tables
- Don't bother computing check value after successful inflateSync()
- Correct comment in crc32.c
- Add use of the ARMv8 crc32 instructions when requested
- Use ARM crc32 instructions if the ARM architecture has them
- Explicitly note that the 32-bit check values are 32 bits
- Avoid adding empty gzip member after gzflush with Z_FINISH
- Fix memory leak on error in gzlog.c
- Fix error in comment on the polynomial representation of a byte
- Clarify gz* function interfaces, referring to parameter names
- Change macro name in inflate.c to avoid collision in VxWorks
- Correct typo in blast.c
- Improve portability of contrib/minizip
- Fix indentation in minizip's zip.c
- Replace black/white with allow/block. (theresa-m)
- minizip warning fix if MAXU32 already defined. (gvollant)
- Fix unztell64() in minizip to work past 4GB. (Daniël Hörchner)
- Clean up minizip to reduce warnings for testing
- Add fallthrough comments for gcc
- Eliminate use of ULL constants
- Separate out address sanitizing from warnings in configure
- Remove destructive aspects of make distclean
- Check for cc masquerading as gcc or clang in configure
- Fix crc32.c to compile local functions only if used

Changes in 1.2.11 (15 Jan 2017)
- Fix deflate stored bug when pulling last block from window
- Permit immediate deflateParams changes before any deflate input

Changes in 1.2.10 (2 Jan 2017)
- Avoid warnings on snprintf() return value
- Fix bug in deflate_stored() for zero-length input

- Use zlib header window size if windowBits is 0 in inflateInit2()
- Remove compressBound() call in deflate.c to avoid linking compress.o
- Replace use of errno in gz* with functions, support WinCE [Alves]
- Provide alternative to perror() in minigzip.c for WinCE [Alves]
- Don't use _vsnprintf on later versions of MSVC [Lowman]
- Add CMake build script and input file [Lowman]
- Update contrib/minizip to 1.1 [Svensson, Vollant]
- Moved nintendods directory from contrib to .
- Moved nintendods directory from contrib to root
- Replace gzio.c with a new set of routines with the same functionality
- Add gzbuffer(), gzoffset(), gzclose_r(), gzclose_w() as part of above
- Update contrib/minizip to 1.1b
- Change gzeof() to return 0 on error instead of -1 to agree with zlib.h

Changes in 1.2.3.4 (21 Dec 2009)
- Use old school .SUFFIXES in Makefile.in for FreeBSD compatibility

- Add cast in trees.c t avoid a warning [Oberhumer]
- Avoid some warnings in fitblk.c, gun.c, gzjoin.c in examples [Oberhumer]
- Update make_vms.com [Zinser]
- Initialize state->write in inflateReset() since copied in inflate_fast()
- Be more strict on incomplete code sets in inflate_table() and increase
  ENOUGH and MAXD -- this repairs a possible security vulnerability for
  invalid inflate input.  Thanks to Tavis Ormandy and Markus Oberhumer for
  discovering the vulnerability and providing test cases.
  discovering the vulnerability and providing test cases
- Add ia64 support to configure for HP-UX [Smith]
- Add error return to gzread() for format or i/o error [Levin]
- Use malloc.h for OS/2 [Necasek]

Changes in 1.2.2.3 (27 May 2005)
- Replace 1U constants in inflate.c and inftrees.c for 64-bit compile
- Typecast fread() return values in gzio.c [Vollant]

- Increase sprintf() buffer size in gzdopen() to allow for large numbers
- Add INFLATE_STRICT to check distances against zlib header
- Improve WinCE errno handling and comments [Chang]
- Remove comment about no gzip header processing in FAQ
- Add Z_FIXED strategy option to deflateInit2() to force fixed trees
- Add updated make_vms.com [Coghlan], update README
- Create a new "examples" directory, move gzappend.c there, add zpipe.c,
  fitblk.c, gzlog.[ch], gzjoin.c, and zlib_how.html.
  fitblk.c, gzlog.[ch], gzjoin.c, and zlib_how.html
- Add FAQ entry and comments in deflate.c on uninitialized memory access
- Add Solaris 9 make options in configure [Gilbert]
- Allow strerror() usage in gzio.c for STDC
- Fix DecompressBuf in contrib/delphi/ZLib.pas [ManChesTer]
- Update contrib/masmx86/inffas32.asm and gvmat32.asm [Vollant]
- Use z_off_t for adler32_combine() and crc32_combine() lengths
- Make adler32() much faster for small len

Changes in 1.2.1.1 (9 January 2004)
- Update email address in README
- Several FAQ updates
- Fix a big fat bug in inftrees.c that prevented decoding valid
  dynamic blocks with only literals and no distance codes --
  Thanks to "Hot Emu" for the bug report and sample file
- Add a note to puff.c on no distance codes case.
- Add a note to puff.c on no distance codes case

Changes in 1.2.1 (17 November 2003)
- Remove a tab in contrib/gzappend/gzappend.c
- Update some interfaces in contrib for new zlib functions
- Update zlib version number in some contrib entries
- Add Windows CE definition for ptrdiff_t in zutil.h [Mai, Truta]
- Support shared libraries on Hurd and KFreeBSD [Brown]

    - Note permitted values of flush parameter of inflate()
- Add some FAQs (and even answers) to the FAQ
- Add contrib/inflate86/ for x86 faster inflate
- Add contrib/blast/ for PKWare Data Compression Library decompression
- Add contrib/puff/ simple inflate for deflate format description

Changes in 1.1.4 (11 March 2002)
- ZFREE was repeated on same allocation on some error conditions.
- ZFREE was repeated on same allocation on some error conditions
  This creates a security problem described in
  http://www.zlib.org/advisory-2002-03-11.txt
- Returned incorrect error (Z_MEM_ERROR) on some invalid data
- Avoid accesses before window for invalid distances with inflate window
  less than 32K.
  less than 32K
- force windowBits > 8 to avoid a bug in the encoder for a window size
  of 256 bytes. (A complete fix will be available in 1.1.5).
  of 256 bytes. (A complete fix will be available in 1.1.5)

Changes in 1.1.3 (9 July 1998)
- fix "an inflate input buffer bug that shows up on rare but persistent
  occasions" (Mark)
- fix gzread and gztell for concatenated .gz files (Didier Le Botlan)
- fix gzseek(..., SEEK_SET) in write mode
- fix crc check after a gzeek (Frank Faubert)

- move Makefile.sas to amiga/Makefile.sas

Changes in 1.1.1 (27 Feb 98)
- fix macros _tr_tally_* in deflate.h for debug mode  (Glenn Randers-Pehrson)
- remove block truncation heuristic which had very marginal effect for zlib
  (smaller lit_bufsize than in gzip 1.2.4) and degraded a little the
  compression ratio on some files. This also allows inlining _tr_tally for
  matches in deflate_slow.
  matches in deflate_slow
- added msdos/Makefile.w32 for WIN32 Microsoft Visual C++ (Bob Frazier)

Changes in 1.1.0 (24 Feb 98)
- do not return STREAM_END prematurely in inflate (John Bowler)
- revert to the zlib 1.0.8 inflate to avoid the gcc 2.8.0 bug (Jeremy Buhler)
- compile with -DFASTEST to get compression code optimized for speed only
- in minigzip, try mmap'ing the input file first (Miguel Albrecht)

Changes in 1.0.8 (27 Jan 1998)
- fixed offsets in contrib/asm386/gvmat32.asm (Gilles Vollant)
- fix gzgetc and gzputc for big endian systems (Markus Oberhumer)
- added compress2() to allow setting the compression level
- include sys/types.h to get off_t on some systems (Marc Lehmann & QingLong)
- use constant arrays for the static trees in trees.c instead of computing
  them at run time (thanks to Ken Raeburn for this suggestion). To create
  trees.h, compile with GEN_TREES_H and run "make test".
  trees.h, compile with GEN_TREES_H and run "make test"
- check return code of example in "make test" and display result
- pass minigzip command line options to file_compress
- simplifying code of inflateSync to avoid gcc 2.8 bug

- support CC="gcc -Wall" in configure -s (QingLong)
- avoid a flush caused by ftell in gzopen for write mode (Ken Raeburn)
- fix test for shared library support to avoid compiler warnings

- add inflateSyncPoint in zconf.h
- fix list of exported functions in nt/zlib.dnt and mdsos/zlib.def

Changes in 1.0.6 (19 Jan 1998)
- add functions gzprintf, gzputc, gzgetc, gztell, gzeof, gzseek, gzrewind and
  gzsetparams (thanks to Roland Giersig and Kevin Ruland for some of this code)
- Fix a deflate bug occurring only with compression level 0 (thanks to
  Andy Buckler for finding this one).
- In minigzip, pass transparently also the first byte for .Z files.
  Andy Buckler for finding this one)
- In minigzip, pass transparently also the first byte for .Z files
- return Z_BUF_ERROR instead of Z_OK if output buffer full in uncompress()
- check Z_FINISH in inflate (thanks to Marc Schluper)
- Implement deflateCopy (thanks to Adam Costello)
- make static libraries by default in configure, add --shared option.
- make static libraries by default in configure, add --shared option
- move MSDOS or Windows specific files to directory msdos
- suppress the notion of partial flush to simplify the interface
  (but the symbol Z_PARTIAL_FLUSH is kept for compatibility with 1.0.4)
- suppress history buffer provided by application to simplify the interface
  (this feature was not implemented anyway in 1.0.4)
- next_in and avail_in must be initialized before calling inflateInit or
  inflateInit2
- add EXPORT in all exported functions (for Windows DLL)
- added Makefile.nt (thanks to Stephen Williams)
- added the unsupported "contrib" directory:
   contrib/asm386/ by Gilles Vollant <info@winimage.com>
        386 asm code replacing longest_match().
        386 asm code replacing longest_match()
   contrib/iostream/ by Kevin Ruland <kevin@rodin.wustl.edu>
        A C++ I/O streams interface to the zlib gz* functions
   contrib/iostream2/  by Tyge Løvset <Tyge.Lovset@cmr.no>
        Another C++ I/O streams interface
   contrib/untgz/  by "Pedro A. Aranda Guti\irrez" <paag@tid.es>
        A very simple tar.gz file extractor using zlib
   contrib/visual-basic.txt by Carlos Rios <c_rios@sonda.cl>
        How to use compress(), uncompress() and the gz* functions from VB.
        How to use compress(), uncompress() and the gz* functions from VB
- pass params -f (filtered data), -h (huffman only), -1 to -9 (compression
  level) in minigzip (thanks to Tom Lane)

- use const for rommable constants in deflate
- added test for gzseek and gztell in example.c
- add undocumented function inflateSyncPoint() (hack for Paul Mackerras)
- add undocumented function zError to convert error code to string
  (for Tim Smithers)
- Allow compilation of gzio with -DNO_DEFLATE to avoid the compression code.
- Use default memcpy for Symantec MSDOS compiler.
- Allow compilation of gzio with -DNO_DEFLATE to avoid the compression code
- Use default memcpy for Symantec MSDOS compiler
- Add EXPORT keyword for check_func (needed for Windows DLL)
- add current directory to LD_LIBRARY_PATH for "make test"
- create also a link for libz.so.1
- added support for FUJITSU UXP/DS (thanks to Toshiaki Nomura)
- use $(SHAREDLIB) instead of libz.so in Makefile.in (for HPUX)
- added -soname for Linux in configure (Chun-Chung Chen,
- assign numbers to the exported functions in zlib.def (for Windows DLL)
- add advice in zlib.h for best usage of deflateSetDictionary
- work around compiler bug on Atari (cast Z_NULL in call of s->checkfn)
- allow compilation with ANSI keywords only enabled for TurboC in large model
- avoid "versionString"[0] (Borland bug)
- add NEED_DUMMY_RETURN for Borland
- use variable z_verbose for tracing in debug mode (L. Peter Deutsch).
- use variable z_verbose for tracing in debug mode (L. Peter Deutsch)
- allow compilation with CC
- defined STDC for OS/2 (David Charlap)
- limit external names to 8 chars for MVS (Thomas Lund)
- in minigzip.c, use static buffers only for 16-bit systems
- fix suffix check for "minigzip -d foo.gz"
- do not return an error for the 2nd of two consecutive gzflush() (Felix Lee)
- use _fdopen instead of fdopen for MSC >= 6.0 (Thomas Fanslau)
- added makelcc.bat for lcc-win32 (Tom St Denis)
- in Makefile.dj2, use copy and del instead of install and rm (Frank Donahoe)
- Avoid expanded $Id$. Use "rcs -kb" or "cvs admin -kb" to avoid Id expansion.
- Avoid expanded $Id$. Use "rcs -kb" or "cvs admin -kb" to avoid Id expansion
- check for unistd.h in configure (for off_t)
- remove useless check parameter in inflate_blocks_free
- avoid useless assignment of s->check to itself in inflate_blocks_new
- do not flush twice in gzclose (thanks to Ken Raeburn)
- rename FOPEN as F_OPEN to avoid clash with /usr/include/sys/file.h
- use NO_ERRNO_H instead of enumeration of operating systems with errno.h
- work around buggy fclose on pipes for HP/UX
- support zlib DLL with BORLAND C++ 5.0 (thanks to Glenn Randers-Pehrson)
- fix configure if CC is already equal to gcc

Changes in 1.0.5 (3 Jan 98)
- Fix inflate to terminate gracefully when fed corrupted or invalid data
- Use const for rommable constants in inflate
- Eliminate memory leaks on error conditions in inflate
- Removed some vestigial code in inflate
- Update web address in README

Changes in 1.0.4 (24 Jul 96)
- In very rare conditions, deflate(s, Z_FINISH) could fail to produce an EOF
  bit, so the decompressor could decompress all the correct data but went
  on to attempt decompressing extra garbage data. This affected minigzip too.
  on to attempt decompressing extra garbage data. This affected minigzip too
- zlibVersion and gzerror return const char* (needed for DLL)
- port to RISCOS (no fdopen, no multiple dots, no unlink, no fileno)
- use z_error only for DEBUG (avoid problem with DLLs)

Changes in 1.0.3 (2 Jul 96)
- use z_streamp instead of z_stream *, which is now a far pointer in MSDOS
  small and medium models; this makes the library incompatible with previous

- updated Makefile.dj2
- added algorithm.doc

Changes in 1.0.1 (20 May 96) [1.0 skipped to avoid confusion]
- fix array overlay in deflate.c which sometimes caused bad compressed data
- fix inflate bug with empty stored block
- fix MSDOS medium model which was broken in 0.99
- fix deflateParams() which could generate bad compressed data.
- fix deflateParams() which could generate bad compressed data
- Bytef is define'd instead of typedef'ed (work around Borland bug)
- added an INDEX file
- new makefiles for DJGPP (Makefile.dj2), 32-bit Borland (Makefile.b32),
  Watcom (Makefile.wat), Amiga SAS/C (Makefile.sas)
- speed up adler32 for modern machines without auto-increment
- added -ansi for IRIX in configure
- static_init_done in trees.c is an int
- define unlink as delete for VMS
- fix configure for QNX
- add configure branch for SCO and HPUX
- avoid many warnings (unused variables, dead assignments, etc...)
- no fdopen for BeOS
- fix the Watcom fix for 32 bit mode (define FAR as empty)
- removed redefinition of Byte for MKWERKS
- work around an MWKERKS bug (incorrect merge of all .h files)

Changes in 0.99 (27 Jan 96)
- allow preset dictionary shared between compressor and decompressor
- allow compression level 0 (no compression)
- add deflateParams in zlib.h: allow dynamic change of compression level
  and compression strategy.
  and compression strategy
- test large buffers and deflateParams in example.c
- add optional "configure" to build zlib as a shared library
- suppress Makefile.qnx, use configure instead
- fixed deflate for 64-bit systems (detected on Cray)
- fixed inflate_blocks for 64-bit systems (detected on Alpha)
- declare Z_DEFLATED in zlib.h (possible parameter for deflateInit2)
- always return Z_BUF_ERROR when deflate() has nothing to do

- check for WIN32 in zconf.h and zutil.c (avoid farmalloc in 32-bit mode)
- fix typo in Make_vms.com (f$trnlnm -> f$getsyi)
- in fcalloc, normalize pointer if size > 65520 bytes
- don't use special fcalloc for 32 bit Borland C++
- use STDC instead of __GO32__ to avoid redeclaring exit, calloc, etc...
- use Z_BINARY instead of BINARY
- document that gzclose after gzdopen will close the file
- allow "a" as mode in gzopen.
- allow "a" as mode in gzopen
- fix error checking in gzread
- allow skipping .gz extra-field on pipes
- added reference to Perl interface in README
- put the crc table in FAR data (I dislike more and more the medium model :)
- added get_crc_table
- added a dimension to all arrays (Borland C can't count).
- added a dimension to all arrays (Borland C can't count)
- workaround Borland C bug in declaration of inflate_codes_new & inflate_fast
- guard against multiple inclusion of *.h (for precompiled header on Mac)
- Watcom C pretends to be Microsoft C small model even in 32 bit mode.
- Watcom C pretends to be Microsoft C small model even in 32 bit mode
- don't use unsized arrays to avoid silly warnings by Visual C++:
     warning C4746: 'inflate_mask' : unsized array treated as  '__far'
     (what's wrong with far data in far model?).
     (what's wrong with far data in far model?)
- define enum out of inflate_blocks_state to allow compilation with C++

Changes in 0.95 (16 Aug 95)
- fix MSDOS small and medium model (now easier to adapt to any compiler)
- inlined send_bits
- fix the final (:-) bug for deflate with flush (output was correct but
  not completely flushed in rare occasions).
  not completely flushed in rare occasions)
- default window size is same for compression and decompression
  (it's now sufficient to set MAX_WBITS in zconf.h).
  (it's now sufficient to set MAX_WBITS in zconf.h)
- voidp -> voidpf and voidnp -> voidp (for consistency with other
  typedefs and because voidnp was not near in large model).
  typedefs and because voidnp was not near in large model)

Changes in 0.94 (13 Aug 95)
- support MSDOS medium model
- fix deflate with flush (could sometimes generate bad output)
- fix deflateReset (zlib header was incorrectly suppressed)
- added support for VMS
- allow a compression level in gzopen()
- gzflush now calls fflush
- For deflate with flush, flush even if no more input is provided.
- For deflate with flush, flush even if no more input is provided
- rename libgz.a as libz.a
- avoid complex expression in infcodes.c triggering Turbo C bug
- work around a problem with gcc on Alpha (in INSERT_STRING)
- don't use inline functions (problem with some gcc versions)
- allow renaming of Byte, uInt, etc... with #define.
- allow renaming of Byte, uInt, etc... with #define
- avoid warning about (unused) pointer before start of array in deflate.c
- avoid various warnings in gzio.c, example.c, infblock.c, adler32.c, zutil.c
- avoid reserved word 'new' in trees.c

Changes in 0.93 (25 June 95)
- temporarily disable inline functions
- make deflate deterministic
- give enough lookahead for PARTIAL_FLUSH
- Set binary mode for stdin/stdout in minigzip.c for OS/2
- don't even use signed char in inflate (not portable enough)
- fix inflate memory leak for segmented architectures

Changes in 0.92 (3 May 95)
- don't assume that char is signed (problem on SGI)
- Clear bit buffer when starting a stored block
- no memcpy on Pyramid
- suppressed inftest.c
- optimized fill_window, put longest_match inline for gcc
- optimized inflate on stored blocks.
- optimized inflate on stored blocks
- untabify all sources to simplify patches

Changes in 0.91 (2 May 95)
- Default MEM_LEVEL is 8 (not 9 for Unix) as documented in zlib.h
- Document the memory requirements in zconf.h
- added "make install"
- fix sync search logic in inflateSync
- deflate(Z_FULL_FLUSH) now works even if output buffer too short
- after inflateSync, don't scare people with just "lo world"
- added support for DJGPP

Changes in 0.9 (1 May 95)
- don't assume that zalloc clears the allocated memory (the TurboC bug
  was Mark's bug after all :)
- let again gzread copy uncompressed data unchanged (was working in 0.71)
- deflate(Z_FULL_FLUSH), inflateReset and inflateSync are now fully implemented
- added a test of inflateSync in example.c
- moved MAX_WBITS to zconf.h because users might want to change that.
- moved MAX_WBITS to zconf.h because users might want to change that
- document explicitly that zalloc(64K) on MSDOS must return a normalized
  pointer (zero offset)
- added Makefiles for Microsoft C, Turbo C, Borland C++
- faster crc32()

Changes in 0.8 (29 April 95)
- added fast inflate (inffast.c)
- deflate(Z_FINISH) now returns Z_STREAM_END when done. Warning: this
  is incompatible with previous versions of zlib which returned Z_OK.
  is incompatible with previous versions of zlib which returned Z_OK
- work around a TurboC compiler bug (bad code for b << 0, see infutil.h)
  (actually that was not a compiler bug, see 0.81 above)
- gzread no longer reads one extra byte in certain cases
- In gzio destroy(), don't reference a freed structure
- avoid many warnings for MSDOS
- avoid the ERROR symbol which is used by MS Windows

Changes in 0.71 (14 April 95)
- Fixed more MSDOS compilation problems :( There is still a bug with
  TurboC large model.
  TurboC large model

Changes in 0.7 (14 April 95)
- Added full inflate support.
- Added full inflate support
- Simplified the crc32() interface. The pre- and post-conditioning
  (one's complement) is now done inside crc32(). WARNING: this is
  incompatible with previous versions; see zlib.h for the new usage.
  incompatible with previous versions; see zlib.h for the new usage

Changes in 0.61 (12 April 95)
- workaround for a bug in TurboC. example and minigzip now work on MSDOS.
- workaround for a bug in TurboC. example and minigzip now work on MSDOS

Changes in 0.6 (11 April 95)
- added minigzip.c
- added gzdopen to reopen a file descriptor as gzFile
- added transparent reading of non-gziped files in gzread.
- added transparent reading of non-gziped files in gzread
- fixed bug in gzread (don't read crc as data)
- fixed bug in destroy (gzio.c) (don't return Z_STREAM_END for gzclose).
- fixed bug in destroy (gzio.c) (don't return Z_STREAM_END for gzclose)
- don't allocate big arrays in the stack (for MSDOS)
- fix some MSDOS compilation problems

Changes in 0.5:
- do real compression in deflate.c. Z_PARTIAL_FLUSH is supported but
  not yet Z_FULL_FLUSH.
  not yet Z_FULL_FLUSH
- support decompression but only in a single step (forced Z_FINISH)
- added opaque object for zalloc and zfree.
- added opaque object for zalloc and zfree
- added deflateReset and inflateReset
- added a variable zlib_version for consistency checking.
- renamed the 'filter' parameter of deflateInit2 as 'strategy'.
  Added Z_FILTERED and Z_HUFFMAN_ONLY constants.
- added a variable zlib_version for consistency checking
- renamed the 'filter' parameter of deflateInit2 as 'strategy'
  Added Z_FILTERED and Z_HUFFMAN_ONLY constants

Changes in 0.4:
- avoid "zip" everywhere, use zlib instead of ziplib.
- avoid "zip" everywhere, use zlib instead of ziplib
- suppress Z_BLOCK_FLUSH, interpret Z_PARTIAL_FLUSH as block flush
  if compression method == 8.
  if compression method == 8
- added adler32 and crc32
- renamed deflateOptions as deflateInit2, call one or the other but not both
- added the method parameter for deflateInit2.
- added the method parameter for deflateInit2
- added inflateInit2
- simplied considerably deflateInit and inflateInit by not supporting
  user-provided history buffer. This is supported only in deflateInit2
  and inflateInit2.
  and inflateInit2

Changes in 0.3:
- prefix all macro names with Z_
- use Z_FINISH instead of deflateEnd to finish compression.
- use Z_FINISH instead of deflateEnd to finish compression
- added Z_HUFFMAN_ONLY
- added gzerror()

LDFLAGS=
TEST_LDFLAGS=-L. libz.a
LDSHARED=$(CC)
CPP=$(CC) -E

STATICLIB=libz.a
SHAREDLIB=libz.so
SHAREDLIBV=libz.so.1.2.11
SHAREDLIBV=libz.so.1.2.12
SHAREDLIBM=libz.so.1
LIBS=$(STATICLIB) $(SHAREDLIBV)

AR=ar
ARFLAGS=rc
RANLIB=ranlib
LDCONFIG=ldconfig

teststatic: static
	@TMPST=tmpst_$$; \
	if echo hello world | ./minigzip | ./minigzip -d && ./example $$TMPST ; then \
	  echo '		*** zlib test OK ***'; \
	else \
	  echo '		*** zlib test FAILED ***'; false; \
	fi; \
	rm -f $$TMPST
	fi
	@rm -f tmpst_$$

testshared: shared
	@LD_LIBRARY_PATH=`pwd`:$(LD_LIBRARY_PATH) ; export LD_LIBRARY_PATH; \
	LD_LIBRARYN32_PATH=`pwd`:$(LD_LIBRARYN32_PATH) ; export LD_LIBRARYN32_PATH; \
	DYLD_LIBRARY_PATH=`pwd`:$(DYLD_LIBRARY_PATH) ; export DYLD_LIBRARY_PATH; \
	SHLIB_PATH=`pwd`:$(SHLIB_PATH) ; export SHLIB_PATH; \
	TMPSH=tmpsh_$$; \
	if echo hello world | ./minigzipsh | ./minigzipsh -d && ./examplesh $$TMPSH; then \
	  echo '		*** zlib shared test OK ***'; \
	else \
	  echo '		*** zlib shared test FAILED ***'; false; \
	fi; \
	rm -f $$TMPSH
	fi
	@rm -f tmpsh_$$

test64: all64
	@TMP64=tmp64_$$; \
	if echo hello world | ./minigzip64 | ./minigzip64 -d && ./example64 $$TMP64; then \
	  echo '		*** zlib 64-bit test OK ***'; \
	else \
	  echo '		*** zlib 64-bit test FAILED ***'; false; \
	fi; \
	rm -f $$TMP64
	fi
	@rm -f tmp64_$$

infcover.o: $(SRCDIR)test/infcover.c $(SRCDIR)zlib.h zconf.h
	$(CC) $(CFLAGS) $(ZINCOUT) -c -o $@ $(SRCDIR)test/infcover.c

infcover: infcover.o libz.a
	$(CC) $(CFLAGS) -o $@ infcover.o libz.a

	   libz.* foo.gz so_locations \
	   _match.s maketree contrib/infback9/*.o
	rm -rf objs
	rm -f *.gcda *.gcno *.gcov
	rm -f contrib/infback9/*.gcda contrib/infback9/*.gcno contrib/infback9/*.gcov

maintainer-clean: distclean
distclean: clean zconf zconf.h.cmakein docs
distclean: clean zconf zconf.h.cmakein
	rm -f Makefile zlib.pc configure.log
	-@rm -f .DS_Store
	@if [ -f Makefile.in ]; then \
	printf 'all:\n\t-@echo "Please use ./configure first.  Thank you."\n' > Makefile ; \
	printf '\ndistclean:\n\tmake -f Makefile.in distclean\n' >> Makefile ; \
	touch -r $(SRCDIR)Makefile.in Makefile ; fi
	@if [ ! -f zconf.h.in ]; then rm -f zconf.h zconf.h.cmakein ; fi
	@if [ ! -f zlib.3 ]; then rm -f zlib.3.pdf ; fi

tags:
	etags $(SRCDIR)*.[ch]

adler32.o zutil.o: $(SRCDIR)zutil.h $(SRCDIR)zlib.h zconf.h
gzclose.o gzlib.o gzread.o gzwrite.o: $(SRCDIR)zlib.h zconf.h $(SRCDIR)gzguts.h
compress.o example.o minigzip.o uncompr.o: $(SRCDIR)zlib.h zconf.h

ZLIB DATA COMPRESSION LIBRARY

zlib 1.2.11 is a general purpose data compression library.  All the code is
zlib 1.2.12 is a general purpose data compression library.  All the code is
thread safe.  The data format used by the zlib library is described by RFCs
(Request for Comments) 1950 to 1952 in the files
http://tools.ietf.org/html/rfc1950 (zlib format), rfc1951 (deflate format) and
rfc1952 (gzip format).

All functions of the compression library are documented in the file zlib.h
(volunteer to write man pages welcome, contact zlib@gzip.org).  A usage example

PLEASE read the zlib FAQ http://zlib.net/zlib_faq.html before asking for help.

Mark Nelson <markn@ieee.org> wrote an article about zlib for the Jan.  1997
issue of Dr.  Dobb's Journal; a copy of the article is available at
http://marknelson.us/1997/01/01/zlib-engine/ .

The changes made in version 1.2.11 are documented in the file ChangeLog.
The changes made in version 1.2.12 are documented in the file ChangeLog.

Unsupported third party contributions are provided in directory contrib/ .

zlib is available in Java using the java.util.zip package, documented at
http://java.sun.com/developer/technicalArticles/Programming/compression/ .

A Perl interface to zlib written by Paul Marquess <pmqs@cpan.org> is available

  The deflate format used by zlib was defined by Phil Katz.  The deflate and
  zlib specifications were written by L.  Peter Deutsch.  Thanks to all the
  people who reported problems and suggested various improvements in zlib; they
  are too numerous to cite here.

Copyright notice:

 (C) 1995-2017 Jean-loup Gailly and Mark Adler
 (C) 1995-2022 Jean-loup Gailly and Mark Adler

  This software is provided 'as-is', without any express or implied
  warranty.  In no event will the authors be held liable for any damages
  arising from the use of this software.

  Permission is granted to anyone to use this software for any purpose,
  including commercial applications, and to alter it and redistribute it

  Jean-loup Gailly        Mark Adler
  jloup@gzip.org          madler@alumni.caltech.edu

If you use the zlib library in a product, we would appreciate *not* receiving
lengthy legal documents to sign.  The sources are provided for free but without
warranty of any kind.  The library has been entirely written by Jean-loup
Gailly and Mark Adler; it does not include third-party code.
Gailly and Mark Adler; it does not include third-party code.  We make all
contributions to and distributions of this project solely in our personal
capacity, and are not conveying any rights to any intellectual property of
any third parties.

If you redistribute modified sources, we would appreciate that you include in
the file ChangeLog history information documenting your changes.  Please read
the FAQ for more information on the distribution of modified source versions.

cover=0
zprefix=0
zconst=0
build64=0
gcc=0
warn=0
debug=0
sanitize=0
old_cc="$CC"
old_cflags="$CFLAGS"
OBJC='$(OBJZ) $(OBJG)'
PIC_OBJC='$(PIC_OBJZ) $(PIC_OBJG)'

# leave this script, optionally in a bad way
leave()

    -6* | --64) build64=1; shift ;;
    -a*=* | --archs=*) ARCHS=`echo $1 | sed 's/.*=//'`; shift ;;
    --sysconfdir=*) echo "ignored option: --sysconfdir" | tee -a configure.log; shift ;;
    --localstatedir=*) echo "ignored option: --localstatedir" | tee -a configure.log; shift ;;
    -c* | --const) zconst=1; shift ;;
    -w* | --warn) warn=1; shift ;;
    -d* | --debug) debug=1; shift ;;
    --sanitize) sanitize=1; shift ;;
    *)
      echo "unknown option: $1" | tee -a configure.log
      echo "$0 --help for help" | tee -a configure.log
      leave 1;;
    esac
done

# check for gcc vs. cc and set compile and link flags based on the system identified by uname
cat > $test.c <<EOF
extern int getchar();
int hello() {return getchar();}
EOF

if test -z "$CC"; then
test -z "$CC" && echo Checking for ${CROSS_PREFIX}gcc... | tee -a configure.log
cc=${CC-${CROSS_PREFIX}gcc}
  echo Checking for ${CROSS_PREFIX}gcc... | tee -a configure.log
  if ${CROSS_PREFIX}gcc -v >/dev/null 2>&1; then
    cc=${CROSS_PREFIX}gcc
  else
    cc=${CROSS_PREFIX}cc
  fi
else
  cc=${CC}
fi

cflags=${CFLAGS-"-O3"}
# to force the asm version use: CFLAGS="-O3 -DASMV" ./configure
case "$cc" in
  *gcc*) gcc=1 ;;
  *clang*) gcc=1 ;;
esac
case `$cc -v 2>&1` in

  if test "$warn" -eq 1; then
    if test "$zconst" -eq 1; then
      CFLAGS="${CFLAGS} -Wall -Wextra -Wcast-qual -pedantic -DZLIB_CONST"
    else
      CFLAGS="${CFLAGS} -Wall -Wextra -pedantic"
    fi
  fi
  if test $sanitize -eq 1; then
    CFLAGS="${CFLAGS} -fsanitize=address"
  fi
  if test $debug -eq 1; then
    CFLAGS="${CFLAGS} -DZLIB_DEBUG"
    SFLAGS="${SFLAGS} -DZLIB_DEBUG"
  fi
  if test -z "$uname"; then
    uname=`(uname -s || echo unknown) 2>/dev/null`
  fi

    test "`( $* ) 2>&1 | tee -a configure.log`" = ""
  }
  echo - using any output from compiler to indicate an error >> configure.log
else
  try()
  {
    show $*
    ( $* ) >> configure.log 2>&1
    got=`( $* ) 2>&1`
    ret=$?
    if test "$got" != ""; then
      printf "%s\n" "$got" >> configure.log
    fi
    if test $ret -ne 0; then
      echo "(exit code "$ret")" >> configure.log
    fi
    return $ret
  }
fi

tryboth()
{
  show $*
  got=`( $* ) 2>&1`
  ret=$?
  if test "$got" != ""; then
  printf %s "$got" >> configure.log
    printf "%s\n" "$got" >> configure.log
  fi
  if test $ret -ne 0; then
    echo "(exit code "$ret")" >> configure.log
    return $ret
  fi
  test "$got" = ""
}

cat > $test.c << EOF
int foo() { return 0; }

cat > $test.c <<EOF
#include <stdio.h>
#include <stdlib.h>
size_t dummy = 0;
EOF
if try $CC -c $CFLAGS $test.c; then
  echo "Checking for size_t... Yes." | tee -a configure.log
  need_sizet=0
else
  echo "Checking for size_t... No." | tee -a configure.log
  need_sizet=1
fi

  # find a size_t integer type
echo >> configure.log

  # check for long long
# find the size_t integer type, if needed
if test $need_sizet -eq 1; then
  cat > $test.c <<EOF
  cat > $test.c << EOF
long long dummy = 0;
EOF
  if try $CC -c $CFLAGS $test.c; then
    echo "Checking for long long... Yes." | tee -a configure.log
    cat > $test.c <<EOF
#include <stdio.h>
int main(void) {

    return 0;
}
EOF
  fi
  if try $CC $CFLAGS -o $test $test.c; then
    sizet=`./$test`
    echo "Checking for a pointer-size integer type..." $sizet"." | tee -a configure.log
    CFLAGS="${CFLAGS} -DNO_SIZE_T=${sizet}"
    SFLAGS="${SFLAGS} -DNO_SIZE_T=${sizet}"
  else
    echo "Failed to find a pointer-size integer type." | tee -a configure.log
    echo "Checking for a pointer-size integer type... not found." | tee -a configure.log
    leave 1
  fi
fi

if test $need_sizet -eq 1; then
  CFLAGS="${CFLAGS} -DNO_SIZE_T=${sizet}"
  SFLAGS="${SFLAGS} -DNO_SIZE_T=${sizet}"
fi

echo >> configure.log

# check for large file support, and if none, check for fseeko()
cat > $test.c <<EOF
#include <sys/types.h>

echo SFLAGS = $SFLAGS >> configure.log
echo SHAREDLIB = $SHAREDLIB >> configure.log
echo SHAREDLIBM = $SHAREDLIBM >> configure.log
echo SHAREDLIBV = $SHAREDLIBV >> configure.log
echo STATICLIB = $STATICLIB >> configure.log
echo TEST = $TEST >> configure.log
echo VER = $VER >> configure.log
echo Z_U4 = $Z_U4 >> configure.log
echo SRCDIR = $SRCDIR >> configure.log
echo exec_prefix = $exec_prefix >> configure.log
echo includedir = $includedir >> configure.log
echo libdir = $libdir >> configure.log
echo mandir = $mandir >> configure.log
echo prefix = $prefix >> configure.log
echo sharedlibdir = $sharedlibdir >> configure.log

All files under this contrib directory are UNSUPPORTED. There were
provided by users of zlib and were not tested by the authors of zlib.
Use at your own risk. Please contact the authors of the contributions
for help about these, not the zlib authors. Thanks.


ada/        by Dmitriy Anisimkov <anisimkov@yahoo.com>
        Support for Ada
        See http://zlib-ada.sourceforge.net/

amd64/      by Mikhail Teterin <mi@ALDAN.algebra.com>
        asm code for AMD64
        See patch at http://www.freebsd.org/cgi/query-pr.cgi?pr=bin/96393

asm686/     by Brian Raiter <breadbox@muppetlabs.com>
        asm code for Pentium and PPro/PII, using the AT&T (GNU as) syntax
        See http://www.muppetlabs.com/~breadbox/software/assembly.html

blast/      by Mark Adler <madler@alumni.caltech.edu>
        Decompressor for output of PKWare Data Compression Library (DCL)

delphi/     by Cosmin Truta <cosmint@cs.ubbcluj.ro>
        Support for Delphi and C++ Builder

dotzlib/    by Henrik Ravn <henrik@ravn.com>
        Support for Microsoft .Net and Visual C++ .Net

gcc_gvmat64/by Gilles Vollant <info@winimage.com>
        GCC Version of x86 64-bit (AMD64 and Intel EM64t) code for x64
        assembler to replace longest_match() and inflate_fast()

infback9/   by Mark Adler <madler@alumni.caltech.edu>
        Unsupported diffs to infback to decode the deflate64 format

inflate86/  by Chris Anderson <christop@charm.net>
        Tuned x86 gcc asm code to replace inflate_fast()

iostream/   by Kevin Ruland <kevin@rodin.wustl.edu>
        A C++ I/O streams interface to the zlib gz* functions

iostream2/  by Tyge Løvset <Tyge.Lovset@cmr.no>
        Another C++ I/O streams interface

iostream3/  by Ludwig Schwardt <schwardt@sun.ac.za>
            and Kevin Ruland <kevin@rodin.wustl.edu>
        Yet another C++ I/O streams interface

masmx64/    by Gilles Vollant <info@winimage.com>
        x86 64-bit (AMD64 and Intel EM64t) code for x64 assembler to
        replace longest_match() and inflate_fast(),  also masm x86
        64-bits translation of Chris Anderson inflate_fast()

masmx86/    by Gilles Vollant <info@winimage.com>
        x86 asm code to replace longest_match() and inflate_fast(),
        for Visual C++ and MASM (32 bits).
        Based on Brian Raiter (asm686) and Chris Anderson (inflate86)

minizip/    by Gilles Vollant <info@winimage.com>
        Mini zip and unzip based on zlib
        Includes Zip64 support by Mathias Svensson <mathias@result42.com>
        See http://www.winimage.com/zLibDll/minizip.html

pascal/     by Bob Dellaca <bobdl@xtra.co.nz> et al.
        Support for Pascal

/*
 * match.S -- optimized version of longest_match()
 * based on the similar work by Gilles Vollant, and Brian Raiter, written 1998
 *
 * This is free software; you can redistribute it and/or modify it
 * under the terms of the BSD License. Use by owners of Che Guevarra
 * parafernalia is prohibited, where possible, and highly discouraged
 * elsewhere.
 */

#ifndef NO_UNDERLINE
#	define	match_init	_match_init
#	define	longest_match	_longest_match
#endif

#define	scanend		ebx
#define	scanendw	bx
#define	chainlenwmask	edx /* high word: current chain len low word: s->wmask */
#define	curmatch	rsi
#define	curmatchd	esi
#define	windowbestlen	r8
#define	scanalign	r9
#define	scanalignd	r9d
#define	window		r10
#define	bestlen		r11
#define	bestlend	r11d
#define	scanstart	r12d
#define	scanstartw	r12w
#define scan		r13
#define nicematch	r14d
#define	limit		r15
#define	limitd		r15d
#define prev		rcx

/*
 * The 258 is a "magic number, not a parameter -- changing it
 * breaks the hell loose
 */
#define	MAX_MATCH	(258)
#define	MIN_MATCH	(3)
#define	MIN_LOOKAHEAD	(MAX_MATCH + MIN_MATCH + 1)
#define	MAX_MATCH_8	((MAX_MATCH + 7) & ~7)

/* stack frame offsets */
#define	LocalVarsSize	(112)
#define _chainlenwmask	( 8-LocalVarsSize)(%rsp)
#define _windowbestlen	(16-LocalVarsSize)(%rsp)
#define save_r14        (24-LocalVarsSize)(%rsp)
#define save_rsi        (32-LocalVarsSize)(%rsp)
#define save_rbx        (40-LocalVarsSize)(%rsp)
#define save_r12        (56-LocalVarsSize)(%rsp)
#define save_r13        (64-LocalVarsSize)(%rsp)
#define save_r15        (80-LocalVarsSize)(%rsp)


.globl	match_init, longest_match

/*
 * On AMD64 the first argument of a function (in our case -- the pointer to
 * deflate_state structure) is passed in %rdi, hence our offsets below are
 * all off of that.
 */

/* you can check the structure offset by running

#include <stdlib.h>
#include <stdio.h>
#include "deflate.h"

void print_depl()
{
deflate_state ds;
deflate_state *s=&ds;
printf("size pointer=%u\n",(int)sizeof(void*));

printf("#define dsWSize         (%3u)(%%rdi)\n",(int)(((char*)&(s->w_size))-((char*)s)));
printf("#define dsWMask         (%3u)(%%rdi)\n",(int)(((char*)&(s->w_mask))-((char*)s)));
printf("#define dsWindow        (%3u)(%%rdi)\n",(int)(((char*)&(s->window))-((char*)s)));
printf("#define dsPrev          (%3u)(%%rdi)\n",(int)(((char*)&(s->prev))-((char*)s)));
printf("#define dsMatchLen      (%3u)(%%rdi)\n",(int)(((char*)&(s->match_length))-((char*)s)));
printf("#define dsPrevMatch     (%3u)(%%rdi)\n",(int)(((char*)&(s->prev_match))-((char*)s)));
printf("#define dsStrStart      (%3u)(%%rdi)\n",(int)(((char*)&(s->strstart))-((char*)s)));
printf("#define dsMatchStart    (%3u)(%%rdi)\n",(int)(((char*)&(s->match_start))-((char*)s)));
printf("#define dsLookahead     (%3u)(%%rdi)\n",(int)(((char*)&(s->lookahead))-((char*)s)));
printf("#define dsPrevLen       (%3u)(%%rdi)\n",(int)(((char*)&(s->prev_length))-((char*)s)));
printf("#define dsMaxChainLen   (%3u)(%%rdi)\n",(int)(((char*)&(s->max_chain_length))-((char*)s)));
printf("#define dsGoodMatch     (%3u)(%%rdi)\n",(int)(((char*)&(s->good_match))-((char*)s)));
printf("#define dsNiceMatch     (%3u)(%%rdi)\n",(int)(((char*)&(s->nice_match))-((char*)s)));
}

*/


/*
  to compile for XCode 3.2 on MacOSX x86_64
  - run "gcc -g -c -DXCODE_MAC_X64_STRUCTURE amd64-match.S"
 */


#ifndef CURRENT_LINX_XCODE_MAC_X64_STRUCTURE
#define dsWSize		( 68)(%rdi)
#define dsWMask		( 76)(%rdi)
#define dsWindow	( 80)(%rdi)
#define dsPrev		( 96)(%rdi)
#define dsMatchLen	(144)(%rdi)
#define dsPrevMatch	(148)(%rdi)
#define dsStrStart	(156)(%rdi)
#define dsMatchStart	(160)(%rdi)
#define dsLookahead	(164)(%rdi)
#define dsPrevLen	(168)(%rdi)
#define dsMaxChainLen	(172)(%rdi)
#define dsGoodMatch	(188)(%rdi)
#define dsNiceMatch	(192)(%rdi)

#else 

#ifndef STRUCT_OFFSET
#	define STRUCT_OFFSET	(0)
#endif


#define dsWSize		( 56 + STRUCT_OFFSET)(%rdi)
#define dsWMask		( 64 + STRUCT_OFFSET)(%rdi)
#define dsWindow	( 72 + STRUCT_OFFSET)(%rdi)
#define dsPrev		( 88 + STRUCT_OFFSET)(%rdi)
#define dsMatchLen	(136 + STRUCT_OFFSET)(%rdi)
#define dsPrevMatch	(140 + STRUCT_OFFSET)(%rdi)
#define dsStrStart	(148 + STRUCT_OFFSET)(%rdi)
#define dsMatchStart	(152 + STRUCT_OFFSET)(%rdi)
#define dsLookahead	(156 + STRUCT_OFFSET)(%rdi)
#define dsPrevLen	(160 + STRUCT_OFFSET)(%rdi)
#define dsMaxChainLen	(164 + STRUCT_OFFSET)(%rdi)
#define dsGoodMatch	(180 + STRUCT_OFFSET)(%rdi)
#define dsNiceMatch	(184 + STRUCT_OFFSET)(%rdi)

#endif




.text

/* uInt longest_match(deflate_state *deflatestate, IPos curmatch) */

longest_match:
/*
 * Retrieve the function arguments. %curmatch will hold cur_match
 * throughout the entire function (passed via rsi on amd64).
 * rdi will hold the pointer to the deflate_state (first arg on amd64)
 */
		mov     %rsi, save_rsi
		mov     %rbx, save_rbx
		mov	%r12, save_r12
		mov     %r13, save_r13
		mov     %r14, save_r14
		mov     %r15, save_r15

/* uInt wmask = s->w_mask;						*/
/* unsigned chain_length = s->max_chain_length;				*/
/* if (s->prev_length >= s->good_match) {				*/
/*     chain_length >>= 2;						*/
/* }									*/

		movl	dsPrevLen, %eax
		movl	dsGoodMatch, %ebx
		cmpl	%ebx, %eax
		movl	dsWMask, %eax
		movl	dsMaxChainLen, %chainlenwmask
		jl	LastMatchGood
		shrl	$2, %chainlenwmask
LastMatchGood:

/* chainlen is decremented once beforehand so that the function can	*/
/* use the sign flag instead of the zero flag for the exit test.	*/
/* It is then shifted into the high word, to make room for the wmask	*/
/* value, which it will always accompany.				*/

		decl	%chainlenwmask
		shll	$16, %chainlenwmask
		orl	%eax, %chainlenwmask

/* if ((uInt)nice_match > s->lookahead) nice_match = s->lookahead;	*/

		movl	dsNiceMatch, %eax
		movl	dsLookahead, %ebx
		cmpl	%eax, %ebx
		jl	LookaheadLess
		movl	%eax, %ebx
LookaheadLess:	movl	%ebx, %nicematch

/* register Bytef *scan = s->window + s->strstart;			*/

		mov	dsWindow, %window
		movl	dsStrStart, %limitd
		lea	(%limit, %window), %scan

/* Determine how many bytes the scan ptr is off from being		*/
/* dword-aligned.							*/

		mov	%scan, %scanalign
		negl	%scanalignd
		andl	$3, %scanalignd

/* IPos limit = s->strstart > (IPos)MAX_DIST(s) ?			*/
/*     s->strstart - (IPos)MAX_DIST(s) : NIL;				*/

		movl	dsWSize, %eax
		subl	$MIN_LOOKAHEAD, %eax
		xorl	%ecx, %ecx
		subl	%eax, %limitd
		cmovng	%ecx, %limitd

/* int best_len = s->prev_length;					*/

		movl	dsPrevLen, %bestlend

/* Store the sum of s->window + best_len in %windowbestlen locally, and in memory.	*/

		lea	(%window, %bestlen), %windowbestlen
		mov	%windowbestlen, _windowbestlen

/* register ush scan_start = *(ushf*)scan;				*/
/* register ush scan_end   = *(ushf*)(scan+best_len-1);			*/
/* Posf *prev = s->prev;						*/

		movzwl	(%scan), %scanstart
		movzwl	-1(%scan, %bestlen), %scanend
		mov	dsPrev, %prev

/* Jump into the main loop.						*/

		movl	%chainlenwmask, _chainlenwmask
		jmp	LoopEntry

.balign 16

/* do {
 *     match = s->window + cur_match;
 *     if (*(ushf*)(match+best_len-1) != scan_end ||
 *         *(ushf*)match != scan_start) continue;
 *     [...]
 * } while ((cur_match = prev[cur_match & wmask]) > limit
 *          && --chain_length != 0);
 *
 * Here is the inner loop of the function. The function will spend the
 * majority of its time in this loop, and majority of that time will
 * be spent in the first ten instructions.
 */
LookupLoop:
		andl	%chainlenwmask, %curmatchd
		movzwl	(%prev, %curmatch, 2), %curmatchd
		cmpl	%limitd, %curmatchd
		jbe	LeaveNow
		subl	$0x00010000, %chainlenwmask
		js	LeaveNow
LoopEntry:	cmpw	-1(%windowbestlen, %curmatch), %scanendw
		jne	LookupLoop
		cmpw	%scanstartw, (%window, %curmatch)
		jne	LookupLoop

/* Store the current value of chainlen.					*/
		movl	%chainlenwmask, _chainlenwmask

/* %scan is the string under scrutiny, and %prev to the string we	*/
/* are hoping to match it up with. In actuality, %esi and %edi are	*/
/* both pointed (MAX_MATCH_8 - scanalign) bytes ahead, and %edx is	*/
/* initialized to -(MAX_MATCH_8 - scanalign).				*/

		mov	$(-MAX_MATCH_8), %rdx
		lea	(%curmatch, %window), %windowbestlen
		lea	MAX_MATCH_8(%windowbestlen, %scanalign), %windowbestlen
		lea	MAX_MATCH_8(%scan, %scanalign), %prev

/* the prefetching below makes very little difference... */
		prefetcht1	(%windowbestlen, %rdx)
		prefetcht1	(%prev, %rdx)

/*
 * Test the strings for equality, 8 bytes at a time. At the end,
 * adjust %rdx so that it is offset to the exact byte that mismatched.
 *
 * It should be confessed that this loop usually does not represent
 * much of the total running time. Replacing it with a more
 * straightforward "rep cmpsb" would not drastically degrade
 * performance -- unrolling it, for example, makes no difference.
 */

#undef USE_SSE	/* works, but is 6-7% slower, than non-SSE... */

LoopCmps:
#ifdef USE_SSE
		/* Preload the SSE registers */
		movdqu	  (%windowbestlen, %rdx), %xmm1
		movdqu	  (%prev, %rdx), %xmm2
		pcmpeqb	%xmm2, %xmm1
		movdqu	16(%windowbestlen, %rdx), %xmm3
		movdqu	16(%prev, %rdx), %xmm4
		pcmpeqb	%xmm4, %xmm3
		movdqu	32(%windowbestlen, %rdx), %xmm5
		movdqu	32(%prev, %rdx), %xmm6
		pcmpeqb	%xmm6, %xmm5
		movdqu	48(%windowbestlen, %rdx), %xmm7
		movdqu	48(%prev, %rdx), %xmm8
		pcmpeqb	%xmm8, %xmm7

		/* Check the comparisions' results */
		pmovmskb %xmm1, %rax
		notw	%ax
		bsfw	%ax, %ax
		jnz	LeaveLoopCmps
		
		/* this is the only iteration of the loop with a possibility of having
		   incremented rdx by 0x108 (each loop iteration add 16*4 = 0x40 
		   and (0x40*4)+8=0x108 */
		add	$8, %rdx
		jz LenMaximum
		add	$8, %rdx

		
		pmovmskb %xmm3, %rax
		notw	%ax
		bsfw	%ax, %ax
		jnz	LeaveLoopCmps
		
		
		add	$16, %rdx


		pmovmskb %xmm5, %rax
		notw	%ax
		bsfw	%ax, %ax
		jnz	LeaveLoopCmps
		
		add	$16, %rdx


		pmovmskb %xmm7, %rax
		notw	%ax
		bsfw	%ax, %ax
		jnz	LeaveLoopCmps
		
		add	$16, %rdx
		
		jmp	LoopCmps
LeaveLoopCmps:	add	%rax, %rdx
#else
		mov	(%windowbestlen, %rdx), %rax
		xor	(%prev, %rdx), %rax
		jnz	LeaveLoopCmps
		
		mov	8(%windowbestlen, %rdx), %rax
		xor	8(%prev, %rdx), %rax
		jnz	LeaveLoopCmps8

		mov	16(%windowbestlen, %rdx), %rax
		xor	16(%prev, %rdx), %rax
		jnz	LeaveLoopCmps16
				
		add	$24, %rdx
		jnz	LoopCmps
		jmp	LenMaximum
#	if 0
/*
 * This three-liner is tantalizingly simple, but bsf is a slow instruction,
 * and the complicated alternative down below is quite a bit faster. Sad...
 */

LeaveLoopCmps:	bsf	%rax, %rax /* find the first non-zero bit */
		shrl	$3, %eax /* divide by 8 to get the byte */
		add	%rax, %rdx
#	else
LeaveLoopCmps16:
		add	$8, %rdx
LeaveLoopCmps8:
		add	$8, %rdx
LeaveLoopCmps:	testl   $0xFFFFFFFF, %eax /* Check the first 4 bytes */
		jnz     Check16
		add     $4, %rdx
		shr     $32, %rax
Check16:        testw   $0xFFFF, %ax
		jnz     LenLower
		add	$2, %rdx
		shrl	$16, %eax
LenLower:	subb	$1, %al
		adc	$0, %rdx
#	endif
#endif

/* Calculate the length of the match. If it is longer than MAX_MATCH,	*/
/* then automatically accept it as the best possible match and leave.	*/

		lea	(%prev, %rdx), %rax
		sub	%scan, %rax
		cmpl	$MAX_MATCH, %eax
		jge	LenMaximum

/* If the length of the match is not longer than the best match we	*/
/* have so far, then forget it and return to the lookup loop.		*/

		cmpl	%bestlend, %eax
		jg	LongerMatch
		mov	_windowbestlen, %windowbestlen
		mov	dsPrev, %prev
		movl	_chainlenwmask, %edx
		jmp	LookupLoop

/*         s->match_start = cur_match;					*/
/*         best_len = len;						*/
/*         if (len >= nice_match) break;				*/
/*         scan_end = *(ushf*)(scan+best_len-1);			*/

LongerMatch:
		movl	%eax, %bestlend
		movl	%curmatchd, dsMatchStart
		cmpl	%nicematch, %eax
		jge	LeaveNow

		lea	(%window, %bestlen), %windowbestlen
		mov	%windowbestlen, _windowbestlen

		movzwl	-1(%scan, %rax), %scanend
		mov	dsPrev, %prev
		movl	_chainlenwmask, %chainlenwmask
		jmp	LookupLoop

/* Accept the current string, with the maximum possible length.		*/

LenMaximum:
		movl	$MAX_MATCH, %bestlend
		movl	%curmatchd, dsMatchStart

/* if ((uInt)best_len <= s->lookahead) return (uInt)best_len;		*/
/* return s->lookahead;							*/

LeaveNow:
		movl	dsLookahead, %eax
		cmpl	%eax, %bestlend
		cmovngl	%bestlend, %eax
LookaheadRet:

/* Restore the registers and return from whence we came.			*/

	mov	save_rsi, %rsi
	mov	save_rbx, %rbx
	mov	save_r12, %r12
	mov	save_r13, %r13
	mov	save_r14, %r14
	mov	save_r15, %r15

	ret

match_init:	ret

This is a patched version of zlib, modified to use
Pentium-Pro-optimized assembly code in the deflation algorithm. The
files changed/added by this patch are:

README.686
match.S

The speedup that this patch provides varies, depending on whether the
compiler used to build the original version of zlib falls afoul of the
PPro's speed traps. My own tests show a speedup of around 10-20% at
the default compression level, and 20-30% using -9, against a version
compiled using gcc 2.7.2.3. Your mileage may vary.

Note that this code has been tailored for the PPro/PII in particular,
and will not perform particuarly well on a Pentium.

If you are using an assembler other than GNU as, you will have to
translate match.S to use your assembler's syntax. (Have fun.)

Brian Raiter
breadbox@muppetlabs.com
April, 1998


Added for zlib 1.1.3:

The patches come from
http://www.muppetlabs.com/~breadbox/software/assembly.html

To compile zlib with this asm file, copy match.S to the zlib directory
then do:

CFLAGS="-O3 -DASMV" ./configure
make OBJA=match.o


Update:

I've been ignoring these assembly routines for years, believing that
gcc's generated code had caught up with it sometime around gcc 2.95
and the major rearchitecting of the Pentium 4. However, I recently
learned that, despite what I believed, this code still has some life
in it. On the Pentium 4 and AMD64 chips, it continues to run about 8%
faster than the code produced by gcc 4.1.

In acknowledgement of its continuing usefulness, I've altered the
license to match that of the rest of zlib. Share and Enjoy!

Brian Raiter
breadbox@muppetlabs.com
April, 2007

 * The input function is invoked: len = infun(how, &buf), where buf is set by
 * infun() to point to the input buffer, and infun() returns the number of
 * available bytes there.  If infun() returns zero, then blast() returns with
 * an input error.  (blast() only asks for input if it needs it.)  inhow is for
 * use by the application to pass an input descriptor to infun(), if desired.
 *
 * If left and in are not NULL and *left is not zero when blast() is called,
 * then the *left bytes are *in are consumed for input before infun() is used.
 * then the *left bytes at *in are consumed for input before infun() is used.
 *
 * The output function is invoked: err = outfun(how, buf, len), where the bytes
 * to be written are buf[0..len-1].  If err is not zero, then blast() returns
 * with an output error.  outfun() is always called with len <= 4096.  outhow
 * is for use by the application to pass an output descriptor to outfun(), if
 * desired.
 *

       InBytes = number of bytes in InBuf
  Out: OutBuf = ptr to user-allocated buffer to contain decompressed data
       BufSize = number of bytes in OutBuf   }
procedure DecompressToUserBuf(const InBuf: Pointer; InBytes: Integer;
  const OutBuf: Pointer; BufSize: Integer);

const
  zlib_version = '1.2.11';
  zlib_version = '1.2.12';

type
  EZlibError = class(Exception);
  ECompressionError = class(EZlibError);
  EDecompressionError = class(EZlibError);

implementation

    public class InfoTests
    {
        #region Info tests
        [Test]
        public void Info_Version()
        {
            Info info = new Info();
            Assert.AreEqual("1.2.11", Info.Version);
            Assert.AreEqual("1.2.12", Info.Version);
            Assert.AreEqual(32, info.SizeOfUInt);
            Assert.AreEqual(32, info.SizeOfULong);
            Assert.AreEqual(32, info.SizeOfPointer);
            Assert.AreEqual(32, info.SizeOfOffset);
        }
        #endregion
    }

/* inftree9.c -- generate Huffman trees for efficient decoding
 * Copyright (C) 1995-2017 Mark Adler
 * Copyright (C) 1995-2022 Mark Adler
 * For conditions of distribution and use, see copyright notice in zlib.h
 */

#include "zutil.h"
#include "inftree9.h"

#define MAXBITS 15

const char inflate9_copyright[] =
   " inflate9 1.2.11 Copyright 1995-2017 Mark Adler ";
   " inflate9 1.2.12 Copyright 1995-2022 Mark Adler ";
/*
  If you use the zlib library in a product, an acknowledgment is welcome
  in the documentation of your product. If for some reason you cannot
  include such an acknowledgment, I would appreciate that you keep this
  copyright string in the executable of your product.
 */

    static const unsigned short lbase[31] = { /* Length codes 257..285 base */
        3, 4, 5, 6, 7, 8, 9, 10, 11, 13, 15, 17,
        19, 23, 27, 31, 35, 43, 51, 59, 67, 83, 99, 115,
        131, 163, 195, 227, 3, 0, 0};
    static const unsigned short lext[31] = { /* Length codes 257..285 extra */
        128, 128, 128, 128, 128, 128, 128, 128, 129, 129, 129, 129,
        130, 130, 130, 130, 131, 131, 131, 131, 132, 132, 132, 132,
        133, 133, 133, 133, 144, 77, 202};
        133, 133, 133, 133, 144, 76, 202};
    static const unsigned short dbase[32] = { /* Distance codes 0..31 base */
        1, 2, 3, 4, 5, 7, 9, 13, 17, 25, 33, 49,
        65, 97, 129, 193, 257, 385, 513, 769, 1025, 1537, 2049, 3073,
        4097, 6145, 8193, 12289, 16385, 24577, 32769, 49153};
    static const unsigned short dext[32] = { /* Distance codes 0..31 extra */
        128, 128, 128, 128, 129, 129, 130, 130, 131, 131, 132, 132,
        133, 133, 134, 134, 135, 135, 136, 136, 137, 137, 138, 138,

/*
 * inffast.S is a hand tuned assembler version of:
 *
 * inffast.c -- fast decoding
 * Copyright (C) 1995-2003 Mark Adler
 * For conditions of distribution and use, see copyright notice in zlib.h
 *
 * Copyright (C) 2003 Chris Anderson <christop@charm.net>
 * Please use the copyright conditions above.
 *
 * This version (Jan-23-2003) of inflate_fast was coded and tested under
 * GNU/Linux on a pentium 3, using the gcc-3.2 compiler distribution.  On that
 * machine, I found that gzip style archives decompressed about 20% faster than
 * the gcc-3.2 -O3 -fomit-frame-pointer compiled version.  Your results will
 * depend on how large of a buffer is used for z_stream.next_in & next_out
 * (8K-32K worked best for my 256K cpu cache) and how much overhead there is in
 * stream processing I/O and crc32/addler32.  In my case, this routine used
 * 70% of the cpu time and crc32 used 20%.
 *
 * I am confident that this version will work in the general case, but I have
 * not tested a wide variety of datasets or a wide variety of platforms.
 *
 * Jan-24-2003 -- Added -DUSE_MMX define for slightly faster inflating.
 * It should be a runtime flag instead of compile time flag...
 *
 * Jan-26-2003 -- Added runtime check for MMX support with cpuid instruction.
 * With -DUSE_MMX, only MMX code is compiled.  With -DNO_MMX, only non-MMX code
 * is compiled.  Without either option, runtime detection is enabled.  Runtime
 * detection should work on all modern cpus and the recomended algorithm (flip
 * ID bit on eflags and then use the cpuid instruction) is used in many
 * multimedia applications.  Tested under win2k with gcc-2.95 and gas-2.12
 * distributed with cygwin3.  Compiling with gcc-2.95 -c inffast.S -o
 * inffast.obj generates a COFF object which can then be linked with MSVC++
 * compiled code.  Tested under FreeBSD 4.7 with gcc-2.95.
 *
 * Jan-28-2003 -- Tested Athlon XP... MMX mode is slower than no MMX (and
 * slower than compiler generated code).  Adjusted cpuid check to use the MMX
 * code only for Pentiums < P4 until I have more data on the P4.  Speed
 * improvment is only about 15% on the Athlon when compared with code generated
 * with MSVC++.  Not sure yet, but I think the P4 will also be slower using the
 * MMX mode because many of it's x86 ALU instructions execute in .5 cycles and
 * have less latency than MMX ops.  Added code to buffer the last 11 bytes of
 * the input stream since the MMX code grabs bits in chunks of 32, which
 * differs from the inffast.c algorithm.  I don't think there would have been
 * read overruns where a page boundary was crossed (a segfault), but there
 * could have been overruns when next_in ends on unaligned memory (unintialized
 * memory read).
 *
 * Mar-13-2003 -- P4 MMX is slightly slower than P4 NO_MMX.  I created a C
 * version of the non-MMX code so that it doesn't depend on zstrm and zstate
 * structure offsets which are hard coded in this file.  This was last tested
 * with zlib-1.2.0 which is currently in beta testing, newer versions of this
 * and inffas86.c can be found at http://www.eetbeetee.com/zlib/ and
 * http://www.charm.net/~christop/zlib/
 */


/*
 * if you have underscore linking problems (_inflate_fast undefined), try
 * using -DGAS_COFF
 */
#if ! defined( GAS_COFF ) && ! defined( GAS_ELF )

#if defined( WIN32 ) || defined( __CYGWIN__ )
#define GAS_COFF /* windows object format */
#else
#define GAS_ELF
#endif

#endif /* ! GAS_COFF && ! GAS_ELF */


#if defined( GAS_COFF )

/* coff externals have underscores */
#define inflate_fast _inflate_fast
#define inflate_fast_use_mmx _inflate_fast_use_mmx

#endif /* GAS_COFF */


.file "inffast.S"

.globl inflate_fast

.text
.align 4,0
.L_invalid_literal_length_code_msg:
.string "invalid literal/length code"

.align 4,0
.L_invalid_distance_code_msg:
.string "invalid distance code"

.align 4,0
.L_invalid_distance_too_far_msg:
.string "invalid distance too far back"

#if ! defined( NO_MMX )
.align 4,0
.L_mask: /* mask[N] = ( 1 << N ) - 1 */
.long 0
.long 1
.long 3
.long 7
.long 15
.long 31
.long 63
.long 127
.long 255
.long 511
.long 1023
.long 2047
.long 4095
.long 8191
.long 16383
.long 32767
.long 65535
.long 131071
.long 262143
.long 524287
.long 1048575
.long 2097151
.long 4194303
.long 8388607
.long 16777215
.long 33554431
.long 67108863
.long 134217727
.long 268435455
.long 536870911
.long 1073741823
.long 2147483647
.long 4294967295
#endif /* NO_MMX */

.text

/*
 * struct z_stream offsets, in zlib.h
 */
#define next_in_strm   0   /* strm->next_in */
#define avail_in_strm  4   /* strm->avail_in */
#define next_out_strm  12  /* strm->next_out */
#define avail_out_strm 16  /* strm->avail_out */
#define msg_strm       24  /* strm->msg */
#define state_strm     28  /* strm->state */

/*
 * struct inflate_state offsets, in inflate.h
 */
#define mode_state     0   /* state->mode */
#define wsize_state    32  /* state->wsize */
#define write_state    40  /* state->write */
#define window_state   44  /* state->window */
#define hold_state     48  /* state->hold */
#define bits_state     52  /* state->bits */
#define lencode_state  68  /* state->lencode */
#define distcode_state 72  /* state->distcode */
#define lenbits_state  76  /* state->lenbits */
#define distbits_state 80  /* state->distbits */

/*
 * inflate_fast's activation record
 */
#define local_var_size 64 /* how much local space for vars */
#define strm_sp        88 /* first arg: z_stream * (local_var_size + 24) */
#define start_sp       92 /* second arg: unsigned int (local_var_size + 28) */

/*
 * offsets for local vars on stack
 */
#define out            60  /* unsigned char* */
#define window         56  /* unsigned char* */
#define wsize          52  /* unsigned int */
#define write          48  /* unsigned int */
#define in             44  /* unsigned char* */
#define beg            40  /* unsigned char* */
#define buf            28  /* char[ 12 ] */
#define len            24  /* unsigned int */
#define last           20  /* unsigned char* */
#define end            16  /* unsigned char* */
#define dcode          12  /* code* */
#define lcode           8  /* code* */
#define dmask           4  /* unsigned int */
#define lmask           0  /* unsigned int */

/*
 * typedef enum inflate_mode consts, in inflate.h
 */
#define INFLATE_MODE_TYPE 11  /* state->mode flags enum-ed in inflate.h */
#define INFLATE_MODE_BAD  26


#if ! defined( USE_MMX ) && ! defined( NO_MMX )

#define RUN_TIME_MMX

#define CHECK_MMX    1
#define DO_USE_MMX   2
#define DONT_USE_MMX 3

.globl inflate_fast_use_mmx

.data

.align 4,0
inflate_fast_use_mmx: /* integer flag for run time control 1=check,2=mmx,3=no */
.long CHECK_MMX

#if defined( GAS_ELF )
/* elf info */
.type   inflate_fast_use_mmx,@object
.size   inflate_fast_use_mmx,4
#endif

#endif /* RUN_TIME_MMX */

#if defined( GAS_COFF )
/* coff info: scl 2 = extern, type 32 = function */
.def inflate_fast; .scl 2; .type 32; .endef
#endif

.text

.align 32,0x90
inflate_fast:
        pushl   %edi
        pushl   %esi
        pushl   %ebp
        pushl   %ebx
        pushf   /* save eflags (strm_sp, state_sp assumes this is 32 bits) */
        subl    $local_var_size, %esp
        cld

#define strm_r  %esi
#define state_r %edi

        movl    strm_sp(%esp), strm_r
        movl    state_strm(strm_r), state_r

        /* in = strm->next_in;
         * out = strm->next_out;
         * last = in + strm->avail_in - 11;
         * beg = out - (start - strm->avail_out);
         * end = out + (strm->avail_out - 257);
         */
        movl    avail_in_strm(strm_r), %edx
        movl    next_in_strm(strm_r), %eax

        addl    %eax, %edx      /* avail_in += next_in */
        subl    $11, %edx       /* avail_in -= 11 */

        movl    %eax, in(%esp)
        movl    %edx, last(%esp)

        movl    start_sp(%esp), %ebp
        movl    avail_out_strm(strm_r), %ecx
        movl    next_out_strm(strm_r), %ebx

        subl    %ecx, %ebp      /* start -= avail_out */
        negl    %ebp            /* start = -start */
        addl    %ebx, %ebp      /* start += next_out */

        subl    $257, %ecx      /* avail_out -= 257 */
        addl    %ebx, %ecx      /* avail_out += out */

        movl    %ebx, out(%esp)
        movl    %ebp, beg(%esp)
        movl    %ecx, end(%esp)

        /* wsize = state->wsize;
         * write = state->write;
         * window = state->window;
         * hold = state->hold;
         * bits = state->bits;
         * lcode = state->lencode;
         * dcode = state->distcode;
         * lmask = ( 1 << state->lenbits ) - 1;
         * dmask = ( 1 << state->distbits ) - 1;
         */

        movl    lencode_state(state_r), %eax
        movl    distcode_state(state_r), %ecx

        movl    %eax, lcode(%esp)
        movl    %ecx, dcode(%esp)

        movl    $1, %eax
        movl    lenbits_state(state_r), %ecx
        shll    %cl, %eax
        decl    %eax
        movl    %eax, lmask(%esp)

        movl    $1, %eax
        movl    distbits_state(state_r), %ecx
        shll    %cl, %eax
        decl    %eax
        movl    %eax, dmask(%esp)

        movl    wsize_state(state_r), %eax
        movl    write_state(state_r), %ecx
        movl    window_state(state_r), %edx

        movl    %eax, wsize(%esp)
        movl    %ecx, write(%esp)
        movl    %edx, window(%esp)

        movl    hold_state(state_r), %ebp
        movl    bits_state(state_r), %ebx

#undef strm_r
#undef state_r

#define in_r       %esi
#define from_r     %esi
#define out_r      %edi

        movl    in(%esp), in_r
        movl    last(%esp), %ecx
        cmpl    in_r, %ecx
        ja      .L_align_long           /* if in < last */

        addl    $11, %ecx               /* ecx = &in[ avail_in ] */
        subl    in_r, %ecx              /* ecx = avail_in */
        movl    $12, %eax
        subl    %ecx, %eax              /* eax = 12 - avail_in */
        leal    buf(%esp), %edi
        rep     movsb                   /* memcpy( buf, in, avail_in ) */
        movl    %eax, %ecx
        xorl    %eax, %eax
        rep     stosb         /* memset( &buf[ avail_in ], 0, 12 - avail_in ) */
        leal    buf(%esp), in_r         /* in = buf */
        movl    in_r, last(%esp)        /* last = in, do just one iteration */
        jmp     .L_is_aligned

        /* align in_r on long boundary */
.L_align_long:
        testl   $3, in_r
        jz      .L_is_aligned
        xorl    %eax, %eax
        movb    (in_r), %al
        incl    in_r
        movl    %ebx, %ecx
        addl    $8, %ebx
        shll    %cl, %eax
        orl     %eax, %ebp
        jmp     .L_align_long

.L_is_aligned:
        movl    out(%esp), out_r

#if defined( NO_MMX )
        jmp     .L_do_loop
#endif

#if defined( USE_MMX )
        jmp     .L_init_mmx
#endif

/*** Runtime MMX check ***/

#if defined( RUN_TIME_MMX )
.L_check_mmx:
        cmpl    $DO_USE_MMX, inflate_fast_use_mmx
        je      .L_init_mmx
        ja      .L_do_loop /* > 2 */

        pushl   %eax
        pushl   %ebx
        pushl   %ecx
        pushl   %edx
        pushf
        movl    (%esp), %eax      /* copy eflags to eax */
        xorl    $0x200000, (%esp) /* try toggling ID bit of eflags (bit 21)
                                   * to see if cpu supports cpuid...
                                   * ID bit method not supported by NexGen but
                                   * bios may load a cpuid instruction and
                                   * cpuid may be disabled on Cyrix 5-6x86 */
        popf
        pushf
        popl    %edx              /* copy new eflags to edx */
        xorl    %eax, %edx        /* test if ID bit is flipped */
        jz      .L_dont_use_mmx   /* not flipped if zero */
        xorl    %eax, %eax
        cpuid
        cmpl    $0x756e6547, %ebx /* check for GenuineIntel in ebx,ecx,edx */
        jne     .L_dont_use_mmx
        cmpl    $0x6c65746e, %ecx
        jne     .L_dont_use_mmx
        cmpl    $0x49656e69, %edx
        jne     .L_dont_use_mmx
        movl    $1, %eax
        cpuid                     /* get cpu features */
        shrl    $8, %eax
        andl    $15, %eax
        cmpl    $6, %eax          /* check for Pentium family, is 0xf for P4 */
        jne     .L_dont_use_mmx
        testl   $0x800000, %edx   /* test if MMX feature is set (bit 23) */
        jnz     .L_use_mmx
        jmp     .L_dont_use_mmx
.L_use_mmx:
        movl    $DO_USE_MMX, inflate_fast_use_mmx
        jmp     .L_check_mmx_pop
.L_dont_use_mmx:
        movl    $DONT_USE_MMX, inflate_fast_use_mmx
.L_check_mmx_pop:
        popl    %edx
        popl    %ecx
        popl    %ebx
        popl    %eax
        jmp     .L_check_mmx
#endif


/*** Non-MMX code ***/

#if defined ( NO_MMX ) || defined( RUN_TIME_MMX )

#define hold_r     %ebp
#define bits_r     %bl
#define bitslong_r %ebx

.align 32,0x90
.L_while_test:
        /* while (in < last && out < end)
         */
        cmpl    out_r, end(%esp)
        jbe     .L_break_loop           /* if (out >= end) */

        cmpl    in_r, last(%esp)
        jbe     .L_break_loop

.L_do_loop:
        /* regs: %esi = in, %ebp = hold, %bl = bits, %edi = out
         *
         * do {
         *   if (bits < 15) {
         *     hold |= *((unsigned short *)in)++ << bits;
         *     bits += 16
         *   }
         *   this = lcode[hold & lmask]
         */
        cmpb    $15, bits_r
        ja      .L_get_length_code      /* if (15 < bits) */

        xorl    %eax, %eax
        lodsw                           /* al = *(ushort *)in++ */
        movb    bits_r, %cl             /* cl = bits, needs it for shifting */
        addb    $16, bits_r             /* bits += 16 */
        shll    %cl, %eax
        orl     %eax, hold_r            /* hold |= *((ushort *)in)++ << bits */

.L_get_length_code:
        movl    lmask(%esp), %edx       /* edx = lmask */
        movl    lcode(%esp), %ecx       /* ecx = lcode */
        andl    hold_r, %edx            /* edx &= hold */
        movl    (%ecx,%edx,4), %eax     /* eax = lcode[hold & lmask] */

.L_dolen:
        /* regs: %esi = in, %ebp = hold, %bl = bits, %edi = out
         *
         * dolen:
         *    bits -= this.bits;
         *    hold >>= this.bits
         */
        movb    %ah, %cl                /* cl = this.bits */
        subb    %ah, bits_r             /* bits -= this.bits */
        shrl    %cl, hold_r             /* hold >>= this.bits */

        /* check if op is a literal
         * if (op == 0) {
         *    PUP(out) = this.val;
         *  }
         */
        testb   %al, %al
        jnz     .L_test_for_length_base /* if (op != 0) 45.7% */

        shrl    $16, %eax               /* output this.val char */
        stosb
        jmp     .L_while_test

.L_test_for_length_base:
        /* regs: %esi = in, %ebp = hold, %bl = bits, %edi = out, %edx = len
         *
         * else if (op & 16) {
         *   len = this.val
         *   op &= 15
         *   if (op) {
         *     if (op > bits) {
         *       hold |= *((unsigned short *)in)++ << bits;
         *       bits += 16
         *     }
         *     len += hold & mask[op];
         *     bits -= op;
         *     hold >>= op;
         *   }
         */
#define len_r %edx
        movl    %eax, len_r             /* len = this */
        shrl    $16, len_r              /* len = this.val */
        movb    %al, %cl

        testb   $16, %al
        jz      .L_test_for_second_level_length /* if ((op & 16) == 0) 8% */
        andb    $15, %cl                /* op &= 15 */
        jz      .L_save_len             /* if (!op) */
        cmpb    %cl, bits_r
        jae     .L_add_bits_to_len      /* if (op <= bits) */

        movb    %cl, %ch                /* stash op in ch, freeing cl */
        xorl    %eax, %eax
        lodsw                           /* al = *(ushort *)in++ */
        movb    bits_r, %cl             /* cl = bits, needs it for shifting */
        addb    $16, bits_r             /* bits += 16 */
        shll    %cl, %eax
        orl     %eax, hold_r            /* hold |= *((ushort *)in)++ << bits */
        movb    %ch, %cl                /* move op back to ecx */

.L_add_bits_to_len:
        movl    $1, %eax
        shll    %cl, %eax
        decl    %eax
        subb    %cl, bits_r
        andl    hold_r, %eax            /* eax &= hold */
        shrl    %cl, hold_r
        addl    %eax, len_r             /* len += hold & mask[op] */

.L_save_len:
        movl    len_r, len(%esp)        /* save len */
#undef  len_r

.L_decode_distance:
        /* regs: %esi = in, %ebp = hold, %bl = bits, %edi = out, %edx = dist
         *
         *   if (bits < 15) {
         *     hold |= *((unsigned short *)in)++ << bits;
         *     bits += 16
         *   }
         *   this = dcode[hold & dmask];
         * dodist:
         *   bits -= this.bits;
         *   hold >>= this.bits;
         *   op = this.op;
         */

        cmpb    $15, bits_r
        ja      .L_get_distance_code    /* if (15 < bits) */

        xorl    %eax, %eax
        lodsw                           /* al = *(ushort *)in++ */
        movb    bits_r, %cl             /* cl = bits, needs it for shifting */
        addb    $16, bits_r             /* bits += 16 */
        shll    %cl, %eax
        orl     %eax, hold_r            /* hold |= *((ushort *)in)++ << bits */

.L_get_distance_code:
        movl    dmask(%esp), %edx       /* edx = dmask */
        movl    dcode(%esp), %ecx       /* ecx = dcode */
        andl    hold_r, %edx            /* edx &= hold */
        movl    (%ecx,%edx,4), %eax     /* eax = dcode[hold & dmask] */

#define dist_r %edx
.L_dodist:
        movl    %eax, dist_r            /* dist = this */
        shrl    $16, dist_r             /* dist = this.val */
        movb    %ah, %cl
        subb    %ah, bits_r             /* bits -= this.bits */
        shrl    %cl, hold_r             /* hold >>= this.bits */

        /* if (op & 16) {
         *   dist = this.val
         *   op &= 15
         *   if (op > bits) {
         *     hold |= *((unsigned short *)in)++ << bits;
         *     bits += 16
         *   }
         *   dist += hold & mask[op];
         *   bits -= op;
         *   hold >>= op;
         */
        movb    %al, %cl                /* cl = this.op */

        testb   $16, %al                /* if ((op & 16) == 0) */
        jz      .L_test_for_second_level_dist
        andb    $15, %cl                /* op &= 15 */
        jz      .L_check_dist_one
        cmpb    %cl, bits_r
        jae     .L_add_bits_to_dist     /* if (op <= bits) 97.6% */

        movb    %cl, %ch                /* stash op in ch, freeing cl */
        xorl    %eax, %eax
        lodsw                           /* al = *(ushort *)in++ */
        movb    bits_r, %cl             /* cl = bits, needs it for shifting */
        addb    $16, bits_r             /* bits += 16 */
        shll    %cl, %eax
        orl     %eax, hold_r            /* hold |= *((ushort *)in)++ << bits */
        movb    %ch, %cl                /* move op back to ecx */

.L_add_bits_to_dist:
        movl    $1, %eax
        shll    %cl, %eax
        decl    %eax                    /* (1 << op) - 1 */
        subb    %cl, bits_r
        andl    hold_r, %eax            /* eax &= hold */
        shrl    %cl, hold_r
        addl    %eax, dist_r            /* dist += hold & ((1 << op) - 1) */
        jmp     .L_check_window

.L_check_window:
        /* regs: %esi = from, %ebp = hold, %bl = bits, %edi = out, %edx = dist
         *       %ecx = nbytes
         *
         * nbytes = out - beg;
         * if (dist <= nbytes) {
         *   from = out - dist;
         *   do {
         *     PUP(out) = PUP(from);
         *   } while (--len > 0) {
         * }
         */

        movl    in_r, in(%esp)          /* save in so from can use it's reg */
        movl    out_r, %eax
        subl    beg(%esp), %eax         /* nbytes = out - beg */

        cmpl    dist_r, %eax
        jb      .L_clip_window          /* if (dist > nbytes) 4.2% */

        movl    len(%esp), %ecx
        movl    out_r, from_r
        subl    dist_r, from_r          /* from = out - dist */

        subl    $3, %ecx
        movb    (from_r), %al
        movb    %al, (out_r)
        movb    1(from_r), %al
        movb    2(from_r), %dl
        addl    $3, from_r
        movb    %al, 1(out_r)
        movb    %dl, 2(out_r)
        addl    $3, out_r
        rep     movsb

        movl    in(%esp), in_r          /* move in back to %esi, toss from */
        jmp     .L_while_test

.align 16,0x90
.L_check_dist_one:
        cmpl    $1, dist_r
        jne     .L_check_window
        cmpl    out_r, beg(%esp)
        je      .L_check_window

        decl    out_r
        movl    len(%esp), %ecx
        movb    (out_r), %al
        subl    $3, %ecx

        movb    %al, 1(out_r)
        movb    %al, 2(out_r)
        movb    %al, 3(out_r)
        addl    $4, out_r
        rep     stosb

        jmp     .L_while_test

.align 16,0x90
.L_test_for_second_level_length:
        /* else if ((op & 64) == 0) {
         *   this = lcode[this.val + (hold & mask[op])];
         * }
         */
        testb   $64, %al
        jnz     .L_test_for_end_of_block  /* if ((op & 64) != 0) */

        movl    $1, %eax
        shll    %cl, %eax
        decl    %eax
        andl    hold_r, %eax            /* eax &= hold */
        addl    %edx, %eax              /* eax += this.val */
        movl    lcode(%esp), %edx       /* edx = lcode */
        movl    (%edx,%eax,4), %eax     /* eax = lcode[val + (hold&mask[op])] */
        jmp     .L_dolen

.align 16,0x90
.L_test_for_second_level_dist:
        /* else if ((op & 64) == 0) {
         *   this = dcode[this.val + (hold & mask[op])];
         * }
         */
        testb   $64, %al
        jnz     .L_invalid_distance_code  /* if ((op & 64) != 0) */

        movl    $1, %eax
        shll    %cl, %eax
        decl    %eax
        andl    hold_r, %eax            /* eax &= hold */
        addl    %edx, %eax              /* eax += this.val */
        movl    dcode(%esp), %edx       /* edx = dcode */
        movl    (%edx,%eax,4), %eax     /* eax = dcode[val + (hold&mask[op])] */
        jmp     .L_dodist

.align 16,0x90
.L_clip_window:
        /* regs: %esi = from, %ebp = hold, %bl = bits, %edi = out, %edx = dist
         *       %ecx = nbytes
         *
         * else {
         *   if (dist > wsize) {
         *     invalid distance
         *   }
         *   from = window;
         *   nbytes = dist - nbytes;
         *   if (write == 0) {
         *     from += wsize - nbytes;
         */
#define nbytes_r %ecx
        movl    %eax, nbytes_r
        movl    wsize(%esp), %eax       /* prepare for dist compare */
        negl    nbytes_r                /* nbytes = -nbytes */
        movl    window(%esp), from_r    /* from = window */

        cmpl    dist_r, %eax
        jb      .L_invalid_distance_too_far /* if (dist > wsize) */

        addl    dist_r, nbytes_r        /* nbytes = dist - nbytes */
        cmpl    $0, write(%esp)
        jne     .L_wrap_around_window   /* if (write != 0) */

        subl    nbytes_r, %eax
        addl    %eax, from_r            /* from += wsize - nbytes */

        /* regs: %esi = from, %ebp = hold, %bl = bits, %edi = out, %edx = dist
         *       %ecx = nbytes, %eax = len
         *
         *     if (nbytes < len) {
         *       len -= nbytes;
         *       do {
         *         PUP(out) = PUP(from);
         *       } while (--nbytes);
         *       from = out - dist;
         *     }
         *   }
         */
#define len_r %eax
        movl    len(%esp), len_r
        cmpl    nbytes_r, len_r
        jbe     .L_do_copy1             /* if (nbytes >= len) */

        subl    nbytes_r, len_r         /* len -= nbytes */
        rep     movsb
        movl    out_r, from_r
        subl    dist_r, from_r          /* from = out - dist */
        jmp     .L_do_copy1

        cmpl    nbytes_r, len_r
        jbe     .L_do_copy1             /* if (nbytes >= len) */

        subl    nbytes_r, len_r         /* len -= nbytes */
        rep     movsb
        movl    out_r, from_r
        subl    dist_r, from_r          /* from = out - dist */
        jmp     .L_do_copy1

.L_wrap_around_window:
        /* regs: %esi = from, %ebp = hold, %bl = bits, %edi = out, %edx = dist
         *       %ecx = nbytes, %eax = write, %eax = len
         *
         *   else if (write < nbytes) {
         *     from += wsize + write - nbytes;
         *     nbytes -= write;
         *     if (nbytes < len) {
         *       len -= nbytes;
         *       do {
         *         PUP(out) = PUP(from);
         *       } while (--nbytes);
         *       from = window;
         *       nbytes = write;
         *       if (nbytes < len) {
         *         len -= nbytes;
         *         do {
         *           PUP(out) = PUP(from);
         *         } while(--nbytes);
         *         from = out - dist;
         *       }
         *     }
         *   }
         */
#define write_r %eax
        movl    write(%esp), write_r
        cmpl    write_r, nbytes_r
        jbe     .L_contiguous_in_window /* if (write >= nbytes) */

        addl    wsize(%esp), from_r
        addl    write_r, from_r
        subl    nbytes_r, from_r        /* from += wsize + write - nbytes */
        subl    write_r, nbytes_r       /* nbytes -= write */
#undef write_r

        movl    len(%esp), len_r
        cmpl    nbytes_r, len_r
        jbe     .L_do_copy1             /* if (nbytes >= len) */

        subl    nbytes_r, len_r         /* len -= nbytes */
        rep     movsb
        movl    window(%esp), from_r    /* from = window */
        movl    write(%esp), nbytes_r   /* nbytes = write */
        cmpl    nbytes_r, len_r
        jbe     .L_do_copy1             /* if (nbytes >= len) */

        subl    nbytes_r, len_r         /* len -= nbytes */
        rep     movsb
        movl    out_r, from_r
        subl    dist_r, from_r          /* from = out - dist */
        jmp     .L_do_copy1

.L_contiguous_in_window:
        /* regs: %esi = from, %ebp = hold, %bl = bits, %edi = out, %edx = dist
         *       %ecx = nbytes, %eax = write, %eax = len
         *
         *   else {
         *     from += write - nbytes;
         *     if (nbytes < len) {
         *       len -= nbytes;
         *       do {
         *         PUP(out) = PUP(from);
         *       } while (--nbytes);
         *       from = out - dist;
         *     }
         *   }
         */
#define write_r %eax
        addl    write_r, from_r
        subl    nbytes_r, from_r        /* from += write - nbytes */
#undef write_r

        movl    len(%esp), len_r
        cmpl    nbytes_r, len_r
        jbe     .L_do_copy1             /* if (nbytes >= len) */

        subl    nbytes_r, len_r         /* len -= nbytes */
        rep     movsb
        movl    out_r, from_r
        subl    dist_r, from_r          /* from = out - dist */

.L_do_copy1:
        /* regs: %esi = from, %esi = in, %ebp = hold, %bl = bits, %edi = out
         *       %eax = len
         *
         *     while (len > 0) {
         *       PUP(out) = PUP(from);
         *       len--;
         *     }
         *   }
         * } while (in < last && out < end);
         */
#undef nbytes_r
#define in_r %esi
        movl    len_r, %ecx
        rep     movsb

        movl    in(%esp), in_r          /* move in back to %esi, toss from */
        jmp     .L_while_test

#undef len_r
#undef dist_r

#endif /* NO_MMX || RUN_TIME_MMX */


/*** MMX code ***/

#if defined( USE_MMX ) || defined( RUN_TIME_MMX )

.align 32,0x90
.L_init_mmx:
        emms

#undef  bits_r
#undef  bitslong_r
#define bitslong_r %ebp
#define hold_mm    %mm0
        movd    %ebp, hold_mm
        movl    %ebx, bitslong_r

#define used_mm   %mm1
#define dmask2_mm %mm2
#define lmask2_mm %mm3
#define lmask_mm  %mm4
#define dmask_mm  %mm5
#define tmp_mm    %mm6

        movd    lmask(%esp), lmask_mm
        movq    lmask_mm, lmask2_mm
        movd    dmask(%esp), dmask_mm
        movq    dmask_mm, dmask2_mm
        pxor    used_mm, used_mm
        movl    lcode(%esp), %ebx       /* ebx = lcode */
        jmp     .L_do_loop_mmx

.align 32,0x90
.L_while_test_mmx:
        /* while (in < last && out < end)
         */
        cmpl    out_r, end(%esp)
        jbe     .L_break_loop           /* if (out >= end) */

        cmpl    in_r, last(%esp)
        jbe     .L_break_loop

.L_do_loop_mmx:
        psrlq   used_mm, hold_mm        /* hold_mm >>= last bit length */

        cmpl    $32, bitslong_r
        ja      .L_get_length_code_mmx  /* if (32 < bits) */

        movd    bitslong_r, tmp_mm
        movd    (in_r), %mm7
        addl    $4, in_r
        psllq   tmp_mm, %mm7
        addl    $32, bitslong_r
        por     %mm7, hold_mm           /* hold_mm |= *((uint *)in)++ << bits */

.L_get_length_code_mmx:
        pand    hold_mm, lmask_mm
        movd    lmask_mm, %eax
        movq    lmask2_mm, lmask_mm
        movl    (%ebx,%eax,4), %eax     /* eax = lcode[hold & lmask] */

.L_dolen_mmx:
        movzbl  %ah, %ecx               /* ecx = this.bits */
        movd    %ecx, used_mm
        subl    %ecx, bitslong_r        /* bits -= this.bits */

        testb   %al, %al
        jnz     .L_test_for_length_base_mmx /* if (op != 0) 45.7% */

        shrl    $16, %eax               /* output this.val char */
        stosb
        jmp     .L_while_test_mmx

.L_test_for_length_base_mmx:
#define len_r  %edx
        movl    %eax, len_r             /* len = this */
        shrl    $16, len_r              /* len = this.val */

        testb   $16, %al
        jz      .L_test_for_second_level_length_mmx /* if ((op & 16) == 0) 8% */
        andl    $15, %eax               /* op &= 15 */
        jz      .L_decode_distance_mmx  /* if (!op) */

        psrlq   used_mm, hold_mm        /* hold_mm >>= last bit length */
        movd    %eax, used_mm
        movd    hold_mm, %ecx
        subl    %eax, bitslong_r
        andl    .L_mask(,%eax,4), %ecx
        addl    %ecx, len_r             /* len += hold & mask[op] */

.L_decode_distance_mmx:
        psrlq   used_mm, hold_mm        /* hold_mm >>= last bit length */

        cmpl    $32, bitslong_r
        ja      .L_get_dist_code_mmx    /* if (32 < bits) */

        movd    bitslong_r, tmp_mm
        movd    (in_r), %mm7
        addl    $4, in_r
        psllq   tmp_mm, %mm7
        addl    $32, bitslong_r
        por     %mm7, hold_mm           /* hold_mm |= *((uint *)in)++ << bits */

.L_get_dist_code_mmx:
        movl    dcode(%esp), %ebx       /* ebx = dcode */
        pand    hold_mm, dmask_mm
        movd    dmask_mm, %eax
        movq    dmask2_mm, dmask_mm
        movl    (%ebx,%eax,4), %eax     /* eax = dcode[hold & lmask] */

.L_dodist_mmx:
#define dist_r %ebx
        movzbl  %ah, %ecx               /* ecx = this.bits */
        movl    %eax, dist_r
        shrl    $16, dist_r             /* dist  = this.val */
        subl    %ecx, bitslong_r        /* bits -= this.bits */
        movd    %ecx, used_mm

        testb   $16, %al                /* if ((op & 16) == 0) */
        jz      .L_test_for_second_level_dist_mmx
        andl    $15, %eax               /* op &= 15 */
        jz      .L_check_dist_one_mmx

.L_add_bits_to_dist_mmx:
        psrlq   used_mm, hold_mm        /* hold_mm >>= last bit length */
        movd    %eax, used_mm           /* save bit length of current op */
        movd    hold_mm, %ecx           /* get the next bits on input stream */
        subl    %eax, bitslong_r        /* bits -= op bits */
        andl    .L_mask(,%eax,4), %ecx  /* ecx   = hold & mask[op] */
        addl    %ecx, dist_r            /* dist += hold & mask[op] */

.L_check_window_mmx:
        movl    in_r, in(%esp)          /* save in so from can use it's reg */
        movl    out_r, %eax
        subl    beg(%esp), %eax         /* nbytes = out - beg */

        cmpl    dist_r, %eax
        jb      .L_clip_window_mmx      /* if (dist > nbytes) 4.2% */

        movl    len_r, %ecx
        movl    out_r, from_r
        subl    dist_r, from_r          /* from = out - dist */

        subl    $3, %ecx
        movb    (from_r), %al
        movb    %al, (out_r)
        movb    1(from_r), %al
        movb    2(from_r), %dl
        addl    $3, from_r
        movb    %al, 1(out_r)
        movb    %dl, 2(out_r)
        addl    $3, out_r
        rep     movsb

        movl    in(%esp), in_r          /* move in back to %esi, toss from */
        movl    lcode(%esp), %ebx       /* move lcode back to %ebx, toss dist */
        jmp     .L_while_test_mmx

.align 16,0x90
.L_check_dist_one_mmx:
        cmpl    $1, dist_r
        jne     .L_check_window_mmx
        cmpl    out_r, beg(%esp)
        je      .L_check_window_mmx

        decl    out_r
        movl    len_r, %ecx
        movb    (out_r), %al
        subl    $3, %ecx

        movb    %al, 1(out_r)
        movb    %al, 2(out_r)
        movb    %al, 3(out_r)
        addl    $4, out_r
        rep     stosb

        movl    lcode(%esp), %ebx       /* move lcode back to %ebx, toss dist */
        jmp     .L_while_test_mmx

.align 16,0x90
.L_test_for_second_level_length_mmx:
        testb   $64, %al
        jnz     .L_test_for_end_of_block  /* if ((op & 64) != 0) */

        andl    $15, %eax
        psrlq   used_mm, hold_mm        /* hold_mm >>= last bit length */
        movd    hold_mm, %ecx
        andl    .L_mask(,%eax,4), %ecx
        addl    len_r, %ecx
        movl    (%ebx,%ecx,4), %eax     /* eax = lcode[hold & lmask] */
        jmp     .L_dolen_mmx

.align 16,0x90
.L_test_for_second_level_dist_mmx:
        testb   $64, %al
        jnz     .L_invalid_distance_code  /* if ((op & 64) != 0) */

        andl    $15, %eax
        psrlq   used_mm, hold_mm        /* hold_mm >>= last bit length */
        movd    hold_mm, %ecx
        andl    .L_mask(,%eax,4), %ecx
        movl    dcode(%esp), %eax       /* ecx = dcode */
        addl    dist_r, %ecx
        movl    (%eax,%ecx,4), %eax     /* eax = lcode[hold & lmask] */
        jmp     .L_dodist_mmx

.align 16,0x90
.L_clip_window_mmx:
#define nbytes_r %ecx
        movl    %eax, nbytes_r
        movl    wsize(%esp), %eax       /* prepare for dist compare */
        negl    nbytes_r                /* nbytes = -nbytes */
        movl    window(%esp), from_r    /* from = window */

        cmpl    dist_r, %eax
        jb      .L_invalid_distance_too_far /* if (dist > wsize) */

        addl    dist_r, nbytes_r        /* nbytes = dist - nbytes */
        cmpl    $0, write(%esp)
        jne     .L_wrap_around_window_mmx /* if (write != 0) */

        subl    nbytes_r, %eax
        addl    %eax, from_r            /* from += wsize - nbytes */

        cmpl    nbytes_r, len_r
        jbe     .L_do_copy1_mmx         /* if (nbytes >= len) */

        subl    nbytes_r, len_r         /* len -= nbytes */
        rep     movsb
        movl    out_r, from_r
        subl    dist_r, from_r          /* from = out - dist */
        jmp     .L_do_copy1_mmx

        cmpl    nbytes_r, len_r
        jbe     .L_do_copy1_mmx         /* if (nbytes >= len) */

        subl    nbytes_r, len_r         /* len -= nbytes */
        rep     movsb
        movl    out_r, from_r
        subl    dist_r, from_r          /* from = out - dist */
        jmp     .L_do_copy1_mmx

.L_wrap_around_window_mmx:
#define write_r %eax
        movl    write(%esp), write_r
        cmpl    write_r, nbytes_r
        jbe     .L_contiguous_in_window_mmx /* if (write >= nbytes) */

        addl    wsize(%esp), from_r
        addl    write_r, from_r
        subl    nbytes_r, from_r        /* from += wsize + write - nbytes */
        subl    write_r, nbytes_r       /* nbytes -= write */
#undef write_r

        cmpl    nbytes_r, len_r
        jbe     .L_do_copy1_mmx         /* if (nbytes >= len) */

        subl    nbytes_r, len_r         /* len -= nbytes */
        rep     movsb
        movl    window(%esp), from_r    /* from = window */
        movl    write(%esp), nbytes_r   /* nbytes = write */
        cmpl    nbytes_r, len_r
        jbe     .L_do_copy1_mmx         /* if (nbytes >= len) */

        subl    nbytes_r, len_r         /* len -= nbytes */
        rep     movsb
        movl    out_r, from_r
        subl    dist_r, from_r          /* from = out - dist */
        jmp     .L_do_copy1_mmx

.L_contiguous_in_window_mmx:
#define write_r %eax
        addl    write_r, from_r
        subl    nbytes_r, from_r        /* from += write - nbytes */
#undef write_r

        cmpl    nbytes_r, len_r
        jbe     .L_do_copy1_mmx         /* if (nbytes >= len) */

        subl    nbytes_r, len_r         /* len -= nbytes */
        rep     movsb
        movl    out_r, from_r
        subl    dist_r, from_r          /* from = out - dist */

.L_do_copy1_mmx:
#undef nbytes_r
#define in_r %esi
        movl    len_r, %ecx
        rep     movsb

        movl    in(%esp), in_r          /* move in back to %esi, toss from */
        movl    lcode(%esp), %ebx       /* move lcode back to %ebx, toss dist */
        jmp     .L_while_test_mmx

#undef hold_r
#undef bitslong_r

#endif /* USE_MMX || RUN_TIME_MMX */


/*** USE_MMX, NO_MMX, and RUNTIME_MMX from here on ***/

.L_invalid_distance_code:
        /* else {
         *   strm->msg = "invalid distance code";
         *   state->mode = BAD;
         * }
         */
        movl    $.L_invalid_distance_code_msg, %ecx
        movl    $INFLATE_MODE_BAD, %edx
        jmp     .L_update_stream_state

.L_test_for_end_of_block:
        /* else if (op & 32) {
         *   state->mode = TYPE;
         *   break;
         * }
         */
        testb   $32, %al
        jz      .L_invalid_literal_length_code  /* if ((op & 32) == 0) */

        movl    $0, %ecx
        movl    $INFLATE_MODE_TYPE, %edx
        jmp     .L_update_stream_state

.L_invalid_literal_length_code:
        /* else {
         *   strm->msg = "invalid literal/length code";
         *   state->mode = BAD;
         * }
         */
        movl    $.L_invalid_literal_length_code_msg, %ecx
        movl    $INFLATE_MODE_BAD, %edx
        jmp     .L_update_stream_state

.L_invalid_distance_too_far:
        /* strm->msg = "invalid distance too far back";
         * state->mode = BAD;
         */
        movl    in(%esp), in_r          /* from_r has in's reg, put in back */
        movl    $.L_invalid_distance_too_far_msg, %ecx
        movl    $INFLATE_MODE_BAD, %edx
        jmp     .L_update_stream_state

.L_update_stream_state:
        /* set strm->msg = %ecx, strm->state->mode = %edx */
        movl    strm_sp(%esp), %eax
        testl   %ecx, %ecx              /* if (msg != NULL) */
        jz      .L_skip_msg
        movl    %ecx, msg_strm(%eax)    /* strm->msg = msg */
.L_skip_msg:
        movl    state_strm(%eax), %eax  /* state = strm->state */
        movl    %edx, mode_state(%eax)  /* state->mode = edx (BAD | TYPE) */
        jmp     .L_break_loop

.align 32,0x90
.L_break_loop:

/*
 * Regs:
 *
 * bits = %ebp when mmx, and in %ebx when non-mmx
 * hold = %hold_mm when mmx, and in %ebp when non-mmx
 * in   = %esi
 * out  = %edi
 */

#if defined( USE_MMX ) || defined( RUN_TIME_MMX )

#if defined( RUN_TIME_MMX )

        cmpl    $DO_USE_MMX, inflate_fast_use_mmx
        jne     .L_update_next_in

#endif /* RUN_TIME_MMX */

        movl    %ebp, %ebx

.L_update_next_in:

#endif

#define strm_r  %eax
#define state_r %edx

        /* len = bits >> 3;
         * in -= len;
         * bits -= len << 3;
         * hold &= (1U << bits) - 1;
         * state->hold = hold;
         * state->bits = bits;
         * strm->next_in = in;
         * strm->next_out = out;
         */
        movl    strm_sp(%esp), strm_r
        movl    %ebx, %ecx
        movl    state_strm(strm_r), state_r
        shrl    $3, %ecx
        subl    %ecx, in_r
        shll    $3, %ecx
        subl    %ecx, %ebx
        movl    out_r, next_out_strm(strm_r)
        movl    %ebx, bits_state(state_r)
        movl    %ebx, %ecx

        leal    buf(%esp), %ebx
        cmpl    %ebx, last(%esp)
        jne     .L_buf_not_used         /* if buf != last */

        subl    %ebx, in_r              /* in -= buf */
        movl    next_in_strm(strm_r), %ebx
        movl    %ebx, last(%esp)        /* last = strm->next_in */
        addl    %ebx, in_r              /* in += strm->next_in */
        movl    avail_in_strm(strm_r), %ebx
        subl    $11, %ebx
        addl    %ebx, last(%esp)    /* last = &strm->next_in[ avail_in - 11 ] */

.L_buf_not_used:
        movl    in_r, next_in_strm(strm_r)

        movl    $1, %ebx
        shll    %cl, %ebx
        decl    %ebx

#if defined( USE_MMX ) || defined( RUN_TIME_MMX )

#if defined( RUN_TIME_MMX )

        cmpl    $DO_USE_MMX, inflate_fast_use_mmx
        jne     .L_update_hold

#endif /* RUN_TIME_MMX */

        psrlq   used_mm, hold_mm        /* hold_mm >>= last bit length */
        movd    hold_mm, %ebp

        emms

.L_update_hold:

#endif /* USE_MMX || RUN_TIME_MMX */

        andl    %ebx, %ebp
        movl    %ebp, hold_state(state_r)

#define last_r %ebx

        /* strm->avail_in = in < last ? 11 + (last - in) : 11 - (in - last) */
        movl    last(%esp), last_r
        cmpl    in_r, last_r
        jbe     .L_last_is_smaller     /* if (in >= last) */

        subl    in_r, last_r           /* last -= in */
        addl    $11, last_r            /* last += 11 */
        movl    last_r, avail_in_strm(strm_r)
        jmp     .L_fixup_out
.L_last_is_smaller:
        subl    last_r, in_r           /* in -= last */
        negl    in_r                   /* in = -in */
        addl    $11, in_r              /* in += 11 */
        movl    in_r, avail_in_strm(strm_r)

#undef last_r
#define end_r %ebx

.L_fixup_out:
        /* strm->avail_out = out < end ? 257 + (end - out) : 257 - (out - end)*/
        movl    end(%esp), end_r
        cmpl    out_r, end_r
        jbe     .L_end_is_smaller      /* if (out >= end) */

        subl    out_r, end_r           /* end -= out */
        addl    $257, end_r            /* end += 257 */
        movl    end_r, avail_out_strm(strm_r)
        jmp     .L_done
.L_end_is_smaller:
        subl    end_r, out_r           /* out -= end */
        negl    out_r                  /* out = -out */
        addl    $257, out_r            /* out += 257 */
        movl    out_r, avail_out_strm(strm_r)

#undef end_r
#undef strm_r
#undef state_r

.L_done:
        addl    $local_var_size, %esp
        popf
        popl    %ebx
        popl    %ebp
        popl    %esi
        popl    %edi
        ret

#if defined( GAS_ELF )
/* elf info */
.type inflate_fast,@function
.size inflate_fast,.-inflate_fast
#endif

ml64.exe /Flinffasx64 /c /Zi inffasx64.asm
ml64.exe /Flgvmat64   /c /Zi gvmat64.asm

;uInt longest_match_x64(
;    deflate_state *s,
;    IPos cur_match);                             /* current match */

; gvmat64.asm -- Asm portion of the optimized longest_match for 32 bits x86_64
;  (AMD64 on Athlon 64, Opteron, Phenom
;     and Intel EM64T on Pentium 4 with EM64T, Pentium D, Core 2 Duo, Core I5/I7)
; Copyright (C) 1995-2010 Jean-loup Gailly, Brian Raiter and Gilles Vollant.
;
; File written by Gilles Vollant, by converting to assembly the longest_match
;  from Jean-loup Gailly in deflate.c of zLib and infoZip zip.
;
;  and by taking inspiration on asm686 with masm, optimised assembly code
;        from Brian Raiter, written 1998
;
;  This software is provided 'as-is', without any express or implied
;  warranty.  In no event will the authors be held liable for any damages
;  arising from the use of this software.
;
;  Permission is granted to anyone to use this software for any purpose,
;  including commercial applications, and to alter it and redistribute it
;  freely, subject to the following restrictions:
;
;  1. The origin of this software must not be misrepresented; you must not
;     claim that you wrote the original software. If you use this software
;     in a product, an acknowledgment in the product documentation would be
;     appreciated but is not required.
;  2. Altered source versions must be plainly marked as such, and must not be
;     misrepresented as being the original software
;  3. This notice may not be removed or altered from any source distribution.
;
;
;
;         http://www.zlib.net
;         http://www.winimage.com/zLibDll
;         http://www.muppetlabs.com/~breadbox/software/assembly.html
;
; to compile this file for infozip Zip, I use option:
;   ml64.exe /Flgvmat64 /c /Zi /DINFOZIP gvmat64.asm
;
; to compile this file for zLib, I use option:
;   ml64.exe /Flgvmat64 /c /Zi gvmat64.asm
; Be carrefull to adapt zlib1222add below to your version of zLib
;   (if you use a version of zLib before 1.0.4 or after 1.2.2.2, change
;    value of zlib1222add later)
;
; This file compile with Microsoft Macro Assembler (x64) for AMD64
;
;   ml64.exe is given with Visual Studio 2005/2008/2010 and Windows WDK
;
;   (you can get Windows WDK with ml64 for AMD64 from
;      http://www.microsoft.com/whdc/Devtools/wdk/default.mspx for low price)
;


;uInt longest_match(s, cur_match)
;    deflate_state *s;
;    IPos cur_match;                             /* current match */
.code
longest_match PROC


;LocalVarsSize   equ 88
 LocalVarsSize   equ 72

; register used : rax,rbx,rcx,rdx,rsi,rdi,r8,r9,r10,r11,r12
; free register :  r14,r15
; register can be saved : rsp

 chainlenwmask   equ  rsp + 8 - LocalVarsSize    ; high word: current chain len
                                                 ; low word: s->wmask
;window          equ  rsp + xx - LocalVarsSize   ; local copy of s->window ; stored in r10
;windowbestlen   equ  rsp + xx - LocalVarsSize   ; s->window + bestlen , use r10+r11
;scanstart       equ  rsp + xx - LocalVarsSize   ; first two bytes of string ; stored in r12w
;scanend         equ  rsp + xx - LocalVarsSize   ; last two bytes of string use ebx
;scanalign       equ  rsp + xx - LocalVarsSize   ; dword-misalignment of string r13
;bestlen         equ  rsp + xx - LocalVarsSize   ; size of best match so far -> r11d
;scan            equ  rsp + xx - LocalVarsSize   ; ptr to string wanting match -> r9
IFDEF INFOZIP
ELSE
 nicematch       equ  (rsp + 16 - LocalVarsSize) ; a good enough match size
ENDIF

save_rdi        equ  rsp + 24 - LocalVarsSize
save_rsi        equ  rsp + 32 - LocalVarsSize
save_rbx        equ  rsp + 40 - LocalVarsSize
save_rbp        equ  rsp + 48 - LocalVarsSize
save_r12        equ  rsp + 56 - LocalVarsSize
save_r13        equ  rsp + 64 - LocalVarsSize
;save_r14        equ  rsp + 72 - LocalVarsSize
;save_r15        equ  rsp + 80 - LocalVarsSize


; summary of register usage
; scanend     ebx
; scanendw    bx
; chainlenwmask   edx
; curmatch    rsi
; curmatchd   esi
; windowbestlen   r8
; scanalign   r9
; scanalignd  r9d
; window      r10
; bestlen     r11
; bestlend    r11d
; scanstart   r12d
; scanstartw  r12w
; scan        r13
; nicematch   r14d
; limit       r15
; limitd      r15d
; prev        rcx

;  all the +4 offsets are due to the addition of pending_buf_size (in zlib
;  in the deflate_state structure since the asm code was first written
;  (if you compile with zlib 1.0.4 or older, remove the +4).
;  Note : these value are good with a 8 bytes boundary pack structure


    MAX_MATCH           equ     258
    MIN_MATCH           equ     3
    MIN_LOOKAHEAD       equ     (MAX_MATCH+MIN_MATCH+1)


;;; Offsets for fields in the deflate_state structure. These numbers
;;; are calculated from the definition of deflate_state, with the
;;; assumption that the compiler will dword-align the fields. (Thus,
;;; changing the definition of deflate_state could easily cause this
;;; program to crash horribly, without so much as a warning at
;;; compile time. Sigh.)

;  all the +zlib1222add offsets are due to the addition of fields
;  in zlib in the deflate_state structure since the asm code was first written
;  (if you compile with zlib 1.0.4 or older, use "zlib1222add equ (-4)").
;  (if you compile with zlib between 1.0.5 and 1.2.2.1, use "zlib1222add equ 0").
;  if you compile with zlib 1.2.2.2 or later , use "zlib1222add equ 8").


IFDEF INFOZIP

_DATA   SEGMENT
COMM    window_size:DWORD
; WMask ; 7fff
COMM    window:BYTE:010040H
COMM    prev:WORD:08000H
; MatchLen : unused
; PrevMatch : unused
COMM    strstart:DWORD
COMM    match_start:DWORD
; Lookahead : ignore
COMM    prev_length:DWORD ; PrevLen
COMM    max_chain_length:DWORD
COMM    good_match:DWORD
COMM    nice_match:DWORD
prev_ad equ OFFSET prev
window_ad equ OFFSET window
nicematch equ nice_match
_DATA ENDS
WMask equ 07fffh

ELSE

  IFNDEF zlib1222add
    zlib1222add equ 8
  ENDIF
dsWSize         equ 56+zlib1222add+(zlib1222add/2)
dsWMask         equ 64+zlib1222add+(zlib1222add/2)
dsWindow        equ 72+zlib1222add
dsPrev          equ 88+zlib1222add
dsMatchLen      equ 128+zlib1222add
dsPrevMatch     equ 132+zlib1222add
dsStrStart      equ 140+zlib1222add
dsMatchStart    equ 144+zlib1222add
dsLookahead     equ 148+zlib1222add
dsPrevLen       equ 152+zlib1222add
dsMaxChainLen   equ 156+zlib1222add
dsGoodMatch     equ 172+zlib1222add
dsNiceMatch     equ 176+zlib1222add

window_size     equ [ rcx + dsWSize]
WMask           equ [ rcx + dsWMask]
window_ad       equ [ rcx + dsWindow]
prev_ad         equ [ rcx + dsPrev]
strstart        equ [ rcx + dsStrStart]
match_start     equ [ rcx + dsMatchStart]
Lookahead       equ [ rcx + dsLookahead] ; 0ffffffffh on infozip
prev_length     equ [ rcx + dsPrevLen]
max_chain_length equ [ rcx + dsMaxChainLen]
good_match      equ [ rcx + dsGoodMatch]
nice_match      equ [ rcx + dsNiceMatch]
ENDIF

; parameter 1 in r8(deflate state s), param 2 in rdx (cur match)

; see http://weblogs.asp.net/oldnewthing/archive/2004/01/14/58579.aspx and
; http://msdn.microsoft.com/library/en-us/kmarch/hh/kmarch/64bitAMD_8e951dd2-ee77-4728-8702-55ce4b5dd24a.xml.asp
;
; All registers must be preserved across the call, except for
;   rax, rcx, rdx, r8, r9, r10, and r11, which are scratch.



;;; Save registers that the compiler may be using, and adjust esp to
;;; make room for our stack frame.


;;; Retrieve the function arguments. r8d will hold cur_match
;;; throughout the entire function. edx will hold the pointer to the
;;; deflate_state structure during the function's setup (before
;;; entering the main loop.

; parameter 1 in rcx (deflate_state* s), param 2 in edx -> r8 (cur match)

; this clear high 32 bits of r8, which can be garbage in both r8 and rdx

        mov [save_rdi],rdi
        mov [save_rsi],rsi
        mov [save_rbx],rbx
        mov [save_rbp],rbp
IFDEF INFOZIP
        mov r8d,ecx
ELSE
        mov r8d,edx
ENDIF
        mov [save_r12],r12
        mov [save_r13],r13
;        mov [save_r14],r14
;        mov [save_r15],r15


;;; uInt wmask = s->w_mask;
;;; unsigned chain_length = s->max_chain_length;
;;; if (s->prev_length >= s->good_match) {
;;;     chain_length >>= 2;
;;; }

        mov edi, prev_length
        mov esi, good_match
        mov eax, WMask
        mov ebx, max_chain_length
        cmp edi, esi
        jl  LastMatchGood
        shr ebx, 2
LastMatchGood:

;;; chainlen is decremented once beforehand so that the function can
;;; use the sign flag instead of the zero flag for the exit test.
;;; It is then shifted into the high word, to make room for the wmask
;;; value, which it will always accompany.

        dec ebx
        shl ebx, 16
        or  ebx, eax

;;; on zlib only
;;; if ((uInt)nice_match > s->lookahead) nice_match = s->lookahead;

IFDEF INFOZIP
        mov [chainlenwmask], ebx
; on infozip nice_match = [nice_match]
ELSE
        mov eax, nice_match
        mov [chainlenwmask], ebx
        mov r10d, Lookahead
        cmp r10d, eax
        cmovnl r10d, eax
        mov [nicematch],r10d
ENDIF

;;; register Bytef *scan = s->window + s->strstart;
        mov r10, window_ad
        mov ebp, strstart
        lea r13, [r10 + rbp]

;;; Determine how many bytes the scan ptr is off from being
;;; dword-aligned.

         mov r9,r13
         neg r13
         and r13,3

;;; IPos limit = s->strstart > (IPos)MAX_DIST(s) ?
;;;     s->strstart - (IPos)MAX_DIST(s) : NIL;
IFDEF INFOZIP
        mov eax,07efah ; MAX_DIST = (WSIZE-MIN_LOOKAHEAD) (0x8000-(3+8+1))
ELSE
        mov eax, window_size
        sub eax, MIN_LOOKAHEAD
ENDIF
        xor edi,edi
        sub ebp, eax

        mov r11d, prev_length

        cmovng ebp,edi

;;; int best_len = s->prev_length;


;;; Store the sum of s->window + best_len in esi locally, and in esi.

       lea  rsi,[r10+r11]

;;; register ush scan_start = *(ushf*)scan;
;;; register ush scan_end   = *(ushf*)(scan+best_len-1);
;;; Posf *prev = s->prev;

        movzx r12d,word ptr [r9]
        movzx ebx, word ptr [r9 + r11 - 1]

        mov rdi, prev_ad

;;; Jump into the main loop.

        mov edx, [chainlenwmask]

        cmp bx,word ptr [rsi + r8 - 1]
        jz  LookupLoopIsZero

LookupLoop1:
        and r8d, edx

        movzx   r8d, word ptr [rdi + r8*2]
        cmp r8d, ebp
        jbe LeaveNow
        sub edx, 00010000h
        js  LeaveNow

LoopEntry1:
        cmp bx,word ptr [rsi + r8 - 1]
        jz  LookupLoopIsZero

LookupLoop2:
        and r8d, edx

        movzx   r8d, word ptr [rdi + r8*2]
        cmp r8d, ebp
        jbe LeaveNow
        sub edx, 00010000h
        js  LeaveNow

LoopEntry2:
        cmp bx,word ptr [rsi + r8 - 1]
        jz  LookupLoopIsZero

LookupLoop4:
        and r8d, edx

        movzx   r8d, word ptr [rdi + r8*2]
        cmp r8d, ebp
        jbe LeaveNow
        sub edx, 00010000h
        js  LeaveNow

LoopEntry4:

        cmp bx,word ptr [rsi + r8 - 1]
        jnz LookupLoop1
        jmp LookupLoopIsZero


;;; do {
;;;     match = s->window + cur_match;
;;;     if (*(ushf*)(match+best_len-1) != scan_end ||
;;;         *(ushf*)match != scan_start) continue;
;;;     [...]
;;; } while ((cur_match = prev[cur_match & wmask]) > limit
;;;          && --chain_length != 0);
;;;
;;; Here is the inner loop of the function. The function will spend the
;;; majority of its time in this loop, and majority of that time will
;;; be spent in the first ten instructions.
;;;
;;; Within this loop:
;;; ebx = scanend
;;; r8d = curmatch
;;; edx = chainlenwmask - i.e., ((chainlen << 16) | wmask)
;;; esi = windowbestlen - i.e., (window + bestlen)
;;; edi = prev
;;; ebp = limit

LookupLoop:
        and r8d, edx

        movzx   r8d, word ptr [rdi + r8*2]
        cmp r8d, ebp
        jbe LeaveNow
        sub edx, 00010000h
        js  LeaveNow

LoopEntry:

        cmp bx,word ptr [rsi + r8 - 1]
        jnz LookupLoop1
LookupLoopIsZero:
        cmp     r12w, word ptr [r10 + r8]
        jnz LookupLoop1


;;; Store the current value of chainlen.
        mov [chainlenwmask], edx

;;; Point edi to the string under scrutiny, and esi to the string we
;;; are hoping to match it up with. In actuality, esi and edi are
;;; both pointed (MAX_MATCH_8 - scanalign) bytes ahead, and edx is
;;; initialized to -(MAX_MATCH_8 - scanalign).

        lea rsi,[r8+r10]
        mov rdx, 0fffffffffffffef8h; -(MAX_MATCH_8)
        lea rsi, [rsi + r13 + 0108h] ;MAX_MATCH_8]
        lea rdi, [r9 + r13 + 0108h] ;MAX_MATCH_8]

        prefetcht1 [rsi+rdx]
        prefetcht1 [rdi+rdx]


;;; Test the strings for equality, 8 bytes at a time. At the end,
;;; adjust rdx so that it is offset to the exact byte that mismatched.
;;;
;;; We already know at this point that the first three bytes of the
;;; strings match each other, and they can be safely passed over before
;;; starting the compare loop. So what this code does is skip over 0-3
;;; bytes, as much as necessary in order to dword-align the edi
;;; pointer. (rsi will still be misaligned three times out of four.)
;;;
;;; It should be confessed that this loop usually does not represent
;;; much of the total running time. Replacing it with a more
;;; straightforward "rep cmpsb" would not drastically degrade
;;; performance.


LoopCmps:
        mov rax, [rsi + rdx]
        xor rax, [rdi + rdx]
        jnz LeaveLoopCmps

        mov rax, [rsi + rdx + 8]
        xor rax, [rdi + rdx + 8]
        jnz LeaveLoopCmps8


        mov rax, [rsi + rdx + 8+8]
        xor rax, [rdi + rdx + 8+8]
        jnz LeaveLoopCmps16

        add rdx,8+8+8

        jnz short LoopCmps
        jmp short LenMaximum
LeaveLoopCmps16: add rdx,8
LeaveLoopCmps8: add rdx,8
LeaveLoopCmps:

        test    eax, 0000FFFFh
        jnz LenLower

        test eax,0ffffffffh

        jnz LenLower32

        add rdx,4
        shr rax,32
        or ax,ax
        jnz LenLower

LenLower32:
        shr eax,16
        add rdx,2
LenLower:   sub al, 1
        adc rdx, 0
;;; Calculate the length of the match. If it is longer than MAX_MATCH,
;;; then automatically accept it as the best possible match and leave.

        lea rax, [rdi + rdx]
        sub rax, r9
        cmp eax, MAX_MATCH
        jge LenMaximum

;;; If the length of the match is not longer than the best match we
;;; have so far, then forget it and return to the lookup loop.
;///////////////////////////////////

        cmp eax, r11d
        jg  LongerMatch

        lea rsi,[r10+r11]

        mov rdi, prev_ad
        mov edx, [chainlenwmask]
        jmp LookupLoop

;;;         s->match_start = cur_match;
;;;         best_len = len;
;;;         if (len >= nice_match) break;
;;;         scan_end = *(ushf*)(scan+best_len-1);

LongerMatch:
        mov r11d, eax
        mov match_start, r8d
        cmp eax, [nicematch]
        jge LeaveNow

        lea rsi,[r10+rax]

        movzx   ebx, word ptr [r9 + rax - 1]
        mov rdi, prev_ad
        mov edx, [chainlenwmask]
        jmp LookupLoop

;;; Accept the current string, with the maximum possible length.

LenMaximum:
        mov r11d,MAX_MATCH
        mov match_start, r8d

;;; if ((uInt)best_len <= s->lookahead) return (uInt)best_len;
;;; return s->lookahead;

LeaveNow:
IFDEF INFOZIP
        mov eax,r11d
ELSE
        mov eax, Lookahead
        cmp r11d, eax
        cmovng eax, r11d
ENDIF

;;; Restore the stack and return from whence we came.


        mov rsi,[save_rsi]
        mov rdi,[save_rdi]
        mov rbx,[save_rbx]
        mov rbp,[save_rbp]
        mov r12,[save_r12]
        mov r13,[save_r13]
;        mov r14,[save_r14]
;        mov r15,[save_r15]


        ret 0
; please don't remove this string !
; Your can freely use gvmat64 in any free or commercial app
; but it is far better don't remove the string in the binary!
    db     0dh,0ah,"asm686 with masm, optimised assembly code from Brian Raiter, written 1998, converted to amd 64 by Gilles Vollant 2005",0dh,0ah,0
longest_match   ENDP

match_init PROC
  ret 0
match_init ENDP


END

/* inffas8664.c is a hand tuned assembler version of inffast.c - fast decoding
 * version for AMD64 on Windows using Microsoft C compiler
 *
 * Copyright (C) 1995-2003 Mark Adler
 * For conditions of distribution and use, see copyright notice in zlib.h
 *
 * Copyright (C) 2003 Chris Anderson <christop@charm.net>
 * Please use the copyright conditions above.
 *
 * 2005 - Adaptation to Microsoft C Compiler for AMD64 by Gilles Vollant
 *
 * inffas8664.c call function inffas8664fnc in inffasx64.asm
 *  inffasx64.asm is automatically convert from AMD64 portion of inffas86.c
 *
 * Dec-29-2003 -- I added AMD64 inflate asm support.  This version is also
 * slightly quicker on x86 systems because, instead of using rep movsb to copy
 * data, it uses rep movsw, which moves data in 2-byte chunks instead of single
 * bytes.  I've tested the AMD64 code on a Fedora Core 1 + the x86_64 updates
 * from http://fedora.linux.duke.edu/fc1_x86_64
 * which is running on an Athlon 64 3000+ / Gigabyte GA-K8VT800M system with
 * 1GB ram.  The 64-bit version is about 4% faster than the 32-bit version,
 * when decompressing mozilla-source-1.3.tar.gz.
 *
 * Mar-13-2003 -- Most of this is derived from inffast.S which is derived from
 * the gcc -S output of zlib-1.2.0/inffast.c.  Zlib-1.2.0 is in beta release at
 * the moment.  I have successfully compiled and tested this code with gcc2.96,
 * gcc3.2, icc5.0, msvc6.0.  It is very close to the speed of inffast.S
 * compiled with gcc -DNO_MMX, but inffast.S is still faster on the P3 with MMX
 * enabled.  I will attempt to merge the MMX code into this version.  Newer
 * versions of this and inffast.S can be found at
 * http://www.eetbeetee.com/zlib/ and http://www.charm.net/~christop/zlib/
 *
 */

#include <stdio.h>
#include "zutil.h"
#include "inftrees.h"
#include "inflate.h"
#include "inffast.h"

/* Mark Adler's comments from inffast.c: */

/*
   Decode literal, length, and distance codes and write out the resulting
   literal and match bytes until either not enough input or output is
   available, an end-of-block is encountered, or a data error is encountered.
   When large enough input and output buffers are supplied to inflate(), for
   example, a 16K input buffer and a 64K output buffer, more than 95% of the
   inflate execution time is spent in this routine.

   Entry assumptions:

        state->mode == LEN
        strm->avail_in >= 6
        strm->avail_out >= 258
        start >= strm->avail_out
        state->bits < 8

   On return, state->mode is one of:

        LEN -- ran out of enough output space or enough available input
        TYPE -- reached end of block code, inflate() to interpret next block
        BAD -- error in block data

   Notes:

    - The maximum input bits used by a length/distance pair is 15 bits for the
      length code, 5 bits for the length extra, 15 bits for the distance code,
      and 13 bits for the distance extra.  This totals 48 bits, or six bytes.
      Therefore if strm->avail_in >= 6, then there is enough input to avoid
      checking for available input while decoding.

    - The maximum bytes that a single length/distance pair can output is 258
      bytes, which is the maximum length that can be coded.  inflate_fast()
      requires strm->avail_out >= 258 for each loop to avoid checking for
      output space.
 */



    typedef struct inffast_ar {
/* 64   32                               x86  x86_64 */
/* ar offset                              register */
/*  0    0 */ void *esp;                /* esp save */
/*  8    4 */ void *ebp;                /* ebp save */
/* 16    8 */ unsigned char FAR *in;    /* esi rsi  local strm->next_in */
/* 24   12 */ unsigned char FAR *last;  /*     r9   while in < last */
/* 32   16 */ unsigned char FAR *out;   /* edi rdi  local strm->next_out */
/* 40   20 */ unsigned char FAR *beg;   /*          inflate()'s init next_out */
/* 48   24 */ unsigned char FAR *end;   /*     r10  while out < end */
/* 56   28 */ unsigned char FAR *window;/*          size of window, wsize!=0 */
/* 64   32 */ code const FAR *lcode;    /* ebp rbp  local strm->lencode */
/* 72   36 */ code const FAR *dcode;    /*     r11  local strm->distcode */
/* 80   40 */ size_t /*unsigned long */hold;       /* edx rdx  local strm->hold */
/* 88   44 */ unsigned bits;            /* ebx rbx  local strm->bits */
/* 92   48 */ unsigned wsize;           /*          window size */
/* 96   52 */ unsigned write;           /*          window write index */
/*100   56 */ unsigned lmask;           /*     r12  mask for lcode */
/*104   60 */ unsigned dmask;           /*     r13  mask for dcode */
/*108   64 */ unsigned len;             /*     r14  match length */
/*112   68 */ unsigned dist;            /*     r15  match distance */
/*116   72 */ unsigned status;          /*          set when state chng*/
    } type_ar;
#ifdef ASMINF

void inflate_fast(strm, start)
z_streamp strm;
unsigned start;         /* inflate()'s starting value for strm->avail_out */
{
    struct inflate_state FAR *state;
    type_ar ar;
    void inffas8664fnc(struct inffast_ar * par);



#if (defined( __GNUC__ ) && defined( __amd64__ ) && ! defined( __i386 )) || (defined(_MSC_VER) && defined(_M_AMD64))
#define PAD_AVAIL_IN 6
#define PAD_AVAIL_OUT 258
#else
#define PAD_AVAIL_IN 5
#define PAD_AVAIL_OUT 257
#endif

    /* copy state to local variables */
    state = (struct inflate_state FAR *)strm->state;

    ar.in = strm->next_in;
    ar.last = ar.in + (strm->avail_in - PAD_AVAIL_IN);
    ar.out = strm->next_out;
    ar.beg = ar.out - (start - strm->avail_out);
    ar.end = ar.out + (strm->avail_out - PAD_AVAIL_OUT);
    ar.wsize = state->wsize;
    ar.write = state->wnext;
    ar.window = state->window;
    ar.hold = state->hold;
    ar.bits = state->bits;
    ar.lcode = state->lencode;
    ar.dcode = state->distcode;
    ar.lmask = (1U << state->lenbits) - 1;
    ar.dmask = (1U << state->distbits) - 1;

    /* decode literals and length/distances until end-of-block or not enough
       input data or output space */

    /* align in on 1/2 hold size boundary */
    while (((size_t)(void *)ar.in & (sizeof(ar.hold) / 2 - 1)) != 0) {
        ar.hold += (unsigned long)*ar.in++ << ar.bits;
        ar.bits += 8;
    }

    inffas8664fnc(&ar);

    if (ar.status > 1) {
        if (ar.status == 2)
            strm->msg = "invalid literal/length code";
        else if (ar.status == 3)
            strm->msg = "invalid distance code";
        else
            strm->msg = "invalid distance too far back";
        state->mode = BAD;
    }
    else if ( ar.status == 1 ) {
        state->mode = TYPE;
    }

    /* return unused bytes (on entry, bits < 8, so in won't go too far back) */
    ar.len = ar.bits >> 3;
    ar.in -= ar.len;
    ar.bits -= ar.len << 3;
    ar.hold &= (1U << ar.bits) - 1;

    /* update state and return */
    strm->next_in = ar.in;
    strm->next_out = ar.out;
    strm->avail_in = (unsigned)(ar.in < ar.last ?
                                PAD_AVAIL_IN + (ar.last - ar.in) :
                                PAD_AVAIL_IN - (ar.in - ar.last));
    strm->avail_out = (unsigned)(ar.out < ar.end ?
                                 PAD_AVAIL_OUT + (ar.end - ar.out) :
                                 PAD_AVAIL_OUT - (ar.out - ar.end));
    state->hold = (unsigned long)ar.hold;
    state->bits = ar.bits;
    return;
}

#endif

; inffasx64.asm is a hand tuned assembler version of inffast.c - fast decoding
; version for AMD64 on Windows using Microsoft C compiler
;
; inffasx64.asm is automatically convert from AMD64 portion of inffas86.c
; inffasx64.asm is called by inffas8664.c, which contain more info.


; to compile this file, I use option
;   ml64.exe /Flinffasx64 /c /Zi inffasx64.asm
;   with Microsoft Macro Assembler (x64) for AMD64
;

; This file compile with Microsoft Macro Assembler (x64) for AMD64
;
;   ml64.exe is given with Visual Studio 2005/2008/2010 and Windows WDK
;
;   (you can get Windows WDK with ml64 for AMD64 from
;      http://www.microsoft.com/whdc/Devtools/wdk/default.mspx for low price)
;


.code
inffas8664fnc PROC

; see http://weblogs.asp.net/oldnewthing/archive/2004/01/14/58579.aspx and
; http://msdn.microsoft.com/library/en-us/kmarch/hh/kmarch/64bitAMD_8e951dd2-ee77-4728-8702-55ce4b5dd24a.xml.asp
;
; All registers must be preserved across the call, except for
;   rax, rcx, rdx, r8, r-9, r10, and r11, which are scratch.


	mov [rsp-8],rsi
	mov [rsp-16],rdi
	mov [rsp-24],r12
	mov [rsp-32],r13
	mov [rsp-40],r14
	mov [rsp-48],r15
	mov [rsp-56],rbx

	mov rax,rcx

	mov	[rax+8], rbp       ; /* save regs rbp and rsp */
	mov	[rax], rsp

	mov	rsp, rax          ; /* make rsp point to &ar */

	mov	rsi, [rsp+16]      ; /* rsi  = in */
	mov	rdi, [rsp+32]      ; /* rdi  = out */
	mov	r9, [rsp+24]       ; /* r9   = last */
	mov	r10, [rsp+48]      ; /* r10  = end */
	mov	rbp, [rsp+64]      ; /* rbp  = lcode */
	mov	r11, [rsp+72]      ; /* r11  = dcode */
	mov	rdx, [rsp+80]      ; /* rdx  = hold */
	mov	ebx, [rsp+88]      ; /* ebx  = bits */
	mov	r12d, [rsp+100]    ; /* r12d = lmask */
	mov	r13d, [rsp+104]    ; /* r13d = dmask */
                                          ; /* r14d = len */
                                          ; /* r15d = dist */


	cld
	cmp	r10, rdi
	je	L_one_time           ; /* if only one decode left */
	cmp	r9, rsi

    jne L_do_loop


L_one_time:
	mov	r8, r12           ; /* r8 = lmask */
	cmp	bl, 32
	ja	L_get_length_code_one_time

	lodsd                         ; /* eax = *(uint *)in++ */
	mov	cl, bl            ; /* cl = bits, needs it for shifting */
	add	bl, 32             ; /* bits += 32 */
	shl	rax, cl
	or	rdx, rax          ; /* hold |= *((uint *)in)++ << bits */
	jmp	L_get_length_code_one_time

ALIGN 4
L_while_test:
	cmp	r10, rdi
	jbe	L_break_loop
	cmp	r9, rsi
	jbe	L_break_loop

L_do_loop:
	mov	r8, r12           ; /* r8 = lmask */
	cmp	bl, 32
	ja	L_get_length_code    ; /* if (32 < bits) */

	lodsd                         ; /* eax = *(uint *)in++ */
	mov	cl, bl            ; /* cl = bits, needs it for shifting */
	add	bl, 32             ; /* bits += 32 */
	shl	rax, cl
	or	rdx, rax          ; /* hold |= *((uint *)in)++ << bits */

L_get_length_code:
	and	r8, rdx            ; /* r8 &= hold */
	mov	eax, [rbp+r8*4]  ; /* eax = lcode[hold & lmask] */

	mov	cl, ah            ; /* cl = this.bits */
	sub	bl, ah            ; /* bits -= this.bits */
	shr	rdx, cl           ; /* hold >>= this.bits */

	test	al, al
	jnz	L_test_for_length_base ; /* if (op != 0) 45.7% */

	mov	r8, r12            ; /* r8 = lmask */
	shr	eax, 16            ; /* output this.val char */
	stosb

L_get_length_code_one_time:
	and	r8, rdx            ; /* r8 &= hold */
	mov	eax, [rbp+r8*4] ; /* eax = lcode[hold & lmask] */

L_dolen:
	mov	cl, ah            ; /* cl = this.bits */
	sub	bl, ah            ; /* bits -= this.bits */
	shr	rdx, cl           ; /* hold >>= this.bits */

	test	al, al
	jnz	L_test_for_length_base ; /* if (op != 0) 45.7% */

	shr	eax, 16            ; /* output this.val char */
	stosb
	jmp	L_while_test

ALIGN 4
L_test_for_length_base:
	mov	r14d, eax         ; /* len = this */
	shr	r14d, 16           ; /* len = this.val */
	mov	cl, al

	test	al, 16
	jz	L_test_for_second_level_length ; /* if ((op & 16) == 0) 8% */
	and	cl, 15             ; /* op &= 15 */
	jz	L_decode_distance    ; /* if (!op) */

L_add_bits_to_len:
	sub	bl, cl
	xor	eax, eax
	inc	eax
	shl	eax, cl
	dec	eax
	and	eax, edx          ; /* eax &= hold */
	shr	rdx, cl
	add	r14d, eax         ; /* len += hold & mask[op] */

L_decode_distance:
	mov	r8, r13           ; /* r8 = dmask */
	cmp	bl, 32
	ja	L_get_distance_code  ; /* if (32 < bits) */

	lodsd                         ; /* eax = *(uint *)in++ */
	mov	cl, bl            ; /* cl = bits, needs it for shifting */
	add	bl, 32             ; /* bits += 32 */
	shl	rax, cl
	or	rdx, rax          ; /* hold |= *((uint *)in)++ << bits */

L_get_distance_code:
	and	r8, rdx           ; /* r8 &= hold */
	mov	eax, [r11+r8*4] ; /* eax = dcode[hold & dmask] */

L_dodist:
	mov	r15d, eax         ; /* dist = this */
	shr	r15d, 16           ; /* dist = this.val */
	mov	cl, ah
	sub	bl, ah            ; /* bits -= this.bits */
	shr	rdx, cl           ; /* hold >>= this.bits */
	mov	cl, al            ; /* cl = this.op */

	test	al, 16             ; /* if ((op & 16) == 0) */
	jz	L_test_for_second_level_dist
	and	cl, 15             ; /* op &= 15 */
	jz	L_check_dist_one

L_add_bits_to_dist:
	sub	bl, cl
	xor	eax, eax
	inc	eax
	shl	eax, cl
	dec	eax                 ; /* (1 << op) - 1 */
	and	eax, edx          ; /* eax &= hold */
	shr	rdx, cl
	add	r15d, eax         ; /* dist += hold & ((1 << op) - 1) */

L_check_window:
	mov	r8, rsi           ; /* save in so from can use it's reg */
	mov	rax, rdi
	sub	rax, [rsp+40]      ; /* nbytes = out - beg */

	cmp	eax, r15d
	jb	L_clip_window        ; /* if (dist > nbytes) 4.2% */

	mov	ecx, r14d         ; /* ecx = len */
	mov	rsi, rdi
	sub	rsi, r15          ; /* from = out - dist */

	sar	ecx, 1
	jnc	L_copy_two           ; /* if len % 2 == 0 */

	rep     movsw
	mov	al, [rsi]
	mov	[rdi], al
	inc	rdi

	mov	rsi, r8           ; /* move in back to %rsi, toss from */
	jmp	L_while_test

L_copy_two:
	rep     movsw
	mov	rsi, r8           ; /* move in back to %rsi, toss from */
	jmp	L_while_test

ALIGN 4
L_check_dist_one:
	cmp	r15d, 1            ; /* if dist 1, is a memset */
	jne	L_check_window
	cmp	[rsp+40], rdi      ; /* if out == beg, outside window */
	je	L_check_window

	mov	ecx, r14d         ; /* ecx = len */
	mov	al, [rdi-1]
	mov	ah, al

	sar	ecx, 1
	jnc	L_set_two
	mov	[rdi], al
	inc	rdi

L_set_two:
	rep     stosw
	jmp	L_while_test

ALIGN 4
L_test_for_second_level_length:
	test	al, 64
	jnz	L_test_for_end_of_block ; /* if ((op & 64) != 0) */

	xor	eax, eax
	inc	eax
	shl	eax, cl
	dec	eax
	and	eax, edx         ; /* eax &= hold */
	add	eax, r14d        ; /* eax += len */
	mov	eax, [rbp+rax*4] ; /* eax = lcode[val+(hold&mask[op])]*/
	jmp	L_dolen

ALIGN 4
L_test_for_second_level_dist:
	test	al, 64
	jnz	L_invalid_distance_code ; /* if ((op & 64) != 0) */

	xor	eax, eax
	inc	eax
	shl	eax, cl
	dec	eax
	and	eax, edx         ; /* eax &= hold */
	add	eax, r15d        ; /* eax += dist */
	mov	eax, [r11+rax*4] ; /* eax = dcode[val+(hold&mask[op])]*/
	jmp	L_dodist

ALIGN 4
L_clip_window:
	mov	ecx, eax         ; /* ecx = nbytes */
	mov	eax, [rsp+92]     ; /* eax = wsize, prepare for dist cmp */
	neg	ecx                ; /* nbytes = -nbytes */

	cmp	eax, r15d
	jb	L_invalid_distance_too_far ; /* if (dist > wsize) */

	add	ecx, r15d         ; /* nbytes = dist - nbytes */
	cmp	dword ptr [rsp+96], 0
	jne	L_wrap_around_window ; /* if (write != 0) */

	mov	rsi, [rsp+56]     ; /* from  = window */
	sub	eax, ecx         ; /* eax  -= nbytes */
	add	rsi, rax         ; /* from += wsize - nbytes */

	mov	eax, r14d        ; /* eax = len */
	cmp	r14d, ecx
	jbe	L_do_copy           ; /* if (nbytes >= len) */

	sub	eax, ecx         ; /* eax -= nbytes */
	rep     movsb
	mov	rsi, rdi
	sub	rsi, r15         ; /* from = &out[ -dist ] */
	jmp	L_do_copy

ALIGN 4
L_wrap_around_window:
	mov	eax, [rsp+96]     ; /* eax = write */
	cmp	ecx, eax
	jbe	L_contiguous_in_window ; /* if (write >= nbytes) */

	mov	esi, [rsp+92]     ; /* from  = wsize */
	add	rsi, [rsp+56]     ; /* from += window */
	add	rsi, rax         ; /* from += write */
	sub	rsi, rcx         ; /* from -= nbytes */
	sub	ecx, eax         ; /* nbytes -= write */

	mov	eax, r14d        ; /* eax = len */
	cmp	eax, ecx
	jbe	L_do_copy           ; /* if (nbytes >= len) */

	sub	eax, ecx         ; /* len -= nbytes */
	rep     movsb
	mov	rsi, [rsp+56]     ; /* from = window */
	mov	ecx, [rsp+96]     ; /* nbytes = write */
	cmp	eax, ecx
	jbe	L_do_copy           ; /* if (nbytes >= len) */

	sub	eax, ecx         ; /* len -= nbytes */
	rep     movsb
	mov	rsi, rdi
	sub	rsi, r15         ; /* from = out - dist */
	jmp	L_do_copy

ALIGN 4
L_contiguous_in_window:
	mov	rsi, [rsp+56]     ; /* rsi = window */
	add	rsi, rax
	sub	rsi, rcx         ; /* from += write - nbytes */

	mov	eax, r14d        ; /* eax = len */
	cmp	eax, ecx
	jbe	L_do_copy           ; /* if (nbytes >= len) */

	sub	eax, ecx         ; /* len -= nbytes */
	rep     movsb
	mov	rsi, rdi
	sub	rsi, r15         ; /* from = out - dist */
	jmp	L_do_copy           ; /* if (nbytes >= len) */

ALIGN 4
L_do_copy:
	mov	ecx, eax         ; /* ecx = len */
	rep     movsb

	mov	rsi, r8          ; /* move in back to %esi, toss from */
	jmp	L_while_test

L_test_for_end_of_block:
	test	al, 32
	jz	L_invalid_literal_length_code
	mov	dword ptr [rsp+116], 1
	jmp	L_break_loop_with_status

L_invalid_literal_length_code:
	mov	dword ptr [rsp+116], 2
	jmp	L_break_loop_with_status

L_invalid_distance_code:
	mov	dword ptr [rsp+116], 3
	jmp	L_break_loop_with_status

L_invalid_distance_too_far:
	mov	dword ptr [rsp+116], 4
	jmp	L_break_loop_with_status

L_break_loop:
	mov	dword ptr [rsp+116], 0

L_break_loop_with_status:
; /* put in, out, bits, and hold back into ar and pop esp */
	mov	[rsp+16], rsi     ; /* in */
	mov	[rsp+32], rdi     ; /* out */
	mov	[rsp+88], ebx     ; /* bits */
	mov	[rsp+80], rdx     ; /* hold */

	mov	rax, [rsp]       ; /* restore rbp and rsp */
	mov	rbp, [rsp+8]
	mov	rsp, rax



	mov rsi,[rsp-8]
	mov rdi,[rsp-16]
	mov r12,[rsp-24]
	mov r13,[rsp-32]
	mov r14,[rsp-40]
	mov r15,[rsp-48]
	mov rbx,[rsp-56]

    ret 0
;          :
;          : "m" (ar)
;          : "memory", "%rax", "%rbx", "%rcx", "%rdx", "%rsi", "%rdi",
;            "%r8", "%r9", "%r10", "%r11", "%r12", "%r13", "%r14", "%r15"
;    );

inffas8664fnc 	ENDP
;_TEXT	ENDS
END

Summary
-------
This directory contains ASM implementations of the functions
longest_match() and inflate_fast(), for 64 bits x86 (both AMD64 and Intel EM64t),
for use with Microsoft Macro Assembler (x64) for AMD64 and Microsoft C++ 64 bits.

gvmat64.asm is written by Gilles Vollant (2005), by using Brian Raiter 686/32 bits
   assembly optimized version from Jean-loup Gailly original longest_match function

inffasx64.asm and inffas8664.c were written by Chris Anderson, by optimizing
   original function from Mark Adler

Use instructions
----------------
Assemble the .asm files using MASM and put the object files into the zlib source
directory.  You can also get object files here:

     http://www.winimage.com/zLibDll/zlib124_masm_obj.zip

define ASMV and ASMINF in your project. Include inffas8664.c in your source tree,
and inffasx64.obj and gvmat64.obj as object to link.


Build instructions
------------------
run bld_64.bat with Microsoft Macro Assembler (x64) for AMD64 (ml64.exe)

ml64.exe is given with Visual Studio 2005, Windows 2003 server DDK

You can get Windows 2003 server DDK with ml64 and cl for AMD64 from
  http://www.microsoft.com/whdc/devtools/ddk/default.mspx for low price)

ml /coff /Zi /c /Flmatch686.lst match686.asm
ml /coff /Zi /c /Flinffas32.lst inffas32.asm

;/* inffas32.asm is a hand tuned assembler version of inffast.c -- fast decoding
; *
; * inffas32.asm is derivated from inffas86.c, with translation of assembly code
; *
; * Copyright (C) 1995-2003 Mark Adler
; * For conditions of distribution and use, see copyright notice in zlib.h
; *
; * Copyright (C) 2003 Chris Anderson <christop@charm.net>
; * Please use the copyright conditions above.
; *
; * Mar-13-2003 -- Most of this is derived from inffast.S which is derived from
; * the gcc -S output of zlib-1.2.0/inffast.c.  Zlib-1.2.0 is in beta release at
; * the moment.  I have successfully compiled and tested this code with gcc2.96,
; * gcc3.2, icc5.0, msvc6.0.  It is very close to the speed of inffast.S
; * compiled with gcc -DNO_MMX, but inffast.S is still faster on the P3 with MMX
; * enabled.  I will attempt to merge the MMX code into this version.  Newer
; * versions of this and inffast.S can be found at
; * http://www.eetbeetee.com/zlib/ and http://www.charm.net/~christop/zlib/
; *
; * 2005 : modification by Gilles Vollant
; */
; For Visual C++ 4.x and higher and ML 6.x and higher
;   ml.exe is in directory \MASM611C of Win95 DDK
;   ml.exe is also distributed in http://www.masm32.com/masmdl.htm
;    and in VC++2003 toolkit at http://msdn.microsoft.com/visualc/vctoolkit2003/
;
;
;   compile with command line option
;   ml  /coff /Zi /c /Flinffas32.lst inffas32.asm

;   if you define NO_GZIP (see inflate.h), compile with
;   ml  /coff /Zi /c /Flinffas32.lst /DNO_GUNZIP inffas32.asm


; zlib122sup is 0 fort zlib 1.2.2.1 and lower
; zlib122sup is 8 fort zlib 1.2.2.2 and more (with addition of dmax and head
;        in inflate_state in inflate.h)
zlib1222sup      equ    8


IFDEF GUNZIP
  INFLATE_MODE_TYPE    equ 11
  INFLATE_MODE_BAD     equ 26
ELSE
  IFNDEF NO_GUNZIP
    INFLATE_MODE_TYPE    equ 11
    INFLATE_MODE_BAD     equ 26
  ELSE
    INFLATE_MODE_TYPE    equ 3
    INFLATE_MODE_BAD     equ 17
  ENDIF
ENDIF


; 75 "inffast.S"
;FILE "inffast.S"

;;;GLOBAL _inflate_fast

;;;SECTION .text



	.586p
	.mmx

	name	inflate_fast_x86
	.MODEL	FLAT

_DATA			segment
inflate_fast_use_mmx:
	dd	1


_TEXT			segment



ALIGN 4
	db	'Fast decoding Code from Chris Anderson'
	db	0

ALIGN 4
invalid_literal_length_code_msg:
	db	'invalid literal/length code'
	db	0

ALIGN 4
invalid_distance_code_msg:
	db	'invalid distance code'
	db	0

ALIGN 4
invalid_distance_too_far_msg:
	db	'invalid distance too far back'
	db	0


ALIGN 4
inflate_fast_mask:
dd	0
dd	1
dd	3
dd	7
dd	15
dd	31
dd	63
dd	127
dd	255
dd	511
dd	1023
dd	2047
dd	4095
dd	8191
dd	16383
dd	32767
dd	65535
dd	131071
dd	262143
dd	524287
dd	1048575
dd	2097151
dd	4194303
dd	8388607
dd	16777215
dd	33554431
dd	67108863
dd	134217727
dd	268435455
dd	536870911
dd	1073741823
dd	2147483647
dd	4294967295


mode_state	 equ	0	;/* state->mode	*/
wsize_state	 equ	(32+zlib1222sup)	;/* state->wsize */
write_state	 equ	(36+4+zlib1222sup)	;/* state->write */
window_state	 equ	(40+4+zlib1222sup)	;/* state->window */
hold_state	 equ	(44+4+zlib1222sup)	;/* state->hold	*/
bits_state	 equ	(48+4+zlib1222sup)	;/* state->bits	*/
lencode_state	 equ	(64+4+zlib1222sup)	;/* state->lencode */
distcode_state	 equ	(68+4+zlib1222sup)	;/* state->distcode */
lenbits_state	 equ	(72+4+zlib1222sup)	;/* state->lenbits */
distbits_state	 equ	(76+4+zlib1222sup)	;/* state->distbits */


;;SECTION .text
; 205 "inffast.S"
;GLOBAL	inflate_fast_use_mmx

;SECTION .data


; GLOBAL inflate_fast_use_mmx:object
;.size inflate_fast_use_mmx, 4
; 226 "inffast.S"
;SECTION .text

ALIGN 4
_inflate_fast proc near
.FPO (16, 4, 0, 0, 1, 0)
	push  edi
	push  esi
	push  ebp
	push  ebx
	pushfd
	sub  esp,64
	cld




	mov  esi, [esp+88]
	mov  edi, [esi+28]







	mov  edx, [esi+4]
	mov  eax, [esi+0]

	add  edx,eax
	sub  edx,11

	mov  [esp+44],eax
	mov  [esp+20],edx

	mov  ebp, [esp+92]
	mov  ecx, [esi+16]
	mov  ebx, [esi+12]

	sub  ebp,ecx
	neg  ebp
	add  ebp,ebx

	sub  ecx,257
	add  ecx,ebx

	mov  [esp+60],ebx
	mov  [esp+40],ebp
	mov  [esp+16],ecx
; 285 "inffast.S"
	mov  eax, [edi+lencode_state]
	mov  ecx, [edi+distcode_state]

	mov  [esp+8],eax
	mov  [esp+12],ecx

	mov  eax,1
	mov  ecx, [edi+lenbits_state]
	shl  eax,cl
	dec  eax
	mov  [esp+0],eax

	mov  eax,1
	mov  ecx, [edi+distbits_state]
	shl  eax,cl
	dec  eax
	mov  [esp+4],eax

	mov  eax, [edi+wsize_state]
	mov  ecx, [edi+write_state]
	mov  edx, [edi+window_state]

	mov  [esp+52],eax
	mov  [esp+48],ecx
	mov  [esp+56],edx

	mov  ebp, [edi+hold_state]
	mov  ebx, [edi+bits_state]
; 321 "inffast.S"
	mov  esi, [esp+44]
	mov  ecx, [esp+20]
	cmp  ecx,esi
	ja   L_align_long

	add  ecx,11
	sub  ecx,esi
	mov  eax,12
	sub  eax,ecx
	lea  edi, [esp+28]
	rep movsb
	mov  ecx,eax
	xor  eax,eax
	rep stosb
	lea  esi, [esp+28]
	mov  [esp+20],esi
	jmp  L_is_aligned


L_align_long:
	test  esi,3
	jz   L_is_aligned
	xor  eax,eax
	mov  al, [esi]
	inc  esi
	mov  ecx,ebx
	add  ebx,8
	shl  eax,cl
	or  ebp,eax
	jmp L_align_long

L_is_aligned:
	mov  edi, [esp+60]
; 366 "inffast.S"
L_check_mmx:
	cmp  dword ptr [inflate_fast_use_mmx],2
	je   L_init_mmx
	ja   L_do_loop

	push  eax
	push  ebx
	push  ecx
	push  edx
	pushfd
	mov  eax, [esp]
	xor  dword ptr [esp],0200000h




	popfd
	pushfd
	pop  edx
	xor  edx,eax
	jz   L_dont_use_mmx
	xor  eax,eax
	cpuid
	cmp  ebx,0756e6547h
	jne  L_dont_use_mmx
	cmp  ecx,06c65746eh
	jne  L_dont_use_mmx
	cmp  edx,049656e69h
	jne  L_dont_use_mmx
	mov  eax,1
	cpuid
	shr  eax,8
	and  eax,15
	cmp  eax,6
	jne  L_dont_use_mmx
	test  edx,0800000h
	jnz  L_use_mmx
	jmp  L_dont_use_mmx
L_use_mmx:
	mov  dword ptr [inflate_fast_use_mmx],2
	jmp  L_check_mmx_pop
L_dont_use_mmx:
	mov  dword ptr [inflate_fast_use_mmx],3
L_check_mmx_pop:
	pop  edx
	pop  ecx
	pop  ebx
	pop  eax
	jmp  L_check_mmx
; 426 "inffast.S"
ALIGN 4
L_do_loop:
; 437 "inffast.S"
	cmp  bl,15
	ja   L_get_length_code

	xor  eax,eax
	lodsw
	mov  cl,bl
	add  bl,16
	shl  eax,cl
	or  ebp,eax

L_get_length_code:
	mov  edx, [esp+0]
	mov  ecx, [esp+8]
	and  edx,ebp
	mov  eax, [ecx+edx*4]

L_dolen:






	mov  cl,ah
	sub  bl,ah
	shr  ebp,cl






	test  al,al
	jnz   L_test_for_length_base

	shr  eax,16
	stosb

L_while_test:


	cmp  [esp+16],edi
	jbe  L_break_loop

	cmp  [esp+20],esi
	ja   L_do_loop
	jmp  L_break_loop

L_test_for_length_base:
; 502 "inffast.S"
	mov  edx,eax
	shr  edx,16
	mov  cl,al

	test  al,16
	jz   L_test_for_second_level_length
	and  cl,15
	jz   L_save_len
	cmp  bl,cl
	jae  L_add_bits_to_len

	mov  ch,cl
	xor  eax,eax
	lodsw
	mov  cl,bl
	add  bl,16
	shl  eax,cl
	or  ebp,eax
	mov  cl,ch

L_add_bits_to_len:
	mov  eax,1
	shl  eax,cl
	dec  eax
	sub  bl,cl
	and  eax,ebp
	shr  ebp,cl
	add  edx,eax

L_save_len:
	mov  [esp+24],edx


L_decode_distance:
; 549 "inffast.S"
	cmp  bl,15
	ja   L_get_distance_code

	xor  eax,eax
	lodsw
	mov  cl,bl
	add  bl,16
	shl  eax,cl
	or  ebp,eax

L_get_distance_code:
	mov  edx, [esp+4]
	mov  ecx, [esp+12]
	and  edx,ebp
	mov  eax, [ecx+edx*4]


L_dodist:
	mov  edx,eax
	shr  edx,16
	mov  cl,ah
	sub  bl,ah
	shr  ebp,cl
; 584 "inffast.S"
	mov  cl,al

	test  al,16
	jz  L_test_for_second_level_dist
	and  cl,15
	jz  L_check_dist_one
	cmp  bl,cl
	jae  L_add_bits_to_dist

	mov  ch,cl
	xor  eax,eax
	lodsw
	mov  cl,bl
	add  bl,16
	shl  eax,cl
	or  ebp,eax
	mov  cl,ch

L_add_bits_to_dist:
	mov  eax,1
	shl  eax,cl
	dec  eax
	sub  bl,cl
	and  eax,ebp
	shr  ebp,cl
	add  edx,eax
	jmp  L_check_window

L_check_window:
; 625 "inffast.S"
	mov  [esp+44],esi
	mov  eax,edi
	sub  eax, [esp+40]

	cmp  eax,edx
	jb   L_clip_window

	mov  ecx, [esp+24]
	mov  esi,edi
	sub  esi,edx

	sub  ecx,3
	mov  al, [esi]
	mov  [edi],al
	mov  al, [esi+1]
	mov  dl, [esi+2]
	add  esi,3
	mov  [edi+1],al
	mov  [edi+2],dl
	add  edi,3
	rep movsb

	mov  esi, [esp+44]
	jmp  L_while_test

ALIGN 4
L_check_dist_one:
	cmp  edx,1
	jne  L_check_window
	cmp  [esp+40],edi
	je  L_check_window

	dec  edi
	mov  ecx, [esp+24]
	mov  al, [edi]
	sub  ecx,3

	mov  [edi+1],al
	mov  [edi+2],al
	mov  [edi+3],al
	add  edi,4
	rep stosb

	jmp  L_while_test

ALIGN 4
L_test_for_second_level_length:




	test  al,64
	jnz   L_test_for_end_of_block

	mov  eax,1
	shl  eax,cl
	dec  eax
	and  eax,ebp
	add  eax,edx
	mov  edx, [esp+8]
	mov  eax, [edx+eax*4]
	jmp  L_dolen

ALIGN 4
L_test_for_second_level_dist:




	test  al,64
	jnz   L_invalid_distance_code

	mov  eax,1
	shl  eax,cl
	dec  eax
	and  eax,ebp
	add  eax,edx
	mov  edx, [esp+12]
	mov  eax, [edx+eax*4]
	jmp  L_dodist

ALIGN 4
L_clip_window:
; 721 "inffast.S"
	mov  ecx,eax
	mov  eax, [esp+52]
	neg  ecx
	mov  esi, [esp+56]

	cmp  eax,edx
	jb   L_invalid_distance_too_far

	add  ecx,edx
	cmp  dword ptr [esp+48],0
	jne  L_wrap_around_window

	sub  eax,ecx
	add  esi,eax
; 749 "inffast.S"
	mov  eax, [esp+24]
	cmp  eax,ecx
	jbe  L_do_copy1

	sub  eax,ecx
	rep movsb
	mov  esi,edi
	sub  esi,edx
	jmp  L_do_copy1

	cmp  eax,ecx
	jbe  L_do_copy1

	sub  eax,ecx
	rep movsb
	mov  esi,edi
	sub  esi,edx
	jmp  L_do_copy1

L_wrap_around_window:
; 793 "inffast.S"
	mov  eax, [esp+48]
	cmp  ecx,eax
	jbe  L_contiguous_in_window

	add  esi, [esp+52]
	add  esi,eax
	sub  esi,ecx
	sub  ecx,eax


	mov  eax, [esp+24]
	cmp  eax,ecx
	jbe  L_do_copy1

	sub  eax,ecx
	rep movsb
	mov  esi, [esp+56]
	mov  ecx, [esp+48]
	cmp  eax,ecx
	jbe  L_do_copy1

	sub  eax,ecx
	rep movsb
	mov  esi,edi
	sub  esi,edx
	jmp  L_do_copy1

L_contiguous_in_window:
; 836 "inffast.S"
	add  esi,eax
	sub  esi,ecx


	mov  eax, [esp+24]
	cmp  eax,ecx
	jbe  L_do_copy1

	sub  eax,ecx
	rep movsb
	mov  esi,edi
	sub  esi,edx

L_do_copy1:
; 862 "inffast.S"
	mov  ecx,eax
	rep movsb

	mov  esi, [esp+44]
	jmp  L_while_test
; 878 "inffast.S"
ALIGN 4
L_init_mmx:
	emms





	movd mm0,ebp
	mov  ebp,ebx
; 896 "inffast.S"
	movd mm4,dword ptr [esp+0]
	movq mm3,mm4
	movd mm5,dword ptr [esp+4]
	movq mm2,mm5
	pxor mm1,mm1
	mov  ebx, [esp+8]
	jmp  L_do_loop_mmx

ALIGN 4
L_do_loop_mmx:
	psrlq mm0,mm1

	cmp  ebp,32
	ja  L_get_length_code_mmx

	movd mm6,ebp
	movd mm7,dword ptr [esi]
	add  esi,4
	psllq mm7,mm6
	add  ebp,32
	por mm0,mm7

L_get_length_code_mmx:
	pand mm4,mm0
	movd eax,mm4
	movq mm4,mm3
	mov  eax, [ebx+eax*4]

L_dolen_mmx:
	movzx  ecx,ah
	movd mm1,ecx
	sub  ebp,ecx

	test  al,al
	jnz L_test_for_length_base_mmx

	shr  eax,16
	stosb

L_while_test_mmx:


	cmp  [esp+16],edi
	jbe L_break_loop

	cmp  [esp+20],esi
	ja L_do_loop_mmx
	jmp L_break_loop

L_test_for_length_base_mmx:

	mov  edx,eax
	shr  edx,16

	test  al,16
	jz  L_test_for_second_level_length_mmx
	and  eax,15
	jz L_decode_distance_mmx

	psrlq mm0,mm1
	movd mm1,eax
	movd ecx,mm0
	sub  ebp,eax
	and  ecx, [inflate_fast_mask+eax*4]
	add  edx,ecx

L_decode_distance_mmx:
	psrlq mm0,mm1

	cmp  ebp,32
	ja L_get_dist_code_mmx

	movd mm6,ebp
	movd mm7,dword ptr [esi]
	add  esi,4
	psllq mm7,mm6
	add  ebp,32
	por mm0,mm7

L_get_dist_code_mmx:
	mov  ebx, [esp+12]
	pand mm5,mm0
	movd eax,mm5
	movq mm5,mm2
	mov  eax, [ebx+eax*4]

L_dodist_mmx:

	movzx  ecx,ah
	mov  ebx,eax
	shr  ebx,16
	sub  ebp,ecx
	movd mm1,ecx

	test  al,16
	jz L_test_for_second_level_dist_mmx
	and  eax,15
	jz L_check_dist_one_mmx

L_add_bits_to_dist_mmx:
	psrlq mm0,mm1
	movd mm1,eax
	movd ecx,mm0
	sub  ebp,eax
	and  ecx, [inflate_fast_mask+eax*4]
	add  ebx,ecx

L_check_window_mmx:
	mov  [esp+44],esi
	mov  eax,edi
	sub  eax, [esp+40]

	cmp  eax,ebx
	jb L_clip_window_mmx

	mov  ecx,edx
	mov  esi,edi
	sub  esi,ebx

	sub  ecx,3
	mov  al, [esi]
	mov  [edi],al
	mov  al, [esi+1]
	mov  dl, [esi+2]
	add  esi,3
	mov  [edi+1],al
	mov  [edi+2],dl
	add  edi,3
	rep movsb

	mov  esi, [esp+44]
	mov  ebx, [esp+8]
	jmp  L_while_test_mmx

ALIGN 4
L_check_dist_one_mmx:
	cmp  ebx,1
	jne  L_check_window_mmx
	cmp  [esp+40],edi
	je   L_check_window_mmx

	dec  edi
	mov  ecx,edx
	mov  al, [edi]
	sub  ecx,3

	mov  [edi+1],al
	mov  [edi+2],al
	mov  [edi+3],al
	add  edi,4
	rep stosb

	mov  ebx, [esp+8]
	jmp  L_while_test_mmx

ALIGN 4
L_test_for_second_level_length_mmx:
	test  al,64
	jnz L_test_for_end_of_block

	and  eax,15
	psrlq mm0,mm1
	movd ecx,mm0
	and  ecx, [inflate_fast_mask+eax*4]
	add  ecx,edx
	mov  eax, [ebx+ecx*4]
	jmp L_dolen_mmx

ALIGN 4
L_test_for_second_level_dist_mmx:
	test  al,64
	jnz L_invalid_distance_code

	and  eax,15
	psrlq mm0,mm1
	movd ecx,mm0
	and  ecx, [inflate_fast_mask+eax*4]
	mov  eax, [esp+12]
	add  ecx,ebx
	mov  eax, [eax+ecx*4]
	jmp  L_dodist_mmx

ALIGN 4
L_clip_window_mmx:

	mov  ecx,eax
	mov  eax, [esp+52]
	neg  ecx
	mov  esi, [esp+56]

	cmp  eax,ebx
	jb  L_invalid_distance_too_far

	add  ecx,ebx
	cmp  dword ptr [esp+48],0
	jne  L_wrap_around_window_mmx

	sub  eax,ecx
	add  esi,eax

	cmp  edx,ecx
	jbe  L_do_copy1_mmx

	sub  edx,ecx
	rep movsb
	mov  esi,edi
	sub  esi,ebx
	jmp  L_do_copy1_mmx

	cmp  edx,ecx
	jbe  L_do_copy1_mmx

	sub  edx,ecx
	rep movsb
	mov  esi,edi
	sub  esi,ebx
	jmp  L_do_copy1_mmx

L_wrap_around_window_mmx:

	mov  eax, [esp+48]
	cmp  ecx,eax
	jbe  L_contiguous_in_window_mmx

	add  esi, [esp+52]
	add  esi,eax
	sub  esi,ecx
	sub  ecx,eax


	cmp  edx,ecx
	jbe  L_do_copy1_mmx

	sub  edx,ecx
	rep movsb
	mov  esi, [esp+56]
	mov  ecx, [esp+48]
	cmp  edx,ecx
	jbe  L_do_copy1_mmx

	sub  edx,ecx
	rep movsb
	mov  esi,edi
	sub  esi,ebx
	jmp  L_do_copy1_mmx

L_contiguous_in_window_mmx:

	add  esi,eax
	sub  esi,ecx


	cmp  edx,ecx
	jbe  L_do_copy1_mmx

	sub  edx,ecx
	rep movsb
	mov  esi,edi
	sub  esi,ebx

L_do_copy1_mmx:


	mov  ecx,edx
	rep movsb

	mov  esi, [esp+44]
	mov  ebx, [esp+8]
	jmp  L_while_test_mmx
; 1174 "inffast.S"
L_invalid_distance_code:





	mov  ecx, invalid_distance_code_msg
	mov  edx,INFLATE_MODE_BAD
	jmp  L_update_stream_state

L_test_for_end_of_block:





	test  al,32
	jz  L_invalid_literal_length_code

	mov  ecx,0
	mov  edx,INFLATE_MODE_TYPE
	jmp  L_update_stream_state

L_invalid_literal_length_code:





	mov  ecx, invalid_literal_length_code_msg
	mov  edx,INFLATE_MODE_BAD
	jmp  L_update_stream_state

L_invalid_distance_too_far:



	mov  esi, [esp+44]
	mov  ecx, invalid_distance_too_far_msg
	mov  edx,INFLATE_MODE_BAD
	jmp  L_update_stream_state

L_update_stream_state:

	mov  eax, [esp+88]
	test  ecx,ecx
	jz  L_skip_msg
	mov  [eax+24],ecx
L_skip_msg:
	mov  eax, [eax+28]
	mov  [eax+mode_state],edx
	jmp  L_break_loop

ALIGN 4
L_break_loop:
; 1243 "inffast.S"
	cmp  dword ptr [inflate_fast_use_mmx],2
	jne  L_update_next_in



	mov  ebx,ebp

L_update_next_in:
; 1266 "inffast.S"
	mov  eax, [esp+88]
	mov  ecx,ebx
	mov  edx, [eax+28]
	shr  ecx,3
	sub  esi,ecx
	shl  ecx,3
	sub  ebx,ecx
	mov  [eax+12],edi
	mov  [edx+bits_state],ebx
	mov  ecx,ebx

	lea  ebx, [esp+28]
	cmp  [esp+20],ebx
	jne  L_buf_not_used

	sub  esi,ebx
	mov  ebx, [eax+0]
	mov  [esp+20],ebx
	add  esi,ebx
	mov  ebx, [eax+4]
	sub  ebx,11
	add  [esp+20],ebx

L_buf_not_used:
	mov  [eax+0],esi

	mov  ebx,1
	shl  ebx,cl
	dec  ebx





	cmp  dword ptr [inflate_fast_use_mmx],2
	jne  L_update_hold



	psrlq mm0,mm1
	movd ebp,mm0

	emms

L_update_hold:



	and  ebp,ebx
	mov  [edx+hold_state],ebp




	mov  ebx, [esp+20]
	cmp  ebx,esi
	jbe  L_last_is_smaller

	sub  ebx,esi
	add  ebx,11
	mov  [eax+4],ebx
	jmp  L_fixup_out
L_last_is_smaller:
	sub  esi,ebx
	neg  esi
	add  esi,11
	mov  [eax+4],esi




L_fixup_out:

	mov  ebx, [esp+16]
	cmp  ebx,edi
	jbe  L_end_is_smaller

	sub  ebx,edi
	add  ebx,257
	mov  [eax+16],ebx
	jmp  L_done
L_end_is_smaller:
	sub  edi,ebx
	neg  edi
	add  edi,257
	mov  [eax+16],edi





L_done:
	add  esp,64
	popfd
	pop  ebx
	pop  ebp
	pop  esi
	pop  edi
	ret
_inflate_fast endp

_TEXT	ends
end

; match686.asm -- Asm portion of the optimized longest_match for 32 bits x86
; Copyright (C) 1995-1996 Jean-loup Gailly, Brian Raiter and Gilles Vollant.
; File written by Gilles Vollant, by converting match686.S from Brian Raiter
; for MASM. This is as assembly version of longest_match
;  from Jean-loup Gailly in deflate.c
;
;         http://www.zlib.net
;         http://www.winimage.com/zLibDll
;         http://www.muppetlabs.com/~breadbox/software/assembly.html
;
; For Visual C++ 4.x and higher and ML 6.x and higher
;   ml.exe is distributed in
;  http://www.microsoft.com/downloads/details.aspx?FamilyID=7a1c9da0-0510-44a2-b042-7ef370530c64
;
; this file contain two implementation of longest_match
;
;  this longest_match was written by Brian raiter (1998), optimized for Pentium Pro
;   (and the faster known version of match_init on modern Core 2 Duo and AMD Phenom)
;
;  for using an assembly version of longest_match, you need define ASMV in project
;
;    compile the asm file running
;           ml /coff /Zi /c /Flmatch686.lst match686.asm
;    and do not include match686.obj in your project
;
; note: contrib of zLib 1.2.3 and earlier contained both a deprecated version for
;  Pentium (prior Pentium Pro) and this version for Pentium Pro and modern processor
;  with autoselect (with cpu detection code)
;  if you want support the old pentium optimization, you can still use these version
;
; this file is not optimized for old pentium, but it compatible with all x86 32 bits
; processor (starting 80386)
;
;
; see below : zlib1222add must be adjuster if you use a zlib version < 1.2.2.2

;uInt longest_match(s, cur_match)
;    deflate_state *s;
;    IPos cur_match;                             /* current match */

    NbStack         equ     76
    cur_match       equ     dword ptr[esp+NbStack-0]
    str_s           equ     dword ptr[esp+NbStack-4]
; 5 dword on top (ret,ebp,esi,edi,ebx)
    adrret          equ     dword ptr[esp+NbStack-8]
    pushebp         equ     dword ptr[esp+NbStack-12]
    pushedi         equ     dword ptr[esp+NbStack-16]
    pushesi         equ     dword ptr[esp+NbStack-20]
    pushebx         equ     dword ptr[esp+NbStack-24]

    chain_length    equ     dword ptr [esp+NbStack-28]
    limit           equ     dword ptr [esp+NbStack-32]
    best_len        equ     dword ptr [esp+NbStack-36]
    window          equ     dword ptr [esp+NbStack-40]
    prev            equ     dword ptr [esp+NbStack-44]
    scan_start      equ      word ptr [esp+NbStack-48]
    wmask           equ     dword ptr [esp+NbStack-52]
    match_start_ptr equ     dword ptr [esp+NbStack-56]
    nice_match      equ     dword ptr [esp+NbStack-60]
    scan            equ     dword ptr [esp+NbStack-64]

    windowlen       equ     dword ptr [esp+NbStack-68]
    match_start     equ     dword ptr [esp+NbStack-72]
    strend          equ     dword ptr [esp+NbStack-76]
    NbStackAdd      equ     (NbStack-24)

    .386p

    name    gvmatch
    .MODEL  FLAT



;  all the +zlib1222add offsets are due to the addition of fields
;  in zlib in the deflate_state structure since the asm code was first written
;  (if you compile with zlib 1.0.4 or older, use "zlib1222add equ (-4)").
;  (if you compile with zlib between 1.0.5 and 1.2.2.1, use "zlib1222add equ 0").
;  if you compile with zlib 1.2.2.2 or later , use "zlib1222add equ 8").

    zlib1222add         equ     8

;  Note : these value are good with a 8 bytes boundary pack structure
    dep_chain_length    equ     74h+zlib1222add
    dep_window          equ     30h+zlib1222add
    dep_strstart        equ     64h+zlib1222add
    dep_prev_length     equ     70h+zlib1222add
    dep_nice_match      equ     88h+zlib1222add
    dep_w_size          equ     24h+zlib1222add
    dep_prev            equ     38h+zlib1222add
    dep_w_mask          equ     2ch+zlib1222add
    dep_good_match      equ     84h+zlib1222add
    dep_match_start     equ     68h+zlib1222add
    dep_lookahead       equ     6ch+zlib1222add


_TEXT                   segment

IFDEF NOUNDERLINE
            public  longest_match
            public  match_init
ELSE
            public  _longest_match
            public  _match_init
ENDIF

    MAX_MATCH           equ     258
    MIN_MATCH           equ     3
    MIN_LOOKAHEAD       equ     (MAX_MATCH+MIN_MATCH+1)



MAX_MATCH       equ     258
MIN_MATCH       equ     3
MIN_LOOKAHEAD   equ     (MAX_MATCH + MIN_MATCH + 1)
MAX_MATCH_8_     equ     ((MAX_MATCH + 7) AND 0FFF0h)


;;; stack frame offsets

chainlenwmask   equ  esp + 0    ; high word: current chain len
                    ; low word: s->wmask
window      equ  esp + 4    ; local copy of s->window
windowbestlen   equ  esp + 8    ; s->window + bestlen
scanstart   equ  esp + 16   ; first two bytes of string
scanend     equ  esp + 12   ; last two bytes of string
scanalign   equ  esp + 20   ; dword-misalignment of string
nicematch   equ  esp + 24   ; a good enough match size
bestlen     equ  esp + 28   ; size of best match so far
scan        equ  esp + 32   ; ptr to string wanting match

LocalVarsSize   equ 36
;   saved ebx   byte esp + 36
;   saved edi   byte esp + 40
;   saved esi   byte esp + 44
;   saved ebp   byte esp + 48
;   return address  byte esp + 52
deflatestate    equ  esp + 56   ; the function arguments
curmatch    equ  esp + 60

;;; Offsets for fields in the deflate_state structure. These numbers
;;; are calculated from the definition of deflate_state, with the
;;; assumption that the compiler will dword-align the fields. (Thus,
;;; changing the definition of deflate_state could easily cause this
;;; program to crash horribly, without so much as a warning at
;;; compile time. Sigh.)

dsWSize     equ 36+zlib1222add
dsWMask     equ 44+zlib1222add
dsWindow    equ 48+zlib1222add
dsPrev      equ 56+zlib1222add
dsMatchLen  equ 88+zlib1222add
dsPrevMatch equ 92+zlib1222add
dsStrStart  equ 100+zlib1222add
dsMatchStart    equ 104+zlib1222add
dsLookahead equ 108+zlib1222add
dsPrevLen   equ 112+zlib1222add
dsMaxChainLen   equ 116+zlib1222add
dsGoodMatch equ 132+zlib1222add
dsNiceMatch equ 136+zlib1222add


;;; match686.asm -- Pentium-Pro-optimized version of longest_match()
;;; Written for zlib 1.1.2
;;; Copyright (C) 1998 Brian Raiter <breadbox@muppetlabs.com>
;;; You can look at http://www.muppetlabs.com/~breadbox/software/assembly.html
;;;
;;
;;  This software is provided 'as-is', without any express or implied
;;  warranty.  In no event will the authors be held liable for any damages
;;  arising from the use of this software.
;;
;;  Permission is granted to anyone to use this software for any purpose,
;;  including commercial applications, and to alter it and redistribute it
;;  freely, subject to the following restrictions:
;;
;;  1. The origin of this software must not be misrepresented; you must not
;;     claim that you wrote the original software. If you use this software
;;     in a product, an acknowledgment in the product documentation would be
;;     appreciated but is not required.
;;  2. Altered source versions must be plainly marked as such, and must not be
;;     misrepresented as being the original software
;;  3. This notice may not be removed or altered from any source distribution.
;;

;GLOBAL _longest_match, _match_init


;SECTION    .text

;;; uInt longest_match(deflate_state *deflatestate, IPos curmatch)

;_longest_match:
    IFDEF NOUNDERLINE
    longest_match       proc near
    ELSE
    _longest_match      proc near
    ENDIF
.FPO (9, 4, 0, 0, 1, 0)

;;; Save registers that the compiler may be using, and adjust esp to
;;; make room for our stack frame.

        push    ebp
        push    edi
        push    esi
        push    ebx
        sub esp, LocalVarsSize

;;; Retrieve the function arguments. ecx will hold cur_match
;;; throughout the entire function. edx will hold the pointer to the
;;; deflate_state structure during the function's setup (before
;;; entering the main loop.

        mov edx, [deflatestate]
        mov ecx, [curmatch]

;;; uInt wmask = s->w_mask;
;;; unsigned chain_length = s->max_chain_length;
;;; if (s->prev_length >= s->good_match) {
;;;     chain_length >>= 2;
;;; }

        mov eax, [edx + dsPrevLen]
        mov ebx, [edx + dsGoodMatch]
        cmp eax, ebx
        mov eax, [edx + dsWMask]
        mov ebx, [edx + dsMaxChainLen]
        jl  LastMatchGood
        shr ebx, 2
LastMatchGood:

;;; chainlen is decremented once beforehand so that the function can
;;; use the sign flag instead of the zero flag for the exit test.
;;; It is then shifted into the high word, to make room for the wmask
;;; value, which it will always accompany.

        dec ebx
        shl ebx, 16
        or  ebx, eax
        mov [chainlenwmask], ebx

;;; if ((uInt)nice_match > s->lookahead) nice_match = s->lookahead;

        mov eax, [edx + dsNiceMatch]
        mov ebx, [edx + dsLookahead]
        cmp ebx, eax
        jl  LookaheadLess
        mov ebx, eax
LookaheadLess:  mov [nicematch], ebx

;;; register Bytef *scan = s->window + s->strstart;

        mov esi, [edx + dsWindow]
        mov [window], esi
        mov ebp, [edx + dsStrStart]
        lea edi, [esi + ebp]
        mov [scan], edi

;;; Determine how many bytes the scan ptr is off from being
;;; dword-aligned.

        mov eax, edi
        neg eax
        and eax, 3
        mov [scanalign], eax

;;; IPos limit = s->strstart > (IPos)MAX_DIST(s) ?
;;;     s->strstart - (IPos)MAX_DIST(s) : NIL;

        mov eax, [edx + dsWSize]
        sub eax, MIN_LOOKAHEAD
        sub ebp, eax
        jg  LimitPositive
        xor ebp, ebp
LimitPositive:

;;; int best_len = s->prev_length;

        mov eax, [edx + dsPrevLen]
        mov [bestlen], eax

;;; Store the sum of s->window + best_len in esi locally, and in esi.

        add esi, eax
        mov [windowbestlen], esi

;;; register ush scan_start = *(ushf*)scan;
;;; register ush scan_end   = *(ushf*)(scan+best_len-1);
;;; Posf *prev = s->prev;

        movzx   ebx, word ptr [edi]
        mov [scanstart], ebx
        movzx   ebx, word ptr [edi + eax - 1]
        mov [scanend], ebx
        mov edi, [edx + dsPrev]

;;; Jump into the main loop.

        mov edx, [chainlenwmask]
        jmp short LoopEntry

align 4

;;; do {
;;;     match = s->window + cur_match;
;;;     if (*(ushf*)(match+best_len-1) != scan_end ||
;;;         *(ushf*)match != scan_start) continue;
;;;     [...]
;;; } while ((cur_match = prev[cur_match & wmask]) > limit
;;;          && --chain_length != 0);
;;;
;;; Here is the inner loop of the function. The function will spend the
;;; majority of its time in this loop, and majority of that time will
;;; be spent in the first ten instructions.
;;;
;;; Within this loop:
;;; ebx = scanend
;;; ecx = curmatch
;;; edx = chainlenwmask - i.e., ((chainlen << 16) | wmask)
;;; esi = windowbestlen - i.e., (window + bestlen)
;;; edi = prev
;;; ebp = limit

LookupLoop:
        and ecx, edx
        movzx   ecx, word ptr [edi + ecx*2]
        cmp ecx, ebp
        jbe LeaveNow
        sub edx, 00010000h
        js  LeaveNow
LoopEntry:  movzx   eax, word ptr [esi + ecx - 1]
        cmp eax, ebx
        jnz LookupLoop
        mov eax, [window]
        movzx   eax, word ptr [eax + ecx]
        cmp eax, [scanstart]
        jnz LookupLoop

;;; Store the current value of chainlen.

        mov [chainlenwmask], edx

;;; Point edi to the string under scrutiny, and esi to the string we
;;; are hoping to match it up with. In actuality, esi and edi are
;;; both pointed (MAX_MATCH_8 - scanalign) bytes ahead, and edx is
;;; initialized to -(MAX_MATCH_8 - scanalign).

        mov esi, [window]
        mov edi, [scan]
        add esi, ecx
        mov eax, [scanalign]
        mov edx, 0fffffef8h; -(MAX_MATCH_8)
        lea edi, [edi + eax + 0108h] ;MAX_MATCH_8]
        lea esi, [esi + eax + 0108h] ;MAX_MATCH_8]

;;; Test the strings for equality, 8 bytes at a time. At the end,
;;; adjust edx so that it is offset to the exact byte that mismatched.
;;;
;;; We already know at this point that the first three bytes of the
;;; strings match each other, and they can be safely passed over before
;;; starting the compare loop. So what this code does is skip over 0-3
;;; bytes, as much as necessary in order to dword-align the edi
;;; pointer. (esi will still be misaligned three times out of four.)
;;;
;;; It should be confessed that this loop usually does not represent
;;; much of the total running time. Replacing it with a more
;;; straightforward "rep cmpsb" would not drastically degrade
;;; performance.

LoopCmps:
        mov eax, [esi + edx]
        xor eax, [edi + edx]
        jnz LeaveLoopCmps
        mov eax, [esi + edx + 4]
        xor eax, [edi + edx + 4]
        jnz LeaveLoopCmps4
        add edx, 8
        jnz LoopCmps
        jmp short LenMaximum
LeaveLoopCmps4: add edx, 4
LeaveLoopCmps:  test    eax, 0000FFFFh
        jnz LenLower
        add edx,  2
        shr eax, 16
LenLower:   sub al, 1
        adc edx, 0

;;; Calculate the length of the match. If it is longer than MAX_MATCH,
;;; then automatically accept it as the best possible match and leave.

        lea eax, [edi + edx]
        mov edi, [scan]
        sub eax, edi
        cmp eax, MAX_MATCH
        jge LenMaximum

;;; If the length of the match is not longer than the best match we
;;; have so far, then forget it and return to the lookup loop.

        mov edx, [deflatestate]
        mov ebx, [bestlen]
        cmp eax, ebx
        jg  LongerMatch
        mov esi, [windowbestlen]
        mov edi, [edx + dsPrev]
        mov ebx, [scanend]
        mov edx, [chainlenwmask]
        jmp LookupLoop

;;;         s->match_start = cur_match;
;;;         best_len = len;
;;;         if (len >= nice_match) break;
;;;         scan_end = *(ushf*)(scan+best_len-1);

LongerMatch:    mov ebx, [nicematch]
        mov [bestlen], eax
        mov [edx + dsMatchStart], ecx
        cmp eax, ebx
        jge LeaveNow
        mov esi, [window]
        add esi, eax
        mov [windowbestlen], esi
        movzx   ebx, word ptr [edi + eax - 1]
        mov edi, [edx + dsPrev]
        mov [scanend], ebx
        mov edx, [chainlenwmask]
        jmp LookupLoop

;;; Accept the current string, with the maximum possible length.

LenMaximum: mov edx, [deflatestate]
        mov dword ptr [bestlen], MAX_MATCH
        mov [edx + dsMatchStart], ecx

;;; if ((uInt)best_len <= s->lookahead) return (uInt)best_len;
;;; return s->lookahead;

LeaveNow:
        mov edx, [deflatestate]
        mov ebx, [bestlen]
        mov eax, [edx + dsLookahead]
        cmp ebx, eax
        jg  LookaheadRet
        mov eax, ebx
LookaheadRet:

;;; Restore the stack and return from whence we came.

        add esp, LocalVarsSize
        pop ebx
        pop esi
        pop edi
        pop ebp

        ret
; please don't remove this string !
; Your can freely use match686 in any free or commercial app if you don't remove the string in the binary!
    db     0dh,0ah,"asm686 with masm, optimised assembly code from Brian Raiter, written 1998",0dh,0ah


    IFDEF NOUNDERLINE
    longest_match       endp
    ELSE
    _longest_match      endp
    ENDIF

    IFDEF NOUNDERLINE
    match_init      proc near
                    ret
    match_init      endp
    ELSE
    _match_init     proc near
                    ret
    _match_init     endp
    ENDIF


_TEXT   ends
end