diff options
Diffstat (limited to '3rdparty/zstd/contrib')
120 files changed, 16166 insertions, 0 deletions
diff --git a/3rdparty/zstd/contrib/VS2005/README.md b/3rdparty/zstd/contrib/VS2005/README.md new file mode 100644 index 00000000000..ec1ef68a066 --- /dev/null +++ b/3rdparty/zstd/contrib/VS2005/README.md @@ -0,0 +1,3 @@ +## Project Support Notice + +The VS2005 Project directory has been moved to the contrib directory in order to indicate that it will no longer be supported. diff --git a/3rdparty/zstd/contrib/VS2005/fullbench/fullbench.vcproj b/3rdparty/zstd/contrib/VS2005/fullbench/fullbench.vcproj new file mode 100644 index 00000000000..98f8593769a --- /dev/null +++ b/3rdparty/zstd/contrib/VS2005/fullbench/fullbench.vcproj @@ -0,0 +1,440 @@ +<?xml version="1.0" encoding="Windows-1252"?> +<VisualStudioProject + ProjectType="Visual C++" + Version="8.00" + Name="fullbench" + ProjectGUID="{CC8F1D1B-BA2F-43E3-A71F-FA415D81AAD3}" + RootNamespace="fullbench" + Keyword="Win32Proj" + > + <Platforms> + <Platform + Name="Win32" + /> + <Platform + Name="x64" + /> + </Platforms> + <ToolFiles> + </ToolFiles> + <Configurations> + <Configuration + Name="Debug|Win32" + OutputDirectory="$(SolutionDir)bin\$(PlatformName)\$(ConfigurationName)" + IntermediateDirectory="$(PlatformName)\$(ConfigurationName)" + ConfigurationType="1" + CharacterSet="2" + > + <Tool + Name="VCPreBuildEventTool" + /> + <Tool + Name="VCCustomBuildTool" + /> + <Tool + Name="VCXMLDataGeneratorTool" + /> + <Tool + Name="VCWebServiceProxyGeneratorTool" + /> + <Tool + Name="VCMIDLTool" + /> + <Tool + Name="VCCLCompilerTool" + Optimization="0" + AdditionalIncludeDirectories="$(SolutionDir)..\..\lib;$(SolutionDir)..\..\lib\common;$(SolutionDir)..\..\lib\legacy;$(SolutionDir)..\..\programs" + PreprocessorDefinitions="WIN32;_DEBUG;_CONSOLE" + MinimalRebuild="true" + BasicRuntimeChecks="3" + RuntimeLibrary="3" + UsePrecompiledHeader="0" + WarningLevel="4" + WarnAsError="true" + DebugInformationFormat="4" + /> + <Tool + Name="VCManagedResourceCompilerTool" + /> + <Tool + Name="VCResourceCompilerTool" + /> + <Tool + Name="VCPreLinkEventTool" + /> + <Tool + Name="VCLinkerTool" + LinkIncremental="2" + GenerateDebugInformation="true" + SubSystem="1" + TargetMachine="1" + /> + <Tool + Name="VCALinkTool" + /> + <Tool + Name="VCManifestTool" + /> + <Tool + Name="VCXDCMakeTool" + /> + <Tool + Name="VCBscMakeTool" + /> + <Tool + Name="VCFxCopTool" + /> + <Tool + Name="VCAppVerifierTool" + /> + <Tool + Name="VCPostBuildEventTool" + /> + </Configuration> + <Configuration + Name="Release|Win32" + OutputDirectory="$(SolutionDir)bin\$(PlatformName)\$(ConfigurationName)" + IntermediateDirectory="$(PlatformName)\$(ConfigurationName)" + ConfigurationType="1" + CharacterSet="2" + WholeProgramOptimization="1" + > + <Tool + Name="VCPreBuildEventTool" + /> + <Tool + Name="VCCustomBuildTool" + /> + <Tool + Name="VCXMLDataGeneratorTool" + /> + <Tool + Name="VCWebServiceProxyGeneratorTool" + /> + <Tool + Name="VCMIDLTool" + /> + <Tool + Name="VCCLCompilerTool" + Optimization="2" + EnableIntrinsicFunctions="true" + OmitFramePointers="true" + AdditionalIncludeDirectories="$(SolutionDir)..\..\lib;$(SolutionDir)..\..\lib\common;$(SolutionDir)..\..\lib\legacy;$(SolutionDir)..\..\programs" + PreprocessorDefinitions="WIN32;NDEBUG;_CONSOLE" + RuntimeLibrary="0" + EnableFunctionLevelLinking="true" + UsePrecompiledHeader="0" + WarningLevel="4" + DebugInformationFormat="3" + /> + <Tool + Name="VCManagedResourceCompilerTool" + /> + <Tool + Name="VCResourceCompilerTool" + /> + <Tool + Name="VCPreLinkEventTool" + /> + <Tool + Name="VCLinkerTool" + LinkIncremental="1" + GenerateDebugInformation="true" + SubSystem="1" + OptimizeReferences="2" + EnableCOMDATFolding="2" + TargetMachine="1" + /> + <Tool + Name="VCALinkTool" + /> + <Tool + Name="VCManifestTool" + /> + <Tool + Name="VCXDCMakeTool" + /> + <Tool + Name="VCBscMakeTool" + /> + <Tool + Name="VCFxCopTool" + /> + <Tool + Name="VCAppVerifierTool" + /> + <Tool + Name="VCPostBuildEventTool" + /> + </Configuration> + <Configuration + Name="Debug|x64" + OutputDirectory="$(SolutionDir)bin\$(PlatformName)\$(ConfigurationName)" + IntermediateDirectory="$(PlatformName)\$(ConfigurationName)" + ConfigurationType="1" + CharacterSet="2" + > + <Tool + Name="VCPreBuildEventTool" + /> + <Tool + Name="VCCustomBuildTool" + /> + <Tool + Name="VCXMLDataGeneratorTool" + /> + <Tool + Name="VCWebServiceProxyGeneratorTool" + /> + <Tool + Name="VCMIDLTool" + TargetEnvironment="3" + /> + <Tool + Name="VCCLCompilerTool" + Optimization="0" + AdditionalIncludeDirectories="$(SolutionDir)..\..\lib;$(SolutionDir)..\..\lib\common;$(SolutionDir)..\..\lib\legacy;$(SolutionDir)..\..\programs" + PreprocessorDefinitions="WIN32;_DEBUG;_CONSOLE" + MinimalRebuild="true" + BasicRuntimeChecks="3" + RuntimeLibrary="3" + UsePrecompiledHeader="0" + WarningLevel="4" + WarnAsError="true" + DebugInformationFormat="3" + /> + <Tool + Name="VCManagedResourceCompilerTool" + /> + <Tool + Name="VCResourceCompilerTool" + /> + <Tool + Name="VCPreLinkEventTool" + /> + <Tool + Name="VCLinkerTool" + LinkIncremental="2" + GenerateDebugInformation="true" + SubSystem="1" + TargetMachine="17" + /> + <Tool + Name="VCALinkTool" + /> + <Tool + Name="VCManifestTool" + /> + <Tool + Name="VCXDCMakeTool" + /> + <Tool + Name="VCBscMakeTool" + /> + <Tool + Name="VCFxCopTool" + /> + <Tool + Name="VCAppVerifierTool" + /> + <Tool + Name="VCPostBuildEventTool" + /> + </Configuration> + <Configuration + Name="Release|x64" + OutputDirectory="$(SolutionDir)bin\$(PlatformName)\$(ConfigurationName)" + IntermediateDirectory="$(PlatformName)\$(ConfigurationName)" + ConfigurationType="1" + CharacterSet="2" + WholeProgramOptimization="1" + > + <Tool + Name="VCPreBuildEventTool" + /> + <Tool + Name="VCCustomBuildTool" + /> + <Tool + Name="VCXMLDataGeneratorTool" + /> + <Tool + Name="VCWebServiceProxyGeneratorTool" + /> + <Tool + Name="VCMIDLTool" + TargetEnvironment="3" + /> + <Tool + Name="VCCLCompilerTool" + Optimization="2" + EnableIntrinsicFunctions="true" + OmitFramePointers="true" + AdditionalIncludeDirectories="$(SolutionDir)..\..\lib;$(SolutionDir)..\..\lib\common;$(SolutionDir)..\..\lib\legacy;$(SolutionDir)..\..\programs" + PreprocessorDefinitions="WIN32;NDEBUG;_CONSOLE" + RuntimeLibrary="0" + EnableFunctionLevelLinking="true" + UsePrecompiledHeader="0" + WarningLevel="4" + DebugInformationFormat="3" + /> + <Tool + Name="VCManagedResourceCompilerTool" + /> + <Tool + Name="VCResourceCompilerTool" + /> + <Tool + Name="VCPreLinkEventTool" + /> + <Tool + Name="VCLinkerTool" + LinkIncremental="1" + GenerateDebugInformation="true" + SubSystem="1" + OptimizeReferences="2" + EnableCOMDATFolding="2" + TargetMachine="17" + /> + <Tool + Name="VCALinkTool" + /> + <Tool + Name="VCManifestTool" + /> + <Tool + Name="VCXDCMakeTool" + /> + <Tool + Name="VCBscMakeTool" + /> + <Tool + Name="VCFxCopTool" + /> + <Tool + Name="VCAppVerifierTool" + /> + <Tool + Name="VCPostBuildEventTool" + /> + </Configuration> + </Configurations> + <References> + </References> + <Files> + <Filter + Name="Source Files" + Filter="cpp;c;cc;cxx;def;odl;idl;hpj;bat;asm;asmx" + UniqueIdentifier="{4FC737F1-C7A5-4376-A066-2A32D752A2FF}" + > + <File + RelativePath="..\..\..\programs\datagen.c" + > + </File> + <File + RelativePath="..\..\..\lib\common\entropy_common.c" + > + </File> + <File + RelativePath="..\..\..\lib\common\error_private.c" + > + </File> + <File + RelativePath="..\..\..\lib\compress\fse_compress.c" + > + </File> + <File + RelativePath="..\..\..\lib\common\fse_decompress.c" + > + </File> + <File + RelativePath="..\..\..\lib\common\xxhash.c" + > + </File> + <File + RelativePath="..\..\..\tests\fullbench.c" + > + </File> + <File + RelativePath="..\..\..\lib\compress\huf_compress.c" + > + </File> + <File + RelativePath="..\..\..\lib\decompress\huf_decompress.c" + > + </File> + <File + RelativePath="..\..\..\lib\common\zstd_common.c" + > + </File> + <File + RelativePath="..\..\..\lib\compress\zstd_compress.c" + > + </File> + <File + RelativePath="..\..\..\lib\decompress\zstd_decompress.c" + > + </File> + </Filter> + <Filter + Name="Header Files" + Filter="h;hpp;hxx;hm;inl;inc;xsd" + UniqueIdentifier="{93995380-89BD-4b04-88EB-625FBE52EBFB}" + > + <File + RelativePath="..\..\..\lib\common\bitstream.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\error_private.h" + > + </File> + <File + RelativePath="..\..\..\lib\zstd_errors.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\fse.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\fse_static.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\huf.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\huf_static.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\xxhash.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\mem.h" + > + </File> + <File + RelativePath="..\..\..\lib\zstd.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\zstd_internal.h" + > + </File> + <File + RelativePath="..\..\..\lib\compress\zstd_opt.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\zstd_static.h" + > + </File> + </Filter> + </Files> + <Globals> + </Globals> +</VisualStudioProject> diff --git a/3rdparty/zstd/contrib/VS2005/fuzzer/fuzzer.vcproj b/3rdparty/zstd/contrib/VS2005/fuzzer/fuzzer.vcproj new file mode 100644 index 00000000000..d182535fe2d --- /dev/null +++ b/3rdparty/zstd/contrib/VS2005/fuzzer/fuzzer.vcproj @@ -0,0 +1,488 @@ +<?xml version="1.0" encoding="Windows-1252"?> +<VisualStudioProject + ProjectType="Visual C++" + Version="8.00" + Name="fuzzer" + ProjectGUID="{A62E89D2-9DDE-42BA-8F9B-9DA74889A6B0}" + RootNamespace="fuzzer" + Keyword="Win32Proj" + > + <Platforms> + <Platform + Name="Win32" + /> + <Platform + Name="x64" + /> + </Platforms> + <ToolFiles> + </ToolFiles> + <Configurations> + <Configuration + Name="Debug|Win32" + OutputDirectory="$(SolutionDir)bin\$(PlatformName)\$(ConfigurationName)" + IntermediateDirectory="$(PlatformName)\$(ConfigurationName)" + ConfigurationType="1" + CharacterSet="2" + > + <Tool + Name="VCPreBuildEventTool" + /> + <Tool + Name="VCCustomBuildTool" + /> + <Tool + Name="VCXMLDataGeneratorTool" + /> + <Tool + Name="VCWebServiceProxyGeneratorTool" + /> + <Tool + Name="VCMIDLTool" + /> + <Tool + Name="VCCLCompilerTool" + Optimization="0" + AdditionalIncludeDirectories="$(SolutionDir)..\..\lib;$(SolutionDir)..\..\lib\common;$(SolutionDir)..\..\lib\dictBuilder;$(SolutionDir)..\..\programs" + PreprocessorDefinitions="WIN32;_DEBUG;_CONSOLE" + MinimalRebuild="true" + BasicRuntimeChecks="3" + RuntimeLibrary="3" + UsePrecompiledHeader="0" + WarningLevel="4" + WarnAsError="true" + DebugInformationFormat="4" + /> + <Tool + Name="VCManagedResourceCompilerTool" + /> + <Tool + Name="VCResourceCompilerTool" + /> + <Tool + Name="VCPreLinkEventTool" + /> + <Tool + Name="VCLinkerTool" + LinkIncremental="2" + GenerateDebugInformation="true" + SubSystem="1" + TargetMachine="1" + /> + <Tool + Name="VCALinkTool" + /> + <Tool + Name="VCManifestTool" + /> + <Tool + Name="VCXDCMakeTool" + /> + <Tool + Name="VCBscMakeTool" + /> + <Tool + Name="VCFxCopTool" + /> + <Tool + Name="VCAppVerifierTool" + /> + <Tool + Name="VCPostBuildEventTool" + /> + </Configuration> + <Configuration + Name="Release|Win32" + OutputDirectory="$(SolutionDir)bin\$(PlatformName)\$(ConfigurationName)" + IntermediateDirectory="$(PlatformName)\$(ConfigurationName)" + ConfigurationType="1" + CharacterSet="2" + WholeProgramOptimization="1" + > + <Tool + Name="VCPreBuildEventTool" + /> + <Tool + Name="VCCustomBuildTool" + /> + <Tool + Name="VCXMLDataGeneratorTool" + /> + <Tool + Name="VCWebServiceProxyGeneratorTool" + /> + <Tool + Name="VCMIDLTool" + /> + <Tool + Name="VCCLCompilerTool" + Optimization="2" + EnableIntrinsicFunctions="true" + OmitFramePointers="true" + AdditionalIncludeDirectories="$(SolutionDir)..\..\lib;$(SolutionDir)..\..\lib\common;$(SolutionDir)..\..\lib\dictBuilder;$(SolutionDir)..\..\programs" + PreprocessorDefinitions="WIN32;NDEBUG;_CONSOLE" + RuntimeLibrary="0" + EnableFunctionLevelLinking="true" + UsePrecompiledHeader="0" + WarningLevel="4" + DebugInformationFormat="3" + /> + <Tool + Name="VCManagedResourceCompilerTool" + /> + <Tool + Name="VCResourceCompilerTool" + /> + <Tool + Name="VCPreLinkEventTool" + /> + <Tool + Name="VCLinkerTool" + LinkIncremental="1" + GenerateDebugInformation="true" + SubSystem="1" + OptimizeReferences="2" + EnableCOMDATFolding="2" + TargetMachine="1" + /> + <Tool + Name="VCALinkTool" + /> + <Tool + Name="VCManifestTool" + /> + <Tool + Name="VCXDCMakeTool" + /> + <Tool + Name="VCBscMakeTool" + /> + <Tool + Name="VCFxCopTool" + /> + <Tool + Name="VCAppVerifierTool" + /> + <Tool + Name="VCPostBuildEventTool" + /> + </Configuration> + <Configuration + Name="Debug|x64" + OutputDirectory="$(SolutionDir)bin\$(PlatformName)\$(ConfigurationName)" + IntermediateDirectory="$(PlatformName)\$(ConfigurationName)" + ConfigurationType="1" + CharacterSet="2" + > + <Tool + Name="VCPreBuildEventTool" + /> + <Tool + Name="VCCustomBuildTool" + /> + <Tool + Name="VCXMLDataGeneratorTool" + /> + <Tool + Name="VCWebServiceProxyGeneratorTool" + /> + <Tool + Name="VCMIDLTool" + TargetEnvironment="3" + /> + <Tool + Name="VCCLCompilerTool" + Optimization="0" + AdditionalIncludeDirectories="$(SolutionDir)..\..\lib;$(SolutionDir)..\..\lib\common;$(SolutionDir)..\..\lib\dictBuilder;$(SolutionDir)..\..\programs" + PreprocessorDefinitions="WIN32;_DEBUG;_CONSOLE" + MinimalRebuild="true" + BasicRuntimeChecks="3" + RuntimeLibrary="3" + UsePrecompiledHeader="0" + WarningLevel="4" + WarnAsError="true" + DebugInformationFormat="3" + /> + <Tool + Name="VCManagedResourceCompilerTool" + /> + <Tool + Name="VCResourceCompilerTool" + /> + <Tool + Name="VCPreLinkEventTool" + /> + <Tool + Name="VCLinkerTool" + LinkIncremental="2" + GenerateDebugInformation="true" + SubSystem="1" + TargetMachine="17" + /> + <Tool + Name="VCALinkTool" + /> + <Tool + Name="VCManifestTool" + /> + <Tool + Name="VCXDCMakeTool" + /> + <Tool + Name="VCBscMakeTool" + /> + <Tool + Name="VCFxCopTool" + /> + <Tool + Name="VCAppVerifierTool" + /> + <Tool + Name="VCPostBuildEventTool" + /> + </Configuration> + <Configuration + Name="Release|x64" + OutputDirectory="$(SolutionDir)bin\$(PlatformName)\$(ConfigurationName)" + IntermediateDirectory="$(PlatformName)\$(ConfigurationName)" + ConfigurationType="1" + CharacterSet="2" + WholeProgramOptimization="1" + > + <Tool + Name="VCPreBuildEventTool" + /> + <Tool + Name="VCCustomBuildTool" + /> + <Tool + Name="VCXMLDataGeneratorTool" + /> + <Tool + Name="VCWebServiceProxyGeneratorTool" + /> + <Tool + Name="VCMIDLTool" + TargetEnvironment="3" + /> + <Tool + Name="VCCLCompilerTool" + Optimization="2" + EnableIntrinsicFunctions="true" + OmitFramePointers="true" + AdditionalIncludeDirectories="$(SolutionDir)..\..\lib;$(SolutionDir)..\..\lib\common;$(SolutionDir)..\..\lib\dictBuilder;$(SolutionDir)..\..\programs" + PreprocessorDefinitions="WIN32;NDEBUG;_CONSOLE" + RuntimeLibrary="0" + EnableFunctionLevelLinking="true" + UsePrecompiledHeader="0" + WarningLevel="4" + DebugInformationFormat="3" + /> + <Tool + Name="VCManagedResourceCompilerTool" + /> + <Tool + Name="VCResourceCompilerTool" + /> + <Tool + Name="VCPreLinkEventTool" + /> + <Tool + Name="VCLinkerTool" + LinkIncremental="1" + GenerateDebugInformation="true" + SubSystem="1" + OptimizeReferences="2" + EnableCOMDATFolding="2" + TargetMachine="17" + /> + <Tool + Name="VCALinkTool" + /> + <Tool + Name="VCManifestTool" + /> + <Tool + Name="VCXDCMakeTool" + /> + <Tool + Name="VCBscMakeTool" + /> + <Tool + Name="VCFxCopTool" + /> + <Tool + Name="VCAppVerifierTool" + /> + <Tool + Name="VCPostBuildEventTool" + /> + </Configuration> + </Configurations> + <References> + </References> + <Files> + <Filter + Name="Source Files" + Filter="cpp;c;cc;cxx;def;odl;idl;hpj;bat;asm;asmx" + UniqueIdentifier="{4FC737F1-C7A5-4376-A066-2A32D752A2FF}" + > + <File + RelativePath="..\..\..\programs\datagen.c" + > + </File> + <File + RelativePath="..\..\..\lib\dictBuilder\cover.c" + > + </File> + <File + RelativePath="..\..\..\lib\dictBuilder\divsufsort.c" + > + </File> + <File + RelativePath="..\..\..\lib\common\entropy_common.c" + > + </File> + <File + RelativePath="..\..\..\lib\common\pool.c" + > + </File> + <File + RelativePath="..\..\..\lib\common\threading.c" + > + </File> + <File + RelativePath="..\..\..\lib\common\error_private.c" + > + </File> + <File + RelativePath="..\..\..\lib\compress\zstdmt_compress.c" + > + </File> + <File + RelativePath="..\..\..\lib\compress\fse_compress.c" + > + </File> + <File + RelativePath="..\..\..\lib\common\fse_decompress.c" + > + </File> + <File + RelativePath="..\..\..\tests\fuzzer.c" + > + </File> + <File + RelativePath="..\..\..\lib\compress\huf_compress.c" + > + </File> + <File + RelativePath="..\..\..\lib\decompress\huf_decompress.c" + > + </File> + <File + RelativePath="..\..\..\lib\common\xxhash.c" + > + </File> + <File + RelativePath="..\..\..\lib\dictBuilder\zdict.c" + > + </File> + <File + RelativePath="..\..\..\lib\common\zstd_common.c" + > + </File> + <File + RelativePath="..\..\..\lib\compress\zstd_compress.c" + > + </File> + <File + RelativePath="..\..\..\lib\decompress\zstd_decompress.c" + > + </File> + </Filter> + <Filter + Name="Header Files" + Filter="h;hpp;hxx;hm;inl;inc;xsd" + UniqueIdentifier="{93995380-89BD-4b04-88EB-625FBE52EBFB}" + > + <File + RelativePath="..\..\..\lib\common\pool.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\threading.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\bitstream.h" + > + </File> + <File + RelativePath="..\..\..\lib\dictBuilder\divsufsort.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\error_private.h" + > + </File> + <File + RelativePath="..\..\..\lib\zstd_errors.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\fse.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\fse_static.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\huf.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\huf_static.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\mem.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\xxhash.h" + > + </File> + <File + RelativePath="..\..\..\lib\zdict.h" + > + </File> + <File + RelativePath="..\..\..\lib\dictBuilder\zdict_static.h" + > + </File> + <File + RelativePath="..\..\..\lib\zstd.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\zstd_internal.h" + > + </File> + <File + RelativePath="..\..\..\lib\compress\zstdmt_compress.h" + > + </File> + <File + RelativePath="..\..\..\lib\compress\zstd_opt.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\zstd_static.h" + > + </File> + </Filter> + </Files> + <Globals> + </Globals> +</VisualStudioProject> diff --git a/3rdparty/zstd/contrib/VS2005/zstd.sln b/3rdparty/zstd/contrib/VS2005/zstd.sln new file mode 100644 index 00000000000..0c31ae126c0 --- /dev/null +++ b/3rdparty/zstd/contrib/VS2005/zstd.sln @@ -0,0 +1,55 @@ +Microsoft Visual Studio Solution File, Format Version 9.00 +# Visual C++ Express 2005 +Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "zstd", "zstd\zstd.vcproj", "{1A2AB08E-5CE7-4C5B-BE55-458157C14051}" +EndProject +Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "fuzzer", "fuzzer\fuzzer.vcproj", "{A62E89D2-9DDE-42BA-8F9B-9DA74889A6B0}" +EndProject +Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "fullbench", "fullbench\fullbench.vcproj", "{CC8F1D1B-BA2F-43E3-A71F-FA415D81AAD3}" +EndProject +Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "zstdlib", "zstdlib\zstdlib.vcproj", "{99DE2A79-7298-4004-A0ED-030D7A3796CA}" +EndProject +Global + GlobalSection(SolutionConfigurationPlatforms) = preSolution + Debug|Win32 = Debug|Win32 + Debug|x64 = Debug|x64 + Release|Win32 = Release|Win32 + Release|x64 = Release|x64 + EndGlobalSection + GlobalSection(ProjectConfigurationPlatforms) = postSolution + {1A2AB08E-5CE7-4C5B-BE55-458157C14051}.Debug|Win32.ActiveCfg = Debug|Win32 + {1A2AB08E-5CE7-4C5B-BE55-458157C14051}.Debug|Win32.Build.0 = Debug|Win32 + {1A2AB08E-5CE7-4C5B-BE55-458157C14051}.Debug|x64.ActiveCfg = Debug|x64 + {1A2AB08E-5CE7-4C5B-BE55-458157C14051}.Debug|x64.Build.0 = Debug|x64 + {1A2AB08E-5CE7-4C5B-BE55-458157C14051}.Release|Win32.ActiveCfg = Release|Win32 + {1A2AB08E-5CE7-4C5B-BE55-458157C14051}.Release|Win32.Build.0 = Release|Win32 + {1A2AB08E-5CE7-4C5B-BE55-458157C14051}.Release|x64.ActiveCfg = Release|x64 + {1A2AB08E-5CE7-4C5B-BE55-458157C14051}.Release|x64.Build.0 = Release|x64 + {A62E89D2-9DDE-42BA-8F9B-9DA74889A6B0}.Debug|Win32.ActiveCfg = Debug|Win32 + {A62E89D2-9DDE-42BA-8F9B-9DA74889A6B0}.Debug|Win32.Build.0 = Debug|Win32 + {A62E89D2-9DDE-42BA-8F9B-9DA74889A6B0}.Debug|x64.ActiveCfg = Debug|x64 + {A62E89D2-9DDE-42BA-8F9B-9DA74889A6B0}.Debug|x64.Build.0 = Debug|x64 + {A62E89D2-9DDE-42BA-8F9B-9DA74889A6B0}.Release|Win32.ActiveCfg = Release|Win32 + {A62E89D2-9DDE-42BA-8F9B-9DA74889A6B0}.Release|Win32.Build.0 = Release|Win32 + {A62E89D2-9DDE-42BA-8F9B-9DA74889A6B0}.Release|x64.ActiveCfg = Release|x64 + {A62E89D2-9DDE-42BA-8F9B-9DA74889A6B0}.Release|x64.Build.0 = Release|x64 + {CC8F1D1B-BA2F-43E3-A71F-FA415D81AAD3}.Debug|Win32.ActiveCfg = Debug|Win32 + {CC8F1D1B-BA2F-43E3-A71F-FA415D81AAD3}.Debug|Win32.Build.0 = Debug|Win32 + {CC8F1D1B-BA2F-43E3-A71F-FA415D81AAD3}.Debug|x64.ActiveCfg = Debug|x64 + {CC8F1D1B-BA2F-43E3-A71F-FA415D81AAD3}.Debug|x64.Build.0 = Debug|x64 + {CC8F1D1B-BA2F-43E3-A71F-FA415D81AAD3}.Release|Win32.ActiveCfg = Release|Win32 + {CC8F1D1B-BA2F-43E3-A71F-FA415D81AAD3}.Release|Win32.Build.0 = Release|Win32 + {CC8F1D1B-BA2F-43E3-A71F-FA415D81AAD3}.Release|x64.ActiveCfg = Release|x64 + {CC8F1D1B-BA2F-43E3-A71F-FA415D81AAD3}.Release|x64.Build.0 = Release|x64 + {99DE2A79-7298-4004-A0ED-030D7A3796CA}.Debug|Win32.ActiveCfg = Debug|Win32 + {99DE2A79-7298-4004-A0ED-030D7A3796CA}.Debug|Win32.Build.0 = Debug|Win32 + {99DE2A79-7298-4004-A0ED-030D7A3796CA}.Debug|x64.ActiveCfg = Debug|x64 + {99DE2A79-7298-4004-A0ED-030D7A3796CA}.Debug|x64.Build.0 = Debug|x64 + {99DE2A79-7298-4004-A0ED-030D7A3796CA}.Release|Win32.ActiveCfg = Release|Win32 + {99DE2A79-7298-4004-A0ED-030D7A3796CA}.Release|Win32.Build.0 = Release|Win32 + {99DE2A79-7298-4004-A0ED-030D7A3796CA}.Release|x64.ActiveCfg = Release|x64 + {99DE2A79-7298-4004-A0ED-030D7A3796CA}.Release|x64.Build.0 = Release|x64 + EndGlobalSection + GlobalSection(SolutionProperties) = preSolution + HideSolutionNode = FALSE + EndGlobalSection +EndGlobal diff --git a/3rdparty/zstd/contrib/VS2005/zstd/zstd.vcproj b/3rdparty/zstd/contrib/VS2005/zstd/zstd.vcproj new file mode 100644 index 00000000000..e37ebee3911 --- /dev/null +++ b/3rdparty/zstd/contrib/VS2005/zstd/zstd.vcproj @@ -0,0 +1,552 @@ +<?xml version="1.0" encoding="Windows-1252"?> +<VisualStudioProject + ProjectType="Visual C++" + Version="8.00" + Name="zstd" + ProjectGUID="{1A2AB08E-5CE7-4C5B-BE55-458157C14051}" + RootNamespace="zstd" + Keyword="Win32Proj" + > + <Platforms> + <Platform + Name="Win32" + /> + <Platform + Name="x64" + /> + </Platforms> + <ToolFiles> + </ToolFiles> + <Configurations> + <Configuration + Name="Debug|Win32" + OutputDirectory="$(SolutionDir)bin\$(PlatformName)\$(ConfigurationName)" + IntermediateDirectory="$(PlatformName)\$(ConfigurationName)" + ConfigurationType="1" + CharacterSet="2" + > + <Tool + Name="VCPreBuildEventTool" + /> + <Tool + Name="VCCustomBuildTool" + /> + <Tool + Name="VCXMLDataGeneratorTool" + /> + <Tool + Name="VCWebServiceProxyGeneratorTool" + /> + <Tool + Name="VCMIDLTool" + /> + <Tool + Name="VCCLCompilerTool" + Optimization="0" + AdditionalIncludeDirectories="$(SolutionDir)..\..\lib;$(SolutionDir)..\..\lib\common;$(SolutionDir)..\..\lib\legacy;$(SolutionDir)..\..\lib\dictBuilder;$(SolutionDir)..\..\lib\compress" + PreprocessorDefinitions="ZSTD_MULTITHREAD=1;ZSTD_LEGACY_SUPPORT=4;WIN32;_DEBUG;_CONSOLE" + MinimalRebuild="true" + BasicRuntimeChecks="3" + RuntimeLibrary="3" + UsePrecompiledHeader="0" + WarningLevel="4" + WarnAsError="true" + DebugInformationFormat="4" + /> + <Tool + Name="VCManagedResourceCompilerTool" + /> + <Tool + Name="VCResourceCompilerTool" + /> + <Tool + Name="VCPreLinkEventTool" + /> + <Tool + Name="VCLinkerTool" + AdditionalDependencies="setargv.obj" + LinkIncremental="2" + GenerateDebugInformation="true" + SubSystem="1" + TargetMachine="1" + /> + <Tool + Name="VCALinkTool" + /> + <Tool + Name="VCManifestTool" + /> + <Tool + Name="VCXDCMakeTool" + /> + <Tool + Name="VCBscMakeTool" + /> + <Tool + Name="VCFxCopTool" + /> + <Tool + Name="VCAppVerifierTool" + /> + <Tool + Name="VCPostBuildEventTool" + /> + </Configuration> + <Configuration + Name="Release|Win32" + OutputDirectory="$(SolutionDir)bin\$(PlatformName)\$(ConfigurationName)" + IntermediateDirectory="$(PlatformName)\$(ConfigurationName)" + ConfigurationType="1" + CharacterSet="2" + WholeProgramOptimization="1" + > + <Tool + Name="VCPreBuildEventTool" + /> + <Tool + Name="VCCustomBuildTool" + /> + <Tool + Name="VCXMLDataGeneratorTool" + /> + <Tool + Name="VCWebServiceProxyGeneratorTool" + /> + <Tool + Name="VCMIDLTool" + /> + <Tool + Name="VCCLCompilerTool" + Optimization="2" + EnableIntrinsicFunctions="true" + OmitFramePointers="true" + AdditionalIncludeDirectories="$(SolutionDir)..\..\lib;$(SolutionDir)..\..\lib\common;$(SolutionDir)..\..\lib\legacy;$(SolutionDir)..\..\lib\dictBuilder;$(SolutionDir)..\..\lib\compress" + PreprocessorDefinitions="ZSTD_MULTITHREAD=1;ZSTD_LEGACY_SUPPORT=4;WIN32;NDEBUG;_CONSOLE" + RuntimeLibrary="0" + EnableFunctionLevelLinking="true" + UsePrecompiledHeader="0" + WarningLevel="4" + DebugInformationFormat="3" + /> + <Tool + Name="VCManagedResourceCompilerTool" + /> + <Tool + Name="VCResourceCompilerTool" + /> + <Tool + Name="VCPreLinkEventTool" + /> + <Tool + Name="VCLinkerTool" + AdditionalDependencies="setargv.obj" + LinkIncremental="1" + GenerateDebugInformation="true" + SubSystem="1" + OptimizeReferences="2" + EnableCOMDATFolding="2" + TargetMachine="1" + /> + <Tool + Name="VCALinkTool" + /> + <Tool + Name="VCManifestTool" + /> + <Tool + Name="VCXDCMakeTool" + /> + <Tool + Name="VCBscMakeTool" + /> + <Tool + Name="VCFxCopTool" + /> + <Tool + Name="VCAppVerifierTool" + /> + <Tool + Name="VCPostBuildEventTool" + /> + </Configuration> + <Configuration + Name="Debug|x64" + OutputDirectory="$(SolutionDir)bin\$(PlatformName)\$(ConfigurationName)" + IntermediateDirectory="$(PlatformName)\$(ConfigurationName)" + ConfigurationType="1" + CharacterSet="2" + > + <Tool + Name="VCPreBuildEventTool" + /> + <Tool + Name="VCCustomBuildTool" + /> + <Tool + Name="VCXMLDataGeneratorTool" + /> + <Tool + Name="VCWebServiceProxyGeneratorTool" + /> + <Tool + Name="VCMIDLTool" + TargetEnvironment="3" + /> + <Tool + Name="VCCLCompilerTool" + Optimization="0" + AdditionalIncludeDirectories="$(SolutionDir)..\..\lib;$(SolutionDir)..\..\lib\common;$(SolutionDir)..\..\lib\legacy;$(SolutionDir)..\..\lib\dictBuilder;$(SolutionDir)..\..\lib\compress" + PreprocessorDefinitions="ZSTD_MULTITHREAD=1;ZSTD_LEGACY_SUPPORT=4;WIN32;_DEBUG;_CONSOLE" + MinimalRebuild="true" + BasicRuntimeChecks="3" + RuntimeLibrary="3" + UsePrecompiledHeader="0" + WarningLevel="4" + WarnAsError="true" + DebugInformationFormat="3" + /> + <Tool + Name="VCManagedResourceCompilerTool" + /> + <Tool + Name="VCResourceCompilerTool" + /> + <Tool + Name="VCPreLinkEventTool" + /> + <Tool + Name="VCLinkerTool" + AdditionalDependencies="setargv.obj" + LinkIncremental="2" + GenerateDebugInformation="true" + SubSystem="1" + TargetMachine="17" + /> + <Tool + Name="VCALinkTool" + /> + <Tool + Name="VCManifestTool" + /> + <Tool + Name="VCXDCMakeTool" + /> + <Tool + Name="VCBscMakeTool" + /> + <Tool + Name="VCFxCopTool" + /> + <Tool + Name="VCAppVerifierTool" + /> + <Tool + Name="VCPostBuildEventTool" + /> + </Configuration> + <Configuration + Name="Release|x64" + OutputDirectory="$(SolutionDir)bin\$(PlatformName)\$(ConfigurationName)" + IntermediateDirectory="$(PlatformName)\$(ConfigurationName)" + ConfigurationType="1" + CharacterSet="2" + WholeProgramOptimization="1" + > + <Tool + Name="VCPreBuildEventTool" + /> + <Tool + Name="VCCustomBuildTool" + /> + <Tool + Name="VCXMLDataGeneratorTool" + /> + <Tool + Name="VCWebServiceProxyGeneratorTool" + /> + <Tool + Name="VCMIDLTool" + TargetEnvironment="3" + /> + <Tool + Name="VCCLCompilerTool" + Optimization="2" + EnableIntrinsicFunctions="true" + OmitFramePointers="true" + AdditionalIncludeDirectories="$(SolutionDir)..\..\lib;$(SolutionDir)..\..\lib\common;$(SolutionDir)..\..\lib\legacy;$(SolutionDir)..\..\lib\dictBuilder;$(SolutionDir)..\..\lib\compress" + PreprocessorDefinitions="ZSTD_MULTITHREAD=1;ZSTD_LEGACY_SUPPORT=4;WIN32;NDEBUG;_CONSOLE" + RuntimeLibrary="0" + EnableFunctionLevelLinking="true" + UsePrecompiledHeader="0" + WarningLevel="4" + DebugInformationFormat="3" + /> + <Tool + Name="VCManagedResourceCompilerTool" + /> + <Tool + Name="VCResourceCompilerTool" + /> + <Tool + Name="VCPreLinkEventTool" + /> + <Tool + Name="VCLinkerTool" + AdditionalDependencies="setargv.obj" + LinkIncremental="1" + GenerateDebugInformation="true" + SubSystem="1" + OptimizeReferences="2" + EnableCOMDATFolding="2" + TargetMachine="17" + /> + <Tool + Name="VCALinkTool" + /> + <Tool + Name="VCManifestTool" + /> + <Tool + Name="VCXDCMakeTool" + /> + <Tool + Name="VCBscMakeTool" + /> + <Tool + Name="VCFxCopTool" + /> + <Tool + Name="VCAppVerifierTool" + /> + <Tool + Name="VCPostBuildEventTool" + /> + </Configuration> + </Configurations> + <References> + </References> + <Files> + <Filter + Name="Source Files" + Filter="cpp;c;cc;cxx;def;odl;idl;hpj;bat;asm;asmx" + UniqueIdentifier="{4FC737F1-C7A5-4376-A066-2A32D752A2FF}" + > + <File + RelativePath="..\..\..\programs\bench.c" + > + </File> + <File + RelativePath="..\..\..\programs\datagen.c" + > + </File> + <File + RelativePath="..\..\..\programs\dibio.c" + > + </File> + <File + RelativePath="..\..\..\lib\dictBuilder\cover.c" + > + </File> + <File + RelativePath="..\..\..\lib\dictBuilder\divsufsort.c" + > + </File> + <File + RelativePath="..\..\..\lib\common\entropy_common.c" + > + </File> + <File + RelativePath="..\..\..\lib\common\error_private.c" + > + </File> + <File + RelativePath="..\..\..\programs\fileio.c" + > + </File> + <File + RelativePath="..\..\..\programs\fileio_asyncio.c" + > + </File> + <File + RelativePath="..\..\..\lib\compress\fse_compress.c" + > + </File> + <File + RelativePath="..\..\..\lib\common\fse_decompress.c" + > + </File> + <File + RelativePath="..\..\..\lib\compress\huf_compress.c" + > + </File> + <File + RelativePath="..\..\..\lib\decompress\huf_decompress.c" + > + </File> + <File + RelativePath="..\..\..\lib\common\xxhash.c" + > + </File> + <File + RelativePath="..\..\..\lib\dictBuilder\zdict.c" + > + </File> + <File + RelativePath="..\..\..\lib\common\zstd_common.c" + > + </File> + <File + RelativePath="..\..\..\lib\compress\zstd_compress.c" + > + </File> + <File + RelativePath="..\..\..\lib\compress\zstdmt_compress.c" + > + </File> + <File + RelativePath="..\..\..\lib\decompress\zstd_decompress.c" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v01.c" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v02.c" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v03.c" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v04.c" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v05.c" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v06.c" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v07.c" + > + </File> + <File + RelativePath="..\..\..\programs\zstdcli.c" + > + </File> + </Filter> + <Filter + Name="Header Files" + Filter="h;hpp;hxx;hm;inl;inc;xsd" + UniqueIdentifier="{93995380-89BD-4b04-88EB-625FBE52EBFB}" + > + <File + RelativePath="..\..\..\lib\common\bitstream.h" + > + </File> + <File + RelativePath="..\..\..\lib\dictBuilder\divsufsort.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\error_private.h" + > + </File> + <File + RelativePath="..\..\..\lib\zstd_errors.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\fse.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\fse_static.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\huf.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\huf_static.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\mem.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\xxhash.h" + > + </File> + <File + RelativePath="..\..\..\lib\zdict.h" + > + </File> + <File + RelativePath="..\..\..\lib\dictBuilder\zdict_static.h" + > + </File> + <File + RelativePath="..\..\..\lib\zstd.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\zstd_internal.h" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_legacy.h" + > + </File> + <File + RelativePath="..\..\..\lib\compress\zstd_opt.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\zstd_static.h" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v01.h" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v02.h" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v03.h" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v04.h" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v05.h" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v06.h" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v07.h" + > + </File> + <File + RelativePath="..\..\..\lib\compress\zstdmt_compress.h" + > + </File> + </Filter> + </Files> + <Globals> + </Globals> +</VisualStudioProject> diff --git a/3rdparty/zstd/contrib/VS2005/zstdlib/zstdlib.vcproj b/3rdparty/zstd/contrib/VS2005/zstdlib/zstdlib.vcproj new file mode 100644 index 00000000000..67ddd2dba3a --- /dev/null +++ b/3rdparty/zstd/contrib/VS2005/zstdlib/zstdlib.vcproj @@ -0,0 +1,546 @@ +<?xml version="1.0" encoding="Windows-1252"?> +<VisualStudioProject + ProjectType="Visual C++" + Version="8.00" + Name="zstdlib" + ProjectGUID="{99DE2A79-7298-4004-A0ED-030D7A3796CA}" + RootNamespace="zstdlib" + Keyword="Win32Proj" + > + <Platforms> + <Platform + Name="Win32" + /> + <Platform + Name="x64" + /> + </Platforms> + <ToolFiles> + </ToolFiles> + <Configurations> + <Configuration + Name="Debug|Win32" + OutputDirectory="$(SolutionDir)bin\$(PlatformName)\$(ConfigurationName)" + IntermediateDirectory="$(PlatformName)\$(ConfigurationName)" + ConfigurationType="2" + CharacterSet="2" + > + <Tool + Name="VCPreBuildEventTool" + /> + <Tool + Name="VCCustomBuildTool" + /> + <Tool + Name="VCXMLDataGeneratorTool" + /> + <Tool + Name="VCWebServiceProxyGeneratorTool" + /> + <Tool + Name="VCMIDLTool" + /> + <Tool + Name="VCCLCompilerTool" + Optimization="0" + AdditionalIncludeDirectories="$(SolutionDir)..\..\lib;$(SolutionDir)..\..\lib\common;$(SolutionDir)..\..\lib\legacy;$(SolutionDir)..\..\programs\legacy;$(SolutionDir)..\..\lib\dictBuilder" + PreprocessorDefinitions="ZSTD_DLL_EXPORT=1;ZSTD_MULTITHREAD=1;ZSTD_LEGACY_SUPPORT=4;WIN32;_DEBUG;_CONSOLE" + MinimalRebuild="true" + BasicRuntimeChecks="3" + RuntimeLibrary="3" + UsePrecompiledHeader="0" + WarningLevel="4" + WarnAsError="true" + DebugInformationFormat="4" + /> + <Tool + Name="VCManagedResourceCompilerTool" + /> + <Tool + Name="VCResourceCompilerTool" + /> + <Tool + Name="VCPreLinkEventTool" + /> + <Tool + Name="VCLinkerTool" + LinkIncremental="2" + GenerateDebugInformation="true" + SubSystem="1" + TargetMachine="1" + /> + <Tool + Name="VCALinkTool" + /> + <Tool + Name="VCManifestTool" + /> + <Tool + Name="VCXDCMakeTool" + /> + <Tool + Name="VCBscMakeTool" + /> + <Tool + Name="VCFxCopTool" + /> + <Tool + Name="VCAppVerifierTool" + /> + <Tool + Name="VCPostBuildEventTool" + /> + </Configuration> + <Configuration + Name="Release|Win32" + OutputDirectory="$(SolutionDir)bin\$(PlatformName)\$(ConfigurationName)" + IntermediateDirectory="$(PlatformName)\$(ConfigurationName)" + ConfigurationType="2" + CharacterSet="2" + WholeProgramOptimization="1" + > + <Tool + Name="VCPreBuildEventTool" + /> + <Tool + Name="VCCustomBuildTool" + /> + <Tool + Name="VCXMLDataGeneratorTool" + /> + <Tool + Name="VCWebServiceProxyGeneratorTool" + /> + <Tool + Name="VCMIDLTool" + /> + <Tool + Name="VCCLCompilerTool" + Optimization="2" + EnableIntrinsicFunctions="true" + OmitFramePointers="true" + AdditionalIncludeDirectories="$(SolutionDir)..\..\lib;$(SolutionDir)..\..\lib\common;$(SolutionDir)..\..\lib\legacy;$(SolutionDir)..\..\programs\legacy;$(SolutionDir)..\..\lib\dictBuilder" + PreprocessorDefinitions="ZSTD_DLL_EXPORT=1;ZSTD_MULTITHREAD=1;ZSTD_LEGACY_SUPPORT=4;WIN32;NDEBUG;_CONSOLE" + RuntimeLibrary="0" + EnableFunctionLevelLinking="true" + UsePrecompiledHeader="0" + WarningLevel="4" + DebugInformationFormat="3" + /> + <Tool + Name="VCManagedResourceCompilerTool" + /> + <Tool + Name="VCResourceCompilerTool" + /> + <Tool + Name="VCPreLinkEventTool" + /> + <Tool + Name="VCLinkerTool" + LinkIncremental="1" + GenerateDebugInformation="true" + SubSystem="1" + OptimizeReferences="2" + EnableCOMDATFolding="2" + TargetMachine="1" + /> + <Tool + Name="VCALinkTool" + /> + <Tool + Name="VCManifestTool" + /> + <Tool + Name="VCXDCMakeTool" + /> + <Tool + Name="VCBscMakeTool" + /> + <Tool + Name="VCFxCopTool" + /> + <Tool + Name="VCAppVerifierTool" + /> + <Tool + Name="VCPostBuildEventTool" + /> + </Configuration> + <Configuration + Name="Debug|x64" + OutputDirectory="$(SolutionDir)bin\$(PlatformName)\$(ConfigurationName)" + IntermediateDirectory="$(PlatformName)\$(ConfigurationName)" + ConfigurationType="2" + CharacterSet="2" + > + <Tool + Name="VCPreBuildEventTool" + /> + <Tool + Name="VCCustomBuildTool" + /> + <Tool + Name="VCXMLDataGeneratorTool" + /> + <Tool + Name="VCWebServiceProxyGeneratorTool" + /> + <Tool + Name="VCMIDLTool" + TargetEnvironment="3" + /> + <Tool + Name="VCCLCompilerTool" + Optimization="0" + AdditionalIncludeDirectories="$(SolutionDir)..\..\lib;$(SolutionDir)..\..\lib\common;$(SolutionDir)..\..\lib\legacy;$(SolutionDir)..\..\programs\legacy;$(SolutionDir)..\..\lib\dictBuilder" + PreprocessorDefinitions="ZSTD_DLL_EXPORT=1;ZSTD_MULTITHREAD=1;ZSTD_LEGACY_SUPPORT=4;WIN32;_DEBUG;_CONSOLE" + MinimalRebuild="true" + BasicRuntimeChecks="3" + RuntimeLibrary="3" + UsePrecompiledHeader="0" + WarningLevel="4" + WarnAsError="true" + DebugInformationFormat="3" + /> + <Tool + Name="VCManagedResourceCompilerTool" + /> + <Tool + Name="VCResourceCompilerTool" + /> + <Tool + Name="VCPreLinkEventTool" + /> + <Tool + Name="VCLinkerTool" + LinkIncremental="2" + GenerateDebugInformation="true" + SubSystem="1" + TargetMachine="17" + /> + <Tool + Name="VCALinkTool" + /> + <Tool + Name="VCManifestTool" + /> + <Tool + Name="VCXDCMakeTool" + /> + <Tool + Name="VCBscMakeTool" + /> + <Tool + Name="VCFxCopTool" + /> + <Tool + Name="VCAppVerifierTool" + /> + <Tool + Name="VCPostBuildEventTool" + /> + </Configuration> + <Configuration + Name="Release|x64" + OutputDirectory="$(SolutionDir)bin\$(PlatformName)\$(ConfigurationName)" + IntermediateDirectory="$(PlatformName)\$(ConfigurationName)" + ConfigurationType="2" + CharacterSet="2" + WholeProgramOptimization="1" + > + <Tool + Name="VCPreBuildEventTool" + /> + <Tool + Name="VCCustomBuildTool" + /> + <Tool + Name="VCXMLDataGeneratorTool" + /> + <Tool + Name="VCWebServiceProxyGeneratorTool" + /> + <Tool + Name="VCMIDLTool" + TargetEnvironment="3" + /> + <Tool + Name="VCCLCompilerTool" + Optimization="2" + EnableIntrinsicFunctions="true" + OmitFramePointers="true" + AdditionalIncludeDirectories="$(SolutionDir)..\..\lib;$(SolutionDir)..\..\lib\common;$(SolutionDir)..\..\lib\legacy;$(SolutionDir)..\..\programs\legacy;$(SolutionDir)..\..\lib\dictBuilder" + PreprocessorDefinitions="ZSTD_DLL_EXPORT=1;ZSTD_MULTITHREAD=1;ZSTD_LEGACY_SUPPORT=4;WIN32;NDEBUG;_CONSOLE" + RuntimeLibrary="0" + EnableFunctionLevelLinking="true" + UsePrecompiledHeader="0" + WarningLevel="4" + DebugInformationFormat="3" + /> + <Tool + Name="VCManagedResourceCompilerTool" + /> + <Tool + Name="VCResourceCompilerTool" + /> + <Tool + Name="VCPreLinkEventTool" + /> + <Tool + Name="VCLinkerTool" + LinkIncremental="1" + GenerateDebugInformation="true" + SubSystem="1" + OptimizeReferences="2" + EnableCOMDATFolding="2" + TargetMachine="17" + /> + <Tool + Name="VCALinkTool" + /> + <Tool + Name="VCManifestTool" + /> + <Tool + Name="VCXDCMakeTool" + /> + <Tool + Name="VCBscMakeTool" + /> + <Tool + Name="VCFxCopTool" + /> + <Tool + Name="VCAppVerifierTool" + /> + <Tool + Name="VCPostBuildEventTool" + /> + </Configuration> + </Configurations> + <References> + </References> + <Files> + <Filter + Name="Source Files" + Filter="cpp;c;cc;cxx;def;odl;idl;hpj;bat;asm;asmx" + UniqueIdentifier="{4FC737F1-C7A5-4376-A066-2A32D752A2FF}" + > + <File + RelativePath="..\..\..\lib\dictBuilder\cover.c" + > + </File> + <File + RelativePath="..\..\..\lib\dictBuilder\divsufsort.c" + > + </File> + <File + RelativePath="..\..\..\lib\common\pool.c" + > + </File> + <File + RelativePath="..\..\..\lib\common\threading.c" + > + </File> + <File + RelativePath="..\..\..\lib\common\entropy_common.c" + > + </File> + <File + RelativePath="..\..\..\lib\common\error_private.c" + > + </File> + <File + RelativePath="..\..\..\lib\compress\fse_compress.c" + > + </File> + <File + RelativePath="..\..\..\lib\common\fse_decompress.c" + > + </File> + <File + RelativePath="..\..\..\lib\compress\huf_compress.c" + > + </File> + <File + RelativePath="..\..\..\lib\decompress\huf_decompress.c" + > + </File> + <File + RelativePath="..\..\..\lib\common\xxhash.c" + > + </File> + <File + RelativePath="..\..\..\lib\dictBuilder\zdict.c" + > + </File> + <File + RelativePath="..\..\..\lib\common\zstd_common.c" + > + </File> + <File + RelativePath="..\..\..\lib\compress\zstdmt_compress.c" + > + </File> + <File + RelativePath="..\..\..\lib\compress\zstd_compress.c" + > + </File> + <File + RelativePath="..\..\..\lib\decompress\zstd_decompress.c" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v01.c" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v02.c" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v03.c" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v04.c" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v05.c" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v06.c" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v07.c" + > + </File> + </Filter> + <Filter + Name="Header Files" + Filter="h;hpp;hxx;hm;inl;inc;xsd" + UniqueIdentifier="{93995380-89BD-4b04-88EB-625FBE52EBFB}" + > + <File + RelativePath="..\..\..\lib\common\bitstream.h" + > + </File> + <File + RelativePath="..\..\..\lib\dictBuilder\divsufsort.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\pool.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\threading.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\error_private.h" + > + </File> + <File + RelativePath="..\..\..\lib\zstd_errors.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\fse.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\fse_static.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\huf.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\huf_static.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\mem.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\xxhash.h" + > + </File> + <File + RelativePath="..\..\..\lib\zdict.h" + > + </File> + <File + RelativePath="..\..\..\lib\dictBuilder\zdict_static.h" + > + </File> + <File + RelativePath="..\..\..\lib\zstd.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\zstd_internal.h" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_legacy.h" + > + </File> + <File + RelativePath="..\..\..\lib\compress\zstd_opt.h" + > + </File> + <File + RelativePath="..\..\..\lib\compress\zstdmt_compress.h" + > + </File> + <File + RelativePath="..\..\..\lib\common\zstd_static.h" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v01.h" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v02.h" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v03.h" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v04.h" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v05.h" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v06.h" + > + </File> + <File + RelativePath="..\..\..\lib\legacy\zstd_v07.h" + > + </File> + </Filter> + <Filter + Name="Resource Files" + Filter="rc;ico;cur;bmp;dlg;rc2;rct;bin;rgs;gif;jpg;jpeg;jpe;resx;tiff;tif;png;wav" + UniqueIdentifier="{67DA6AB6-F800-4c08-8B7A-83BB121AAD01}" + > + </Filter> + </Files> + <Globals> + </Globals> +</VisualStudioProject> diff --git a/3rdparty/zstd/contrib/cleanTabs b/3rdparty/zstd/contrib/cleanTabs new file mode 100755 index 00000000000..215913a90ac --- /dev/null +++ b/3rdparty/zstd/contrib/cleanTabs @@ -0,0 +1,2 @@ +#!/bin/sh +sed -i '' $'s/\t/ /g' ../lib/**/*.{h,c} ../programs/*.{h,c} ../tests/*.c ./**/*.{h,cpp} ../examples/*.c ../zlibWrapper/*.{h,c} diff --git a/3rdparty/zstd/contrib/diagnose_corruption/.gitignore b/3rdparty/zstd/contrib/diagnose_corruption/.gitignore new file mode 100644 index 00000000000..a8e92b69b81 --- /dev/null +++ b/3rdparty/zstd/contrib/diagnose_corruption/.gitignore @@ -0,0 +1 @@ +check_flipped_bits diff --git a/3rdparty/zstd/contrib/diagnose_corruption/Makefile b/3rdparty/zstd/contrib/diagnose_corruption/Makefile new file mode 100644 index 00000000000..ecc9e63952c --- /dev/null +++ b/3rdparty/zstd/contrib/diagnose_corruption/Makefile @@ -0,0 +1,35 @@ +# ################################################################ +# Copyright (c) Meta Platforms, Inc. and affiliates. +# All rights reserved. +# +# This source code is licensed under both the BSD-style license (found in the +# LICENSE file in the root directory of this source tree) and the GPLv2 (found +# in the COPYING file in the root directory of this source tree). +# ################################################################ + +.PHONY: all +all: check_flipped_bits + +ZSTDLIBDIR ?= ../../lib + +CFLAGS ?= -O3 +CFLAGS += -I$(ZSTDLIBDIR) -I$(ZSTDLIBDIR)/common -I$(ZSTDLIBDIR)/compress \ + -I$(ZSTDLIBDIR)/decompress +CFLAGS += -Wall -Wextra -Wcast-qual -Wcast-align -Wshadow \ + -Wstrict-aliasing=1 -Wswitch-enum -Wdeclaration-after-statement \ + -Wstrict-prototypes -Wundef \ + -Wvla -Wformat=2 -Winit-self -Wfloat-equal -Wwrite-strings \ + -Wredundant-decls -Wmissing-prototypes +CFLAGS += $(DEBUGFLAGS) $(MOREFLAGS) +FLAGS = $(CPPFLAGS) $(CFLAGS) $(LDFLAGS) + +.PHONY: $(ZSTDLIBDIR)/libzstd.a +$(ZSTDLIBDIR)/libzstd.a: + $(MAKE) -C $(ZSTDLIBDIR) libzstd.a + +check_flipped_bits: check_flipped_bits.c $(ZSTDLIBDIR)/libzstd.a + $(CC) $(FLAGS) $< -o $@$(EXT) $(ZSTDLIBDIR)/libzstd.a + +.PHONY: clean +clean: + rm -f check_flipped_bits diff --git a/3rdparty/zstd/contrib/diagnose_corruption/check_flipped_bits.c b/3rdparty/zstd/contrib/diagnose_corruption/check_flipped_bits.c new file mode 100644 index 00000000000..09ddd467476 --- /dev/null +++ b/3rdparty/zstd/contrib/diagnose_corruption/check_flipped_bits.c @@ -0,0 +1,400 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ + +#define ZSTD_STATIC_LINKING_ONLY +#include "zstd.h" +#include "zstd_errors.h" + +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <sys/types.h> +#include <sys/stat.h> +#include <unistd.h> + +typedef struct { + char *input; + size_t input_size; + + char *perturbed; /* same size as input */ + + char *output; + size_t output_size; + + const char *dict_file_name; + const char *dict_file_dir_name; + int32_t dict_id; + char *dict; + size_t dict_size; + ZSTD_DDict* ddict; + + ZSTD_DCtx* dctx; + + int success_count; + int error_counts[ZSTD_error_maxCode]; +} stuff_t; + +static void free_stuff(stuff_t* stuff) { + free(stuff->input); + free(stuff->output); + ZSTD_freeDDict(stuff->ddict); + free(stuff->dict); + ZSTD_freeDCtx(stuff->dctx); +} + +static void usage(void) { + fprintf(stderr, "check_flipped_bits input_filename [-d dict] [-D dict_dir]\n"); + fprintf(stderr, "\n"); + fprintf(stderr, "Arguments:\n"); + fprintf(stderr, " -d file: path to a dictionary file to use.\n"); + fprintf(stderr, " -D dir : path to a directory, with files containing dictionaries, of the\n" + " form DICTID.zstd-dict, e.g., 12345.zstd-dict.\n"); + exit(1); +} + +static void print_summary(stuff_t* stuff) { + int error_code; + fprintf(stderr, "%9d successful decompressions\n", stuff->success_count); + for (error_code = 0; error_code < ZSTD_error_maxCode; error_code++) { + int count = stuff->error_counts[error_code]; + if (count) { + fprintf( + stderr, "%9d failed decompressions with message: %s\n", + count, ZSTD_getErrorString(error_code)); + } + } +} + +static char* readFile(const char* filename, size_t* size) { + struct stat statbuf; + int ret; + FILE* f; + char *buf; + size_t bytes_read; + + ret = stat(filename, &statbuf); + if (ret != 0) { + fprintf(stderr, "stat failed: %m\n"); + return NULL; + } + if ((statbuf.st_mode & S_IFREG) != S_IFREG) { + fprintf(stderr, "Input must be regular file\n"); + return NULL; + } + + *size = statbuf.st_size; + + f = fopen(filename, "r"); + if (f == NULL) { + fprintf(stderr, "fopen failed: %m\n"); + return NULL; + } + + buf = malloc(*size); + if (buf == NULL) { + fprintf(stderr, "malloc failed\n"); + fclose(f); + return NULL; + } + + bytes_read = fread(buf, 1, *size, f); + if (bytes_read != *size) { + fprintf(stderr, "failed to read whole file\n"); + fclose(f); + free(buf); + return NULL; + } + + ret = fclose(f); + if (ret != 0) { + fprintf(stderr, "fclose failed: %m\n"); + free(buf); + return NULL; + } + + return buf; +} + +static ZSTD_DDict* readDict(const char* filename, char **buf, size_t* size, int32_t* dict_id) { + ZSTD_DDict* ddict; + *buf = readFile(filename, size); + if (*buf == NULL) { + fprintf(stderr, "Opening dictionary file '%s' failed\n", filename); + return NULL; + } + + ddict = ZSTD_createDDict_advanced(*buf, *size, ZSTD_dlm_byRef, ZSTD_dct_auto, ZSTD_defaultCMem); + if (ddict == NULL) { + fprintf(stderr, "Failed to create ddict.\n"); + return NULL; + } + if (dict_id != NULL) { + *dict_id = ZSTD_getDictID_fromDDict(ddict); + } + return ddict; +} + +static ZSTD_DDict* readDictByID(stuff_t *stuff, int32_t dict_id, char **buf, size_t* size) { + if (stuff->dict_file_dir_name == NULL) { + return NULL; + } else { + size_t dir_name_len = strlen(stuff->dict_file_dir_name); + int dir_needs_separator = 0; + size_t dict_file_name_alloc_size = dir_name_len + 1 /* '/' */ + 10 /* max int32_t len */ + strlen(".zstd-dict") + 1 /* '\0' */; + char *dict_file_name = malloc(dict_file_name_alloc_size); + ZSTD_DDict* ddict; + int32_t read_dict_id; + if (dict_file_name == NULL) { + fprintf(stderr, "malloc failed.\n"); + return 0; + } + + if (dir_name_len > 0 && stuff->dict_file_dir_name[dir_name_len - 1] != '/') { + dir_needs_separator = 1; + } + + snprintf( + dict_file_name, + dict_file_name_alloc_size, + "%s%s%u.zstd-dict", + stuff->dict_file_dir_name, + dir_needs_separator ? "/" : "", + dict_id); + + /* fprintf(stderr, "Loading dict %u from '%s'.\n", dict_id, dict_file_name); */ + + ddict = readDict(dict_file_name, buf, size, &read_dict_id); + if (ddict == NULL) { + fprintf(stderr, "Failed to create ddict from '%s'.\n", dict_file_name); + free(dict_file_name); + return 0; + } + if (read_dict_id != dict_id) { + fprintf(stderr, "Read dictID (%u) does not match expected (%u).\n", read_dict_id, dict_id); + free(dict_file_name); + ZSTD_freeDDict(ddict); + return 0; + } + + free(dict_file_name); + return ddict; + } +} + +static int init_stuff(stuff_t* stuff, int argc, char *argv[]) { + const char* input_filename; + + if (argc < 2) { + usage(); + } + + input_filename = argv[1]; + stuff->input_size = 0; + stuff->input = readFile(input_filename, &stuff->input_size); + if (stuff->input == NULL) { + fprintf(stderr, "Failed to read input file.\n"); + return 0; + } + + stuff->perturbed = malloc(stuff->input_size); + if (stuff->perturbed == NULL) { + fprintf(stderr, "malloc failed.\n"); + return 0; + } + memcpy(stuff->perturbed, stuff->input, stuff->input_size); + + stuff->output_size = ZSTD_DStreamOutSize(); + stuff->output = malloc(stuff->output_size); + if (stuff->output == NULL) { + fprintf(stderr, "malloc failed.\n"); + return 0; + } + + stuff->dict_file_name = NULL; + stuff->dict_file_dir_name = NULL; + stuff->dict_id = 0; + stuff->dict = NULL; + stuff->dict_size = 0; + stuff->ddict = NULL; + + if (argc > 2) { + if (!strcmp(argv[2], "-d")) { + if (argc > 3) { + stuff->dict_file_name = argv[3]; + } else { + usage(); + } + } else + if (!strcmp(argv[2], "-D")) { + if (argc > 3) { + stuff->dict_file_dir_name = argv[3]; + } else { + usage(); + } + } else { + usage(); + } + } + + if (stuff->dict_file_dir_name) { + int32_t dict_id = ZSTD_getDictID_fromFrame(stuff->input, stuff->input_size); + if (dict_id != 0) { + stuff->ddict = readDictByID(stuff, dict_id, &stuff->dict, &stuff->dict_size); + if (stuff->ddict == NULL) { + fprintf(stderr, "Failed to create cached ddict.\n"); + return 0; + } + stuff->dict_id = dict_id; + } + } else + if (stuff->dict_file_name) { + stuff->ddict = readDict(stuff->dict_file_name, &stuff->dict, &stuff->dict_size, &stuff->dict_id); + if (stuff->ddict == NULL) { + fprintf(stderr, "Failed to create ddict from '%s'.\n", stuff->dict_file_name); + return 0; + } + } + + stuff->dctx = ZSTD_createDCtx(); + if (stuff->dctx == NULL) { + return 0; + } + + stuff->success_count = 0; + memset(stuff->error_counts, 0, sizeof(stuff->error_counts)); + + return 1; +} + +static int test_decompress(stuff_t* stuff) { + size_t ret; + ZSTD_inBuffer in = {stuff->perturbed, stuff->input_size, 0}; + ZSTD_outBuffer out = {stuff->output, stuff->output_size, 0}; + ZSTD_DCtx* dctx = stuff->dctx; + int32_t custom_dict_id = ZSTD_getDictID_fromFrame(in.src, in.size); + char *custom_dict = NULL; + size_t custom_dict_size = 0; + ZSTD_DDict* custom_ddict = NULL; + + if (custom_dict_id != 0 && custom_dict_id != stuff->dict_id) { + /* fprintf(stderr, "Instead of dict %u, this perturbed blob wants dict %u.\n", stuff->dict_id, custom_dict_id); */ + custom_ddict = readDictByID(stuff, custom_dict_id, &custom_dict, &custom_dict_size); + } + + ZSTD_DCtx_reset(dctx, ZSTD_reset_session_only); + + if (custom_ddict != NULL) { + ZSTD_DCtx_refDDict(dctx, custom_ddict); + } else { + ZSTD_DCtx_refDDict(dctx, stuff->ddict); + } + + while (in.pos != in.size) { + out.pos = 0; + ret = ZSTD_decompressStream(dctx, &out, &in); + + if (ZSTD_isError(ret)) { + unsigned int code = ZSTD_getErrorCode(ret); + if (code >= ZSTD_error_maxCode) { + fprintf(stderr, "Received unexpected error code!\n"); + exit(1); + } + stuff->error_counts[code]++; + /* + fprintf( + stderr, "Decompression failed: %s\n", ZSTD_getErrorName(ret)); + */ + if (custom_ddict != NULL) { + ZSTD_freeDDict(custom_ddict); + free(custom_dict); + } + return 0; + } + } + + stuff->success_count++; + + if (custom_ddict != NULL) { + ZSTD_freeDDict(custom_ddict); + free(custom_dict); + } + return 1; +} + +static int perturb_bits(stuff_t* stuff) { + size_t pos; + size_t bit; + for (pos = 0; pos < stuff->input_size; pos++) { + unsigned char old_val = stuff->input[pos]; + if (pos % 1000 == 0) { + fprintf(stderr, "Perturbing byte %zu / %zu\n", pos, stuff->input_size); + } + for (bit = 0; bit < 8; bit++) { + unsigned char new_val = old_val ^ (1 << bit); + stuff->perturbed[pos] = new_val; + if (test_decompress(stuff)) { + fprintf( + stderr, + "Flipping byte %zu bit %zu (0x%02x -> 0x%02x) " + "produced a successful decompression!\n", + pos, bit, old_val, new_val); + } + } + stuff->perturbed[pos] = old_val; + } + return 1; +} + +static int perturb_bytes(stuff_t* stuff) { + size_t pos; + size_t new_val; + for (pos = 0; pos < stuff->input_size; pos++) { + unsigned char old_val = stuff->input[pos]; + if (pos % 1000 == 0) { + fprintf(stderr, "Perturbing byte %zu / %zu\n", pos, stuff->input_size); + } + for (new_val = 0; new_val < 256; new_val++) { + stuff->perturbed[pos] = new_val; + if (test_decompress(stuff)) { + fprintf( + stderr, + "Changing byte %zu (0x%02x -> 0x%02x) " + "produced a successful decompression!\n", + pos, old_val, (unsigned char)new_val); + } + } + stuff->perturbed[pos] = old_val; + } + return 1; +} + +int main(int argc, char* argv[]) { + stuff_t stuff; + + if(!init_stuff(&stuff, argc, argv)) { + fprintf(stderr, "Failed to init.\n"); + return 1; + } + + if (test_decompress(&stuff)) { + fprintf(stderr, "Blob already decompresses successfully!\n"); + return 1; + } + + perturb_bits(&stuff); + + perturb_bytes(&stuff); + + print_summary(&stuff); + + free_stuff(&stuff); + + return 0; +} diff --git a/3rdparty/zstd/contrib/docker/Dockerfile b/3rdparty/zstd/contrib/docker/Dockerfile new file mode 100644 index 00000000000..912bf194977 --- /dev/null +++ b/3rdparty/zstd/contrib/docker/Dockerfile @@ -0,0 +1,20 @@ +# Dockerfile +# First image to build the binary +FROM alpine@sha256:69665d02cb32192e52e07644d76bc6f25abeb5410edc1c7a81a10ba3f0efb90a as builder + +RUN apk --no-cache add make gcc libc-dev +COPY . /src +RUN mkdir /pkg && cd /src && make && make DESTDIR=/pkg install + +# Second minimal image to only keep the built binary +FROM alpine@sha256:69665d02cb32192e52e07644d76bc6f25abeb5410edc1c7a81a10ba3f0efb90a + +# Copy the built files +COPY --from=builder /pkg / + +# Copy the license as well +RUN mkdir -p /usr/local/share/licenses/zstd +COPY --from=builder /src/LICENSE /usr/local/share/licences/zstd/ + +# Just run `zstd` if no other command is given +CMD ["/usr/local/bin/zstd"] diff --git a/3rdparty/zstd/contrib/docker/README.md b/3rdparty/zstd/contrib/docker/README.md new file mode 100644 index 00000000000..43f6d7a1ae1 --- /dev/null +++ b/3rdparty/zstd/contrib/docker/README.md @@ -0,0 +1,20 @@ + +## Requirement + +The `Dockerfile` script requires a version of `docker` >= 17.05 + +## Installing docker + +The official docker install docs use a ppa with a modern version available: +https://docs.docker.com/install/linux/docker-ce/ubuntu/ + +## How to run + +`docker build -t zstd .` + +## test + +``` +echo foo | docker run -i --rm zstd | docker run -i --rm zstd zstdcat +foo +``` diff --git a/3rdparty/zstd/contrib/externalSequenceProducer/.gitignore b/3rdparty/zstd/contrib/externalSequenceProducer/.gitignore new file mode 100644 index 00000000000..147710aee78 --- /dev/null +++ b/3rdparty/zstd/contrib/externalSequenceProducer/.gitignore @@ -0,0 +1,2 @@ +# build artifacts +externalSequenceProducer diff --git a/3rdparty/zstd/contrib/externalSequenceProducer/Makefile b/3rdparty/zstd/contrib/externalSequenceProducer/Makefile new file mode 100644 index 00000000000..0591ae01ba5 --- /dev/null +++ b/3rdparty/zstd/contrib/externalSequenceProducer/Makefile @@ -0,0 +1,40 @@ +# ################################################################ +# Copyright (c) Yann Collet, Meta Platforms, Inc. +# All rights reserved. +# +# This source code is licensed under both the BSD-style license (found in the +# LICENSE file in the root directory of this source tree) and the GPLv2 (found +# in the COPYING file in the root directory of this source tree). +# ################################################################ + +PROGDIR = ../../programs +LIBDIR = ../../lib + +LIBZSTD = $(LIBDIR)/libzstd.a + +CPPFLAGS+= -I$(LIBDIR) -I$(LIBDIR)/compress -I$(LIBDIR)/common + +CFLAGS ?= -O3 +CFLAGS += -std=gnu99 +DEBUGFLAGS= -Wall -Wextra -Wcast-qual -Wcast-align -Wshadow \ + -Wstrict-aliasing=1 -Wswitch-enum \ + -Wstrict-prototypes -Wundef -Wpointer-arith \ + -Wvla -Wformat=2 -Winit-self -Wfloat-equal -Wwrite-strings \ + -Wredundant-decls +CFLAGS += $(DEBUGFLAGS) $(MOREFLAGS) + +default: externalSequenceProducer + +all: externalSequenceProducer + +externalSequenceProducer: sequence_producer.c main.c $(LIBZSTD) + $(CC) $(CPPFLAGS) $(CFLAGS) $^ $(LDFLAGS) -o $@ + +.PHONY: $(LIBZSTD) +$(LIBZSTD): + $(MAKE) -C $(LIBDIR) libzstd.a CFLAGS="$(CFLAGS)" + +clean: + $(RM) *.o + $(MAKE) -C $(LIBDIR) clean > /dev/null + $(RM) externalSequenceProducer diff --git a/3rdparty/zstd/contrib/externalSequenceProducer/README.md b/3rdparty/zstd/contrib/externalSequenceProducer/README.md new file mode 100644 index 00000000000..c16a170073b --- /dev/null +++ b/3rdparty/zstd/contrib/externalSequenceProducer/README.md @@ -0,0 +1,14 @@ +externalSequenceProducer +===================== + +`externalSequenceProducer` is a test tool for the Block-Level Sequence Producer API. +It demonstrates how to use the API to perform a simple round-trip test. + +A sample sequence producer is provided in sequence_producer.c, but the user can swap +this out with a different one if desired. The sample sequence producer implements +LZ parsing with a 1KB hashtable. Dictionary-based parsing is not currently supported. + +Command line : +``` +externalSequenceProducer filename +``` diff --git a/3rdparty/zstd/contrib/externalSequenceProducer/main.c b/3rdparty/zstd/contrib/externalSequenceProducer/main.c new file mode 100644 index 00000000000..e67e295383c --- /dev/null +++ b/3rdparty/zstd/contrib/externalSequenceProducer/main.c @@ -0,0 +1,107 @@ +/* + * Copyright (c) Yann Collet, Meta Platforms, Inc. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ + +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <assert.h> + +#define ZSTD_STATIC_LINKING_ONLY +#include "zstd.h" +#include "zstd_errors.h" +#include "sequence_producer.h" // simpleSequenceProducer + +#define CHECK(res) \ +do { \ + if (ZSTD_isError(res)) { \ + printf("ERROR: %s\n", ZSTD_getErrorName(res)); \ + return 1; \ + } \ +} while (0) \ + +int main(int argc, char *argv[]) { + if (argc != 2) { + printf("Usage: externalSequenceProducer <file>\n"); + return 1; + } + + ZSTD_CCtx* const zc = ZSTD_createCCtx(); + + int simpleSequenceProducerState = 0xdeadbeef; + + // Here is the crucial bit of code! + ZSTD_registerSequenceProducer( + zc, + &simpleSequenceProducerState, + simpleSequenceProducer + ); + + { + size_t const res = ZSTD_CCtx_setParameter(zc, ZSTD_c_enableSeqProducerFallback, 1); + CHECK(res); + } + + FILE *f = fopen(argv[1], "rb"); + assert(f); + { + int const ret = fseek(f, 0, SEEK_END); + assert(ret == 0); + } + size_t const srcSize = ftell(f); + { + int const ret = fseek(f, 0, SEEK_SET); + assert(ret == 0); + } + + char* const src = malloc(srcSize + 1); + assert(src); + { + size_t const ret = fread(src, srcSize, 1, f); + assert(ret == 1); + int const ret2 = fclose(f); + assert(ret2 == 0); + } + + size_t const dstSize = ZSTD_compressBound(srcSize); + char* const dst = malloc(dstSize); + assert(dst); + + size_t const cSize = ZSTD_compress2(zc, dst, dstSize, src, srcSize); + CHECK(cSize); + + char* const val = malloc(srcSize); + assert(val); + + { + size_t const res = ZSTD_decompress(val, srcSize, dst, cSize); + CHECK(res); + } + + if (memcmp(src, val, srcSize) == 0) { + printf("Compression and decompression were successful!\n"); + printf("Original size: %lu\n", srcSize); + printf("Compressed size: %lu\n", cSize); + } else { + printf("ERROR: input and validation buffers don't match!\n"); + for (size_t i = 0; i < srcSize; i++) { + if (src[i] != val[i]) { + printf("First bad index: %zu\n", i); + break; + } + } + return 1; + } + + ZSTD_freeCCtx(zc); + free(src); + free(dst); + free(val); + return 0; +} diff --git a/3rdparty/zstd/contrib/externalSequenceProducer/sequence_producer.c b/3rdparty/zstd/contrib/externalSequenceProducer/sequence_producer.c new file mode 100644 index 00000000000..60a2f9572b0 --- /dev/null +++ b/3rdparty/zstd/contrib/externalSequenceProducer/sequence_producer.c @@ -0,0 +1,80 @@ +/* + * Copyright (c) Yann Collet, Meta Platforms, Inc. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ + +#include "zstd_compress_internal.h" +#include "sequence_producer.h" + +#define HSIZE 1024 +static U32 const HLOG = 10; +static U32 const MLS = 4; +static U32 const BADIDX = 0xffffffff; + +size_t simpleSequenceProducer( + void* sequenceProducerState, + ZSTD_Sequence* outSeqs, size_t outSeqsCapacity, + const void* src, size_t srcSize, + const void* dict, size_t dictSize, + int compressionLevel, + size_t windowSize +) { + const BYTE* const istart = (const BYTE*)src; + const BYTE* const iend = istart + srcSize; + const BYTE* ip = istart; + const BYTE* anchor = istart; + size_t seqCount = 0; + U32 hashTable[HSIZE]; + + (void)sequenceProducerState; + (void)dict; + (void)dictSize; + (void)outSeqsCapacity; + (void)compressionLevel; + + { int i; + for (i=0; i < HSIZE; i++) { + hashTable[i] = BADIDX; + } } + + while (ip + MLS < iend) { + size_t const hash = ZSTD_hashPtr(ip, HLOG, MLS); + U32 const matchIndex = hashTable[hash]; + hashTable[hash] = (U32)(ip - istart); + + if (matchIndex != BADIDX) { + const BYTE* const match = istart + matchIndex; + U32 const matchLen = (U32)ZSTD_count(ip, match, iend); + if (matchLen >= ZSTD_MINMATCH_MIN) { + U32 const litLen = (U32)(ip - anchor); + U32 const offset = (U32)(ip - match); + ZSTD_Sequence const seq = { + offset, litLen, matchLen, 0 + }; + + /* Note: it's crucial to stay within the window size! */ + if (offset <= windowSize) { + outSeqs[seqCount++] = seq; + ip += matchLen; + anchor = ip; + continue; + } + } + } + + ip++; + } + + { ZSTD_Sequence const finalSeq = { + 0, (U32)(iend - anchor), 0, 0 + }; + outSeqs[seqCount++] = finalSeq; + } + + return seqCount; +} diff --git a/3rdparty/zstd/contrib/externalSequenceProducer/sequence_producer.h b/3rdparty/zstd/contrib/externalSequenceProducer/sequence_producer.h new file mode 100644 index 00000000000..19f9982ac36 --- /dev/null +++ b/3rdparty/zstd/contrib/externalSequenceProducer/sequence_producer.h @@ -0,0 +1,26 @@ +/* + * Copyright (c) Yann Collet, Meta Platforms, Inc. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ + +#ifndef MATCHFINDER_H +#define MATCHFINDER_H + +#define ZSTD_STATIC_LINKING_ONLY +#include "zstd.h" + +size_t simpleSequenceProducer( + void* sequenceProducerState, + ZSTD_Sequence* outSeqs, size_t outSeqsCapacity, + const void* src, size_t srcSize, + const void* dict, size_t dictSize, + int compressionLevel, + size_t windowSize +); + +#endif diff --git a/3rdparty/zstd/contrib/freestanding_lib/freestanding.py b/3rdparty/zstd/contrib/freestanding_lib/freestanding.py new file mode 100755 index 00000000000..df6983245d6 --- /dev/null +++ b/3rdparty/zstd/contrib/freestanding_lib/freestanding.py @@ -0,0 +1,774 @@ +#!/usr/bin/env python3 +# ################################################################ +# Copyright (c) Meta Platforms, Inc. and affiliates. +# All rights reserved. +# +# This source code is licensed under both the BSD-style license (found in the +# LICENSE file in the root directory of this source tree) and the GPLv2 (found +# in the COPYING file in the root directory of this source tree). +# You may select, at your option, one of the above-listed licenses. +# ########################################################################## + +import argparse +import contextlib +import os +import re +import shutil +import sys +from typing import Optional + + +INCLUDED_SUBDIRS = ["common", "compress", "decompress"] + +SKIPPED_FILES = [ + "common/mem.h", + "common/zstd_deps.h", + "common/pool.c", + "common/pool.h", + "common/threading.c", + "common/threading.h", + "common/zstd_trace.h", + "compress/zstdmt_compress.h", + "compress/zstdmt_compress.c", +] + +XXHASH_FILES = [ + "common/xxhash.c", + "common/xxhash.h", +] + + +class FileLines(object): + def __init__(self, filename): + self.filename = filename + with open(self.filename, "r") as f: + self.lines = f.readlines() + + def write(self): + with open(self.filename, "w") as f: + f.write("".join(self.lines)) + + +class PartialPreprocessor(object): + """ + Looks for simple ifdefs and ifndefs and replaces them. + Handles && and ||. + Has fancy logic to handle translating elifs to ifs. + Only looks for macros in the first part of the expression with no + parens. + Does not handle multi-line macros (only looks in first line). + """ + def __init__(self, defs: [(str, Optional[str])], replaces: [(str, str)], undefs: [str]): + MACRO_GROUP = r"(?P<macro>[a-zA-Z_][a-zA-Z_0-9]*)" + ELIF_GROUP = r"(?P<elif>el)?" + OP_GROUP = r"(?P<op>&&|\|\|)?" + + self._defs = {macro:value for macro, value in defs} + self._replaces = {macro:value for macro, value in replaces} + self._defs.update(self._replaces) + self._undefs = set(undefs) + + self._define = re.compile(r"\s*#\s*define") + self._if = re.compile(r"\s*#\s*if") + self._elif = re.compile(r"\s*#\s*(?P<elif>el)if") + self._else = re.compile(r"\s*#\s*(?P<else>else)") + self._endif = re.compile(r"\s*#\s*endif") + + self._ifdef = re.compile(fr"\s*#\s*if(?P<not>n)?def {MACRO_GROUP}\s*") + self._if_defined = re.compile( + fr"\s*#\s*{ELIF_GROUP}if\s+(?P<not>!)?\s*defined\s*\(\s*{MACRO_GROUP}\s*\)\s*{OP_GROUP}" + ) + self._if_defined_value = re.compile( + fr"\s*#\s*{ELIF_GROUP}if\s+defined\s*\(\s*{MACRO_GROUP}\s*\)\s*" + fr"(?P<op>&&)\s*" + fr"(?P<openp>\()?\s*" + fr"(?P<macro2>[a-zA-Z_][a-zA-Z_0-9]*)\s*" + fr"(?P<cmp>[=><!]+)\s*" + fr"(?P<value>[0-9]*)\s*" + fr"(?P<closep>\))?\s*" + ) + self._if_true = re.compile( + fr"\s*#\s*{ELIF_GROUP}if\s+{MACRO_GROUP}\s*{OP_GROUP}" + ) + + self._c_comment = re.compile(r"/\*.*?\*/") + self._cpp_comment = re.compile(r"//") + + def _log(self, *args, **kwargs): + print(*args, **kwargs) + + def _strip_comments(self, line): + # First strip c-style comments (may include //) + while True: + m = self._c_comment.search(line) + if m is None: + break + line = line[:m.start()] + line[m.end():] + + # Then strip cpp-style comments + m = self._cpp_comment.search(line) + if m is not None: + line = line[:m.start()] + + return line + + def _fixup_indentation(self, macro, replace: [str]): + if len(replace) == 0: + return replace + if len(replace) == 1 and self._define.match(replace[0]) is None: + # If there is only one line, only replace defines + return replace + + + all_pound = True + for line in replace: + if not line.startswith('#'): + all_pound = False + if all_pound: + replace = [line[1:] for line in replace] + + min_spaces = len(replace[0]) + for line in replace: + spaces = 0 + for i, c in enumerate(line): + if c != ' ': + # Non-preprocessor line ==> skip the fixup + if not all_pound and c != '#': + return replace + spaces = i + break + min_spaces = min(min_spaces, spaces) + + replace = [line[min_spaces:] for line in replace] + + if all_pound: + replace = ["#" + line for line in replace] + + return replace + + def _handle_if_block(self, macro, idx, is_true, prepend): + """ + Remove the #if or #elif block starting on this line. + """ + REMOVE_ONE = 0 + KEEP_ONE = 1 + REMOVE_REST = 2 + + if is_true: + state = KEEP_ONE + else: + state = REMOVE_ONE + + line = self._inlines[idx] + is_if = self._if.match(line) is not None + assert is_if or self._elif.match(line) is not None + depth = 0 + + start_idx = idx + + idx += 1 + replace = prepend + finished = False + while idx < len(self._inlines): + line = self._inlines[idx] + # Nested if statement + if self._if.match(line): + depth += 1 + idx += 1 + continue + # We're inside a nested statement + if depth > 0: + if self._endif.match(line): + depth -= 1 + idx += 1 + continue + + # We're at the original depth + + # Looking only for an endif. + # We've found a true statement, but haven't + # completely elided the if block, so we just + # remove the remainder. + if state == REMOVE_REST: + if self._endif.match(line): + if is_if: + # Remove the endif because we took the first if + idx += 1 + finished = True + break + idx += 1 + continue + + if state == KEEP_ONE: + m = self._elif.match(line) + if self._endif.match(line): + replace += self._inlines[start_idx + 1:idx] + idx += 1 + finished = True + break + if self._elif.match(line) or self._else.match(line): + replace += self._inlines[start_idx + 1:idx] + state = REMOVE_REST + idx += 1 + continue + + if state == REMOVE_ONE: + m = self._elif.match(line) + if m is not None: + if is_if: + idx += 1 + b = m.start('elif') + e = m.end('elif') + assert e - b == 2 + replace.append(line[:b] + line[e:]) + finished = True + break + m = self._else.match(line) + if m is not None: + if is_if: + idx += 1 + while self._endif.match(self._inlines[idx]) is None: + replace.append(self._inlines[idx]) + idx += 1 + idx += 1 + finished = True + break + if self._endif.match(line): + if is_if: + # Remove the endif because no other elifs + idx += 1 + finished = True + break + idx += 1 + continue + if not finished: + raise RuntimeError("Unterminated if block!") + + replace = self._fixup_indentation(macro, replace) + + self._log(f"\tHardwiring {macro}") + if start_idx > 0: + self._log(f"\t\t {self._inlines[start_idx - 1][:-1]}") + for x in range(start_idx, idx): + self._log(f"\t\t- {self._inlines[x][:-1]}") + for line in replace: + self._log(f"\t\t+ {line[:-1]}") + if idx < len(self._inlines): + self._log(f"\t\t {self._inlines[idx][:-1]}") + + return idx, replace + + def _preprocess_once(self): + outlines = [] + idx = 0 + changed = False + while idx < len(self._inlines): + line = self._inlines[idx] + sline = self._strip_comments(line) + m = self._ifdef.fullmatch(sline) + if_true = False + if m is None: + m = self._if_defined_value.fullmatch(sline) + if m is None: + m = self._if_defined.match(sline) + if m is None: + m = self._if_true.match(sline) + if_true = (m is not None) + if m is None: + outlines.append(line) + idx += 1 + continue + + groups = m.groupdict() + macro = groups['macro'] + op = groups.get('op') + + if not (macro in self._defs or macro in self._undefs): + outlines.append(line) + idx += 1 + continue + + defined = macro in self._defs + + # Needed variables set: + # resolved: Is the statement fully resolved? + # is_true: If resolved, is the statement true? + ifdef = False + if if_true: + if not defined: + outlines.append(line) + idx += 1 + continue + + defined_value = self._defs[macro] + is_int = True + try: + defined_value = int(defined_value) + except TypeError: + is_int = False + except ValueError: + is_int = False + + resolved = is_int + is_true = (defined_value != 0) + + if resolved and op is not None: + if op == '&&': + resolved = not is_true + else: + assert op == '||' + resolved = is_true + + else: + ifdef = groups.get('not') is None + elseif = groups.get('elif') is not None + + macro2 = groups.get('macro2') + cmp = groups.get('cmp') + value = groups.get('value') + openp = groups.get('openp') + closep = groups.get('closep') + + is_true = (ifdef == defined) + resolved = True + if op is not None: + if op == '&&': + resolved = not is_true + else: + assert op == '||' + resolved = is_true + + if macro2 is not None and not resolved: + assert ifdef and defined and op == '&&' and cmp is not None + # If the statement is true, but we have a single value check, then + # check the value. + defined_value = self._defs[macro] + are_ints = True + try: + defined_value = int(defined_value) + value = int(value) + except TypeError: + are_ints = False + except ValueError: + are_ints = False + if ( + macro == macro2 and + ((openp is None) == (closep is None)) and + are_ints + ): + resolved = True + if cmp == '<': + is_true = defined_value < value + elif cmp == '<=': + is_true = defined_value <= value + elif cmp == '==': + is_true = defined_value == value + elif cmp == '!=': + is_true = defined_value != value + elif cmp == '>=': + is_true = defined_value >= value + elif cmp == '>': + is_true = defined_value > value + else: + resolved = False + + if op is not None and not resolved: + # Remove the first op in the line + spaces + if op == '&&': + opre = op + else: + assert op == '||' + opre = r'\|\|' + needle = re.compile(fr"(?P<if>\s*#\s*(el)?if\s+).*?(?P<op>{opre}\s*)") + match = needle.match(line) + assert match is not None + newline = line[:match.end('if')] + line[match.end('op'):] + + self._log(f"\tHardwiring partially resolved {macro}") + self._log(f"\t\t- {line[:-1]}") + self._log(f"\t\t+ {newline[:-1]}") + + outlines.append(newline) + idx += 1 + continue + + # Skip any statements we cannot fully compute + if not resolved: + outlines.append(line) + idx += 1 + continue + + prepend = [] + if macro in self._replaces: + assert not ifdef + assert op is None + value = self._replaces.pop(macro) + prepend = [f"#define {macro} {value}\n"] + + idx, replace = self._handle_if_block(macro, idx, is_true, prepend) + outlines += replace + changed = True + + return changed, outlines + + def preprocess(self, filename): + with open(filename, 'r') as f: + self._inlines = f.readlines() + changed = True + iters = 0 + while changed: + iters += 1 + changed, outlines = self._preprocess_once() + self._inlines = outlines + + with open(filename, 'w') as f: + f.write(''.join(self._inlines)) + + +class Freestanding(object): + def __init__( + self, zstd_deps: str, mem: str, source_lib: str, output_lib: str, + external_xxhash: bool, xxh64_state: Optional[str], + xxh64_prefix: Optional[str], rewritten_includes: [(str, str)], + defs: [(str, Optional[str])], replaces: [(str, str)], + undefs: [str], excludes: [str], seds: [str], spdx: bool, + ): + self._zstd_deps = zstd_deps + self._mem = mem + self._src_lib = source_lib + self._dst_lib = output_lib + self._external_xxhash = external_xxhash + self._xxh64_state = xxh64_state + self._xxh64_prefix = xxh64_prefix + self._rewritten_includes = rewritten_includes + self._defs = defs + self._replaces = replaces + self._undefs = undefs + self._excludes = excludes + self._seds = seds + self._spdx = spdx + + def _dst_lib_file_paths(self): + """ + Yields all the file paths in the dst_lib. + """ + for root, dirname, filenames in os.walk(self._dst_lib): + for filename in filenames: + filepath = os.path.join(root, filename) + yield filepath + + def _log(self, *args, **kwargs): + print(*args, **kwargs) + + def _copy_file(self, lib_path): + suffixes = [".c", ".h", ".S"] + if not any((lib_path.endswith(suffix) for suffix in suffixes)): + return + if lib_path in SKIPPED_FILES: + self._log(f"\tSkipping file: {lib_path}") + return + if self._external_xxhash and lib_path in XXHASH_FILES: + self._log(f"\tSkipping xxhash file: {lib_path}") + return + + src_path = os.path.join(self._src_lib, lib_path) + dst_path = os.path.join(self._dst_lib, lib_path) + self._log(f"\tCopying: {src_path} -> {dst_path}") + shutil.copyfile(src_path, dst_path) + + def _copy_source_lib(self): + self._log("Copying source library into output library") + + assert os.path.exists(self._src_lib) + os.makedirs(self._dst_lib, exist_ok=True) + self._copy_file("zstd.h") + self._copy_file("zstd_errors.h") + for subdir in INCLUDED_SUBDIRS: + src_dir = os.path.join(self._src_lib, subdir) + dst_dir = os.path.join(self._dst_lib, subdir) + + assert os.path.exists(src_dir) + os.makedirs(dst_dir, exist_ok=True) + + for filename in os.listdir(src_dir): + lib_path = os.path.join(subdir, filename) + self._copy_file(lib_path) + + def _copy_zstd_deps(self): + dst_zstd_deps = os.path.join(self._dst_lib, "common", "zstd_deps.h") + self._log(f"Copying zstd_deps: {self._zstd_deps} -> {dst_zstd_deps}") + shutil.copyfile(self._zstd_deps, dst_zstd_deps) + + def _copy_mem(self): + dst_mem = os.path.join(self._dst_lib, "common", "mem.h") + self._log(f"Copying mem: {self._mem} -> {dst_mem}") + shutil.copyfile(self._mem, dst_mem) + + def _hardwire_preprocessor(self, name: str, value: Optional[str] = None, undef=False): + """ + If value=None then hardwire that it is defined, but not what the value is. + If undef=True then value must be None. + If value='' then the macro is defined to '' exactly. + """ + assert not (undef and value is not None) + for filepath in self._dst_lib_file_paths(): + file = FileLines(filepath) + + def _hardwire_defines(self): + self._log("Hardwiring macros") + partial_preprocessor = PartialPreprocessor(self._defs, self._replaces, self._undefs) + for filepath in self._dst_lib_file_paths(): + partial_preprocessor.preprocess(filepath) + + def _remove_excludes(self): + self._log("Removing excluded sections") + for exclude in self._excludes: + self._log(f"\tRemoving excluded sections for: {exclude}") + begin_re = re.compile(f"BEGIN {exclude}") + end_re = re.compile(f"END {exclude}") + for filepath in self._dst_lib_file_paths(): + file = FileLines(filepath) + outlines = [] + skipped = [] + emit = True + for line in file.lines: + if emit and begin_re.search(line) is not None: + assert end_re.search(line) is None + emit = False + if emit: + outlines.append(line) + else: + skipped.append(line) + if end_re.search(line) is not None: + assert begin_re.search(line) is None + self._log(f"\t\tRemoving excluded section: {exclude}") + for s in skipped: + self._log(f"\t\t\t- {s}") + emit = True + skipped = [] + if not emit: + raise RuntimeError("Excluded section unfinished!") + file.lines = outlines + file.write() + + def _rewrite_include(self, original, rewritten): + self._log(f"\tRewriting include: {original} -> {rewritten}") + regex = re.compile(f"\\s*#\\s*include\\s*(?P<include>{original})") + for filepath in self._dst_lib_file_paths(): + file = FileLines(filepath) + for i, line in enumerate(file.lines): + match = regex.match(line) + if match is None: + continue + s = match.start('include') + e = match.end('include') + file.lines[i] = line[:s] + rewritten + line[e:] + file.write() + + def _rewrite_includes(self): + self._log("Rewriting includes") + for original, rewritten in self._rewritten_includes: + self._rewrite_include(original, rewritten) + + def _replace_xxh64_prefix(self): + if self._xxh64_prefix is None: + return + self._log(f"Replacing XXH64 prefix with {self._xxh64_prefix}") + replacements = [] + if self._xxh64_state is not None: + replacements.append( + (re.compile(r"([^\w]|^)(?P<orig>XXH64_state_t)([^\w]|$)"), self._xxh64_state) + ) + if self._xxh64_prefix is not None: + replacements.append( + (re.compile(r"([^\w]|^)(?P<orig>XXH64)[\(_]"), self._xxh64_prefix) + ) + for filepath in self._dst_lib_file_paths(): + file = FileLines(filepath) + for i, line in enumerate(file.lines): + modified = False + for regex, replacement in replacements: + match = regex.search(line) + while match is not None: + modified = True + b = match.start('orig') + e = match.end('orig') + line = line[:b] + replacement + line[e:] + match = regex.search(line) + if modified: + self._log(f"\t- {file.lines[i][:-1]}") + self._log(f"\t+ {line[:-1]}") + file.lines[i] = line + file.write() + + def _parse_sed(self, sed): + assert sed[0] == 's' + delim = sed[1] + match = re.fullmatch(f's{delim}(.+){delim}(.*){delim}(.*)', sed) + assert match is not None + regex = re.compile(match.group(1)) + format_str = match.group(2) + is_global = match.group(3) == 'g' + return regex, format_str, is_global + + def _process_sed(self, sed): + self._log(f"Processing sed: {sed}") + regex, format_str, is_global = self._parse_sed(sed) + + for filepath in self._dst_lib_file_paths(): + file = FileLines(filepath) + for i, line in enumerate(file.lines): + modified = False + while True: + match = regex.search(line) + if match is None: + break + replacement = format_str.format(match.groups(''), match.groupdict('')) + b = match.start() + e = match.end() + line = line[:b] + replacement + line[e:] + modified = True + if not is_global: + break + if modified: + self._log(f"\t- {file.lines[i][:-1]}") + self._log(f"\t+ {line[:-1]}") + file.lines[i] = line + file.write() + + def _process_seds(self): + self._log("Processing seds") + for sed in self._seds: + self._process_sed(sed) + + def _process_spdx(self): + if not self._spdx: + return + self._log("Processing spdx") + SPDX_C = "// SPDX-License-Identifier: GPL-2.0+ OR BSD-3-Clause\n" + SPDX_H_S = "/* SPDX-License-Identifier: GPL-2.0+ OR BSD-3-Clause */\n" + for filepath in self._dst_lib_file_paths(): + file = FileLines(filepath) + if file.lines[0] == SPDX_C or file.lines[0] == SPDX_H_S: + continue + for line in file.lines: + if "SPDX-License-Identifier" in line: + raise RuntimeError(f"Unexpected SPDX license identifier: {file.filename} {repr(line)}") + if file.filename.endswith(".c"): + file.lines.insert(0, SPDX_C) + elif file.filename.endswith(".h") or file.filename.endswith(".S"): + file.lines.insert(0, SPDX_H_S) + else: + raise RuntimeError(f"Unexpected file extension: {file.filename}") + file.write() + + + + def go(self): + self._copy_source_lib() + self._copy_zstd_deps() + self._copy_mem() + self._hardwire_defines() + self._remove_excludes() + self._rewrite_includes() + self._replace_xxh64_prefix() + self._process_seds() + self._process_spdx() + + +def parse_optional_pair(defines: [str]) -> [(str, Optional[str])]: + output = [] + for define in defines: + parsed = define.split('=') + if len(parsed) == 1: + output.append((parsed[0], None)) + elif len(parsed) == 2: + output.append((parsed[0], parsed[1])) + else: + raise RuntimeError(f"Bad define: {define}") + return output + + +def parse_pair(rewritten_includes: [str]) -> [(str, str)]: + output = [] + for rewritten_include in rewritten_includes: + parsed = rewritten_include.split('=') + if len(parsed) == 2: + output.append((parsed[0], parsed[1])) + else: + raise RuntimeError(f"Bad rewritten include: {rewritten_include}") + return output + + + +def main(name, args): + parser = argparse.ArgumentParser(prog=name) + parser.add_argument("--zstd-deps", default="zstd_deps.h", help="Zstd dependencies file") + parser.add_argument("--mem", default="mem.h", help="Memory module") + parser.add_argument("--source-lib", default="../../lib", help="Location of the zstd library") + parser.add_argument("--output-lib", default="./freestanding_lib", help="Where to output the freestanding zstd library") + parser.add_argument("--xxhash", default=None, help="Alternate external xxhash include e.g. --xxhash='<xxhash.h>'. If set xxhash is not included.") + parser.add_argument("--xxh64-state", default=None, help="Alternate XXH64 state type (excluding _) e.g. --xxh64-state='struct xxh64_state'") + parser.add_argument("--xxh64-prefix", default=None, help="Alternate XXH64 function prefix (excluding _) e.g. --xxh64-prefix=xxh64") + parser.add_argument("--rewrite-include", default=[], dest="rewritten_includes", action="append", help="Rewrite an include REGEX=NEW (e.g. '<stddef\\.h>=<linux/types.h>')") + parser.add_argument("--sed", default=[], dest="seds", action="append", help="Apply a sed replacement. Format: `s/REGEX/FORMAT/[g]`. REGEX is a Python regex. FORMAT is a Python format string formatted by the regex dict.") + parser.add_argument("--spdx", action="store_true", help="Add SPDX License Identifiers") + parser.add_argument("-D", "--define", default=[], dest="defs", action="append", help="Pre-define this macro (can be passed multiple times)") + parser.add_argument("-U", "--undefine", default=[], dest="undefs", action="append", help="Pre-undefine this macro (can be passed multiple times)") + parser.add_argument("-R", "--replace", default=[], dest="replaces", action="append", help="Pre-define this macro and replace the first ifndef block with its definition") + parser.add_argument("-E", "--exclude", default=[], dest="excludes", action="append", help="Exclude all lines between 'BEGIN <EXCLUDE>' and 'END <EXCLUDE>'") + args = parser.parse_args(args) + + # Always remove threading + if "ZSTD_MULTITHREAD" not in args.undefs: + args.undefs.append("ZSTD_MULTITHREAD") + + args.defs = parse_optional_pair(args.defs) + for name, _ in args.defs: + if name in args.undefs: + raise RuntimeError(f"{name} is both defined and undefined!") + + # Always set tracing to 0 + if "ZSTD_NO_TRACE" not in (arg[0] for arg in args.defs): + args.defs.append(("ZSTD_NO_TRACE", None)) + args.defs.append(("ZSTD_TRACE", "0")) + + args.replaces = parse_pair(args.replaces) + for name, _ in args.replaces: + if name in args.undefs or name in args.defs: + raise RuntimeError(f"{name} is both replaced and (un)defined!") + + args.rewritten_includes = parse_pair(args.rewritten_includes) + + external_xxhash = False + if args.xxhash is not None: + external_xxhash = True + args.rewritten_includes.append(('"(\\.\\./common/)?xxhash.h"', args.xxhash)) + + if args.xxh64_prefix is not None: + if not external_xxhash: + raise RuntimeError("--xxh64-prefix may only be used with --xxhash provided") + + if args.xxh64_state is not None: + if not external_xxhash: + raise RuntimeError("--xxh64-state may only be used with --xxhash provided") + + Freestanding( + args.zstd_deps, + args.mem, + args.source_lib, + args.output_lib, + external_xxhash, + args.xxh64_state, + args.xxh64_prefix, + args.rewritten_includes, + args.defs, + args.replaces, + args.undefs, + args.excludes, + args.seds, + args.spdx, + ).go() + +if __name__ == "__main__": + main(sys.argv[0], sys.argv[1:]) diff --git a/3rdparty/zstd/contrib/gen_html/.gitignore b/3rdparty/zstd/contrib/gen_html/.gitignore new file mode 100644 index 00000000000..34461142872 --- /dev/null +++ b/3rdparty/zstd/contrib/gen_html/.gitignore @@ -0,0 +1,3 @@ +# make artefact +gen_html +zstd_manual.html diff --git a/3rdparty/zstd/contrib/gen_html/Makefile b/3rdparty/zstd/contrib/gen_html/Makefile new file mode 100644 index 00000000000..26e870399aa --- /dev/null +++ b/3rdparty/zstd/contrib/gen_html/Makefile @@ -0,0 +1,51 @@ +# ################################################################ +# Copyright (c) Meta Platforms, Inc. and affiliates. +# All rights reserved. +# +# This source code is licensed under both the BSD-style license (found in the +# LICENSE file in the root directory of this source tree) and the GPLv2 (found +# in the COPYING file in the root directory of this source tree). +# ################################################################ + +CXXFLAGS ?= -O3 +CXXFLAGS += -Wall -Wextra -Wcast-qual -Wcast-align -Wshadow -Wstrict-aliasing=1 -Wswitch-enum -Wno-comment +CXXFLAGS += $(MOREFLAGS) +FLAGS = $(CPPFLAGS) $(CXXFLAGS) $(LDFLAGS) + +ZSTDAPI = ../../lib/zstd.h +ZSTDMANUAL = ../../doc/zstd_manual.html +LIBVER_MAJOR_SCRIPT:=`sed -n '/define ZSTD_VERSION_MAJOR/s/.*[[:blank:]]\([0-9][0-9]*\).*/\1/p' < $(ZSTDAPI)` +LIBVER_MINOR_SCRIPT:=`sed -n '/define ZSTD_VERSION_MINOR/s/.*[[:blank:]]\([0-9][0-9]*\).*/\1/p' < $(ZSTDAPI)` +LIBVER_PATCH_SCRIPT:=`sed -n '/define ZSTD_VERSION_RELEASE/s/.*[[:blank:]]\([0-9][0-9]*\).*/\1/p' < $(ZSTDAPI)` +LIBVER_SCRIPT:= $(LIBVER_MAJOR_SCRIPT).$(LIBVER_MINOR_SCRIPT).$(LIBVER_PATCH_SCRIPT) +LIBVER := $(shell echo $(LIBVER_SCRIPT)) + + +# Define *.exe as extension for Windows systems +ifneq (,$(filter Windows%,$(OS))) +EXT =.exe +else +EXT = +endif + + +.PHONY: default +default: gen_html + +.PHONY: all +all: manual + +gen_html: gen_html.cpp + $(CXX) $(FLAGS) $^ -o $@$(EXT) + +$(ZSTDMANUAL): gen_html $(ZSTDAPI) + echo "Update zstd manual in /doc" + ./gen_html $(LIBVER) $(ZSTDAPI) $(ZSTDMANUAL) + +.PHONY: manual +manual: gen_html $(ZSTDMANUAL) + +.PHONY: clean +clean: + @$(RM) gen_html$(EXT) + @echo Cleaning completed diff --git a/3rdparty/zstd/contrib/gen_html/README.md b/3rdparty/zstd/contrib/gen_html/README.md new file mode 100644 index 00000000000..63a4caa2506 --- /dev/null +++ b/3rdparty/zstd/contrib/gen_html/README.md @@ -0,0 +1,31 @@ +gen_html - a program for automatic generation of zstd manual +============================================================ + +#### Introduction + +This simple C++ program generates a single-page HTML manual from `zstd.h`. + +The format of recognized comment blocks is following: +- comments of type `/*!` mean: this is a function declaration; switch comments with declarations +- comments of type `/**` and `/*-` mean: this is a comment; use a `<H2>` header for the first line +- comments of type `/*=` and `/**=` mean: use a `<H3>` header and show also all functions until first empty line +- comments of type `/*X` where `X` is different from above-mentioned are ignored + +Moreover: +- `ZSTDLIB_API` is removed to improve readability +- `typedef` are detected and included even if uncommented +- comments of type `/**<` and `/*!<` are detected and only function declaration is highlighted (bold) + + +#### Usage + +The program requires 3 parameters: +``` +gen_html [zstd_version] [input_file] [output_html] +``` + +To compile program and generate zstd manual we have used: +``` +make +./gen_html.exe 1.1.1 ../../lib/zstd.h zstd_manual.html +``` diff --git a/3rdparty/zstd/contrib/gen_html/gen-zstd-manual.sh b/3rdparty/zstd/contrib/gen_html/gen-zstd-manual.sh new file mode 100755 index 00000000000..57a8b6ea512 --- /dev/null +++ b/3rdparty/zstd/contrib/gen_html/gen-zstd-manual.sh @@ -0,0 +1,9 @@ +#!/bin/sh + +LIBVER_MAJOR_SCRIPT=`sed -n '/define ZSTD_VERSION_MAJOR/s/.*[[:blank:]]\([0-9][0-9]*\).*/\1/p' < ../../lib/zstd.h` +LIBVER_MINOR_SCRIPT=`sed -n '/define ZSTD_VERSION_MINOR/s/.*[[:blank:]]\([0-9][0-9]*\).*/\1/p' < ../../lib/zstd.h` +LIBVER_PATCH_SCRIPT=`sed -n '/define ZSTD_VERSION_RELEASE/s/.*[[:blank:]]\([0-9][0-9]*\).*/\1/p' < ../../lib/zstd.h` +LIBVER_SCRIPT=$LIBVER_MAJOR_SCRIPT.$LIBVER_MINOR_SCRIPT.$LIBVER_PATCH_SCRIPT + +echo ZSTD_VERSION=$LIBVER_SCRIPT +./gen_html $LIBVER_SCRIPT ../../lib/zstd.h ./zstd_manual.html diff --git a/3rdparty/zstd/contrib/gen_html/gen_html.cpp b/3rdparty/zstd/contrib/gen_html/gen_html.cpp new file mode 100644 index 00000000000..1da5879adae --- /dev/null +++ b/3rdparty/zstd/contrib/gen_html/gen_html.cpp @@ -0,0 +1,224 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ + +#include <iostream> +#include <fstream> +#include <sstream> +#include <vector> +using namespace std; + + +/* trim string at the beginning and at the end */ +void trim(string& s, string characters) +{ + size_t p = s.find_first_not_of(characters); + s.erase(0, p); + + p = s.find_last_not_of(characters); + if (string::npos != p) + s.erase(p+1); +} + + +/* trim C++ style comments */ +void trim_comments(string &s) +{ + size_t spos, epos; + + spos = s.find("/*"); + epos = s.find("*/"); + s = s.substr(spos+3, epos-(spos+3)); +} + + +/* get lines until a given terminator */ +vector<string> get_lines(vector<string>& input, int& linenum, string terminator) +{ + vector<string> out; + string line; + size_t epos; + + while ((size_t)linenum < input.size()) { + line = input[linenum]; + + if (terminator.empty() && line.empty()) { linenum--; break; } + + epos = line.find(terminator); + if (!terminator.empty() && epos!=string::npos) { + out.push_back(line); + break; + } + out.push_back(line); + linenum++; + } + return out; +} + + +/* print line with ZSTDLIB_API removed and C++ comments not bold */ +void print_line(stringstream &sout, string line) +{ + size_t spos; + + if (line.substr(0,12) == "ZSTDLIB_API ") line = line.substr(12); + spos = line.find("/*"); + if (spos!=string::npos) { + sout << line.substr(0, spos); + sout << "</b>" << line.substr(spos) << "<b>" << endl; + } else { + // fprintf(stderr, "lines=%s\n", line.c_str()); + sout << line << endl; + } +} + + +int main(int argc, char *argv[]) { + char exclam; + int linenum, chapter = 1; + vector<string> input, lines, comments, chapters; + string line, version; + size_t spos, l; + stringstream sout; + ifstream istream; + ofstream ostream; + + if (argc < 4) { + cout << "usage: " << argv[0] << " [zstd_version] [input_file] [output_html]" << endl; + return 1; + } + + version = "zstd " + string(argv[1]) + " Manual"; + + istream.open(argv[2], ifstream::in); + if (!istream.is_open()) { + cout << "Error opening file " << argv[2] << endl; + return 1; + } + + ostream.open(argv[3], ifstream::out); + if (!ostream.is_open()) { + cout << "Error opening file " << argv[3] << endl; + return 1; + } + + while (getline(istream, line)) { + input.push_back(line); + } + + for (linenum=0; (size_t)linenum < input.size(); linenum++) { + line = input[linenum]; + + /* typedefs are detected and included even if uncommented */ + if (line.substr(0,7) == "typedef" && line.find("{")!=string::npos) { + lines = get_lines(input, linenum, "}"); + sout << "<pre><b>"; + for (l=0; l<lines.size(); l++) { + print_line(sout, lines[l]); + } + sout << "</b></pre><BR>" << endl; + continue; + } + + /* comments of type /**< and /*!< are detected and only function declaration is highlighted (bold) */ + if ((line.find("/**<")!=string::npos || line.find("/*!<")!=string::npos) && line.find("*/")!=string::npos) { + sout << "<pre><b>"; + print_line(sout, line); + sout << "</b></pre><BR>" << endl; + continue; + } + + spos = line.find("/**="); + if (spos==string::npos) { + spos = line.find("/*!"); + if (spos==string::npos) + spos = line.find("/**"); + if (spos==string::npos) + spos = line.find("/*-"); + if (spos==string::npos) + spos = line.find("/*="); + if (spos==string::npos) + continue; + exclam = line[spos+2]; + } + else exclam = '='; + + comments = get_lines(input, linenum, "*/"); + if (!comments.empty()) comments[0] = line.substr(spos+3); + if (!comments.empty()) comments[comments.size()-1] = comments[comments.size()-1].substr(0, comments[comments.size()-1].find("*/")); + for (l=0; l<comments.size(); l++) { + if (comments[l].find(" *")==0) comments[l] = comments[l].substr(2); + else if (comments[l].find(" *")==0) comments[l] = comments[l].substr(3); + trim(comments[l], "*-="); + } + while (!comments.empty() && comments[comments.size()-1].empty()) comments.pop_back(); // remove empty line at the end + while (!comments.empty() && comments[0].empty()) comments.erase(comments.begin()); // remove empty line at the start + + /* comments of type /*! mean: this is a function declaration; switch comments with declarations */ + if (exclam == '!') { + if (!comments.empty()) comments.erase(comments.begin()); /* remove first line like "ZSTD_XXX() :" */ + linenum++; + lines = get_lines(input, linenum, ""); + + sout << "<pre><b>"; + for (l=0; l<lines.size(); l++) { + // fprintf(stderr, "line[%d]=%s\n", l, lines[l].c_str()); + string fline = lines[l]; + if (fline.substr(0, 12) == "ZSTDLIB_API " || + fline.substr(0, 12) == string(12, ' ')) + fline = fline.substr(12); + print_line(sout, fline); + } + sout << "</b><p>"; + for (l=0; l<comments.size(); l++) { + print_line(sout, comments[l]); + } + sout << "</p></pre><BR>" << endl << endl; + } else if (exclam == '=') { /* comments of type /*= and /**= mean: use a <H3> header and show also all functions until first empty line */ + trim(comments[0], " "); + sout << "<h3>" << comments[0] << "</h3><pre>"; + for (l=1; l<comments.size(); l++) { + print_line(sout, comments[l]); + } + sout << "</pre><b><pre>"; + lines = get_lines(input, ++linenum, ""); + for (l=0; l<lines.size(); l++) { + print_line(sout, lines[l]); + } + sout << "</pre></b><BR>" << endl; + } else { /* comments of type /** and /*- mean: this is a comment; use a <H2> header for the first line */ + if (comments.empty()) continue; + + trim(comments[0], " "); + sout << "<a name=\"Chapter" << chapter << "\"></a><h2>" << comments[0] << "</h2><pre>"; + chapters.push_back(comments[0]); + chapter++; + + for (l=1; l<comments.size(); l++) { + print_line(sout, comments[l]); + } + if (comments.size() > 1) + sout << "<BR></pre>" << endl << endl; + else + sout << "</pre>" << endl << endl; + } + } + + ostream << "<html>\n<head>\n<meta http-equiv=\"Content-Type\" content=\"text/html; charset=ISO-8859-1\">\n<title>" << version << "</title>\n</head>\n<body>" << endl; + ostream << "<h1>" << version << "</h1>\n"; + + ostream << "<hr>\n<a name=\"Contents\"></a><h2>Contents</h2>\n<ol>\n"; + for (size_t i=0; i<chapters.size(); i++) + ostream << "<li><a href=\"#Chapter" << i+1 << "\">" << chapters[i].c_str() << "</a></li>\n"; + ostream << "</ol>\n<hr>\n"; + + ostream << sout.str(); + ostream << "</html>" << endl << "</body>" << endl; + + return 0; +} diff --git a/3rdparty/zstd/contrib/largeNbDicts/.gitignore b/3rdparty/zstd/contrib/largeNbDicts/.gitignore new file mode 100644 index 00000000000..e77c4e49679 --- /dev/null +++ b/3rdparty/zstd/contrib/largeNbDicts/.gitignore @@ -0,0 +1,2 @@ +# build artifacts +largeNbDicts diff --git a/3rdparty/zstd/contrib/largeNbDicts/Makefile b/3rdparty/zstd/contrib/largeNbDicts/Makefile new file mode 100644 index 00000000000..40734e62e47 --- /dev/null +++ b/3rdparty/zstd/contrib/largeNbDicts/Makefile @@ -0,0 +1,58 @@ +# ################################################################ +# Copyright (c) Meta Platforms, Inc. and affiliates. +# All rights reserved. +# +# This source code is licensed under both the BSD-style license (found in the +# LICENSE file in the root directory of this source tree) and the GPLv2 (found +# in the COPYING file in the root directory of this source tree). +# ################################################################ + +PROGDIR = ../../programs +LIBDIR = ../../lib + +LIBZSTD = $(LIBDIR)/libzstd.a + +CPPFLAGS+= -I$(LIBDIR) -I$(LIBDIR)/common -I$(LIBDIR)/dictBuilder -I$(PROGDIR) + +CFLAGS ?= -O3 +CFLAGS += -std=gnu99 +DEBUGFLAGS= -Wall -Wextra -Wcast-qual -Wcast-align -Wshadow \ + -Wstrict-aliasing=1 -Wswitch-enum \ + -Wstrict-prototypes -Wundef -Wpointer-arith \ + -Wvla -Wformat=2 -Winit-self -Wfloat-equal -Wwrite-strings \ + -Wredundant-decls +CFLAGS += $(DEBUGFLAGS) $(MOREFLAGS) + + +default: largeNbDicts + +all : largeNbDicts + +largeNbDicts: util.o timefn.o benchfn.o datagen.o xxhash.o largeNbDicts.c $(LIBZSTD) + $(CC) $(CPPFLAGS) $(CFLAGS) $^ $(LDFLAGS) -o $@ + +.PHONY: $(LIBZSTD) +$(LIBZSTD): + $(MAKE) -C $(LIBDIR) libzstd.a CFLAGS="$(CFLAGS)" + +benchfn.o: $(PROGDIR)/benchfn.c + $(CC) $(CPPFLAGS) $(CFLAGS) $^ -c + +timefn.o: $(PROGDIR)/timefn.c + $(CC) $(CPPFLAGS) $(CFLAGS) $^ -c + +datagen.o: $(PROGDIR)/datagen.c + $(CC) $(CPPFLAGS) $(CFLAGS) $^ -c + +util.o: $(PROGDIR)/util.c + $(CC) $(CPPFLAGS) $(CFLAGS) $^ -c + + +xxhash.o : $(LIBDIR)/common/xxhash.c + $(CC) $(CPPFLAGS) $(CFLAGS) $^ -c + + +clean: + $(RM) *.o + $(MAKE) -C $(LIBDIR) clean > /dev/null + $(RM) largeNbDicts diff --git a/3rdparty/zstd/contrib/largeNbDicts/README.md b/3rdparty/zstd/contrib/largeNbDicts/README.md new file mode 100644 index 00000000000..010102c904f --- /dev/null +++ b/3rdparty/zstd/contrib/largeNbDicts/README.md @@ -0,0 +1,33 @@ +largeNbDicts +===================== + +`largeNbDicts` is a benchmark test tool +dedicated to the specific scenario of +dictionary decompression using a very large number of dictionaries. +When dictionaries are constantly changing, they are always "cold", +suffering from increased latency due to cache misses. + +The tool is created in a bid to investigate performance for this scenario, +and experiment mitigation techniques. + +Command line : +``` +largeNbDicts [Options] filename(s) + +Options : +-z : benchmark compression (default) +-d : benchmark decompression +-r : recursively load all files in subdirectories (default: off) +-B# : split input into blocks of size # (default: no split) +-# : use compression level # (default: 3) +-D # : use # as a dictionary (default: create one) +-i# : nb benchmark rounds (default: 6) +--nbBlocks=#: use # blocks for bench (default: one per file) +--nbDicts=# : create # dictionaries for bench (default: one per block) +-h : help (this text) + +Advanced Options (see zstd.h for documentation) : +--dedicated-dict-search +--dict-content-type=# +--dict-attach-pref=# +``` diff --git a/3rdparty/zstd/contrib/largeNbDicts/largeNbDicts.c b/3rdparty/zstd/contrib/largeNbDicts/largeNbDicts.c new file mode 100644 index 00000000000..eeaaf7182e5 --- /dev/null +++ b/3rdparty/zstd/contrib/largeNbDicts/largeNbDicts.c @@ -0,0 +1,1085 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ + +/* largeNbDicts + * This is a benchmark test tool + * dedicated to the specific case of dictionary decompression + * using a very large nb of dictionaries + * thus suffering latency from lots of cache misses. + * It's created in a bid to investigate performance and find optimizations. */ + + +/*--- Dependencies ---*/ + +#include <stddef.h> /* size_t */ +#include <stdlib.h> /* malloc, free, abort, qsort*/ +#include <stdio.h> /* fprintf */ +#include <limits.h> /* UINT_MAX */ +#include <assert.h> /* assert */ + +#include "util.h" +#include "benchfn.h" +#define ZSTD_STATIC_LINKING_ONLY +#include "zstd.h" +#include "zdict.h" + + +/*--- Constants --- */ + +#define KB *(1<<10) +#define MB *(1<<20) + +#define BLOCKSIZE_DEFAULT 0 /* no slicing into blocks */ +#define DICTSIZE (4 KB) +#define CLEVEL_DEFAULT 3 +#define DICT_LOAD_METHOD ZSTD_dlm_byCopy + +#define BENCH_TIME_DEFAULT_S 6 +#define RUN_TIME_DEFAULT_MS 1000 +#define BENCH_TIME_DEFAULT_MS (BENCH_TIME_DEFAULT_S * RUN_TIME_DEFAULT_MS) + +#define DISPLAY_LEVEL_DEFAULT 3 + +#define BENCH_SIZE_MAX (1200 MB) + + +/*--- Macros ---*/ + +#define CONTROL(c) { if (!(c)) abort(); } +#undef MIN +#define MIN(a,b) ((a) < (b) ? (a) : (b)) + + +/*--- Display Macros ---*/ + +#define DISPLAY(...) fprintf(stdout, __VA_ARGS__) +#define DISPLAYLEVEL(l, ...) { if (g_displayLevel>=l) { DISPLAY(__VA_ARGS__); } } +static int g_displayLevel = DISPLAY_LEVEL_DEFAULT; /* 0 : no display, 1: errors, 2 : + result + interaction + warnings, 3 : + progression, 4 : + information */ + + +/*--- buffer_t ---*/ + +typedef struct { + void* ptr; + size_t size; + size_t capacity; +} buffer_t; + +static const buffer_t kBuffNull = { NULL, 0, 0 }; + +/* @return : kBuffNull if any error */ +static buffer_t createBuffer(size_t capacity) +{ + assert(capacity > 0); + void* const ptr = malloc(capacity); + if (ptr==NULL) return kBuffNull; + + buffer_t buffer; + buffer.ptr = ptr; + buffer.capacity = capacity; + buffer.size = 0; + return buffer; +} + +static void freeBuffer(buffer_t buff) +{ + free(buff.ptr); +} + + +static void fillBuffer_fromHandle(buffer_t* buff, FILE* f) +{ + size_t const readSize = fread(buff->ptr, 1, buff->capacity, f); + buff->size = readSize; +} + + +/* @return : kBuffNull if any error */ +static buffer_t createBuffer_fromFile(const char* fileName) +{ + U64 const fileSize = UTIL_getFileSize(fileName); + size_t const bufferSize = (size_t) fileSize; + + if (fileSize == UTIL_FILESIZE_UNKNOWN) return kBuffNull; + assert((U64)bufferSize == fileSize); /* check overflow */ + + { FILE* const f = fopen(fileName, "rb"); + if (f == NULL) return kBuffNull; + + buffer_t buff = createBuffer(bufferSize); + CONTROL(buff.ptr != NULL); + + fillBuffer_fromHandle(&buff, f); + CONTROL(buff.size == buff.capacity); + + fclose(f); /* do nothing specific if fclose() fails */ + return buff; + } +} + + +/* @return : kBuffNull if any error */ +static buffer_t +createDictionaryBuffer(const char* dictionaryName, + const void* srcBuffer, + const size_t* srcBlockSizes, size_t nbBlocks, + size_t requestedDictSize) +{ + if (dictionaryName) { + DISPLAYLEVEL(3, "loading dictionary %s \n", dictionaryName); + return createBuffer_fromFile(dictionaryName); /* note : result might be kBuffNull */ + + } else { + + DISPLAYLEVEL(3, "creating dictionary, of target size %u bytes \n", + (unsigned)requestedDictSize); + void* const dictBuffer = malloc(requestedDictSize); + CONTROL(dictBuffer != NULL); + + assert(nbBlocks <= UINT_MAX); + size_t const dictSize = ZDICT_trainFromBuffer(dictBuffer, requestedDictSize, + srcBuffer, + srcBlockSizes, (unsigned)nbBlocks); + CONTROL(!ZSTD_isError(dictSize)); + + buffer_t result; + result.ptr = dictBuffer; + result.capacity = requestedDictSize; + result.size = dictSize; + return result; + } +} + +/*! BMK_loadFiles() : + * Loads `buffer`, with content from files listed within `fileNamesTable`. + * Fills `buffer` entirely. + * @return : 0 on success, !=0 on error */ +static int loadFiles(void* buffer, size_t bufferSize, + size_t* fileSizes, + const char* const * fileNamesTable, unsigned nbFiles) +{ + size_t pos = 0, totalSize = 0; + + for (unsigned n=0; n<nbFiles; n++) { + U64 fileSize = UTIL_getFileSize(fileNamesTable[n]); + if (UTIL_isDirectory(fileNamesTable[n])) { + fileSizes[n] = 0; + continue; + } + if (fileSize == UTIL_FILESIZE_UNKNOWN) { + fileSizes[n] = 0; + continue; + } + + FILE* const f = fopen(fileNamesTable[n], "rb"); + assert(f!=NULL); + + assert(pos <= bufferSize); + assert(fileSize <= bufferSize - pos); + + { size_t const readSize = fread(((char*)buffer)+pos, 1, (size_t)fileSize, f); + assert(readSize == fileSize); + pos += readSize; + } + fileSizes[n] = (size_t)fileSize; + totalSize += (size_t)fileSize; + fclose(f); + } + + assert(totalSize == bufferSize); + return 0; +} + + + +/*--- slice_collection_t ---*/ + +typedef struct { + void** slicePtrs; + size_t* capacities; + size_t nbSlices; +} slice_collection_t; + +static const slice_collection_t kNullCollection = { NULL, NULL, 0 }; + +static void freeSliceCollection(slice_collection_t collection) +{ + free(collection.slicePtrs); + free(collection.capacities); +} + +/* shrinkSizes() : + * downsizes sizes of slices within collection, according to `newSizes`. + * every `newSizes` entry must be <= than its corresponding collection size */ +void shrinkSizes(slice_collection_t collection, + const size_t* newSizes) /* presumed same size as collection */ +{ + size_t const nbSlices = collection.nbSlices; + for (size_t blockNb = 0; blockNb < nbSlices; blockNb++) { + assert(newSizes[blockNb] <= collection.capacities[blockNb]); + collection.capacities[blockNb] = newSizes[blockNb]; + } +} + + +/* splitSlices() : + * nbSlices : if == 0, nbSlices is automatically determined from srcSlices and blockSize. + * otherwise, creates exactly nbSlices slices, + * by either truncating input (when smaller) + * or repeating input from beginning */ +static slice_collection_t +splitSlices(slice_collection_t srcSlices, size_t blockSize, size_t nbSlices) +{ + if (blockSize==0) blockSize = (size_t)(-1); /* means "do not cut" */ + size_t nbSrcBlocks = 0; + for (size_t ssnb=0; ssnb < srcSlices.nbSlices; ssnb++) { + size_t pos = 0; + while (pos <= srcSlices.capacities[ssnb]) { + nbSrcBlocks++; + pos += blockSize; + } + } + + if (nbSlices == 0) nbSlices = nbSrcBlocks; + + void** const sliceTable = (void**)malloc(nbSlices * sizeof(*sliceTable)); + size_t* const capacities = (size_t*)malloc(nbSlices * sizeof(*capacities)); + if (sliceTable == NULL || capacities == NULL) { + free(sliceTable); + free(capacities); + return kNullCollection; + } + + size_t ssnb = 0; + for (size_t sliceNb=0; sliceNb < nbSlices; ) { + ssnb = (ssnb + 1) % srcSlices.nbSlices; + size_t pos = 0; + char* const ptr = (char*)srcSlices.slicePtrs[ssnb]; + while (pos < srcSlices.capacities[ssnb] && sliceNb < nbSlices) { + size_t const size = MIN(blockSize, srcSlices.capacities[ssnb] - pos); + sliceTable[sliceNb] = ptr + pos; + capacities[sliceNb] = size; + sliceNb++; + pos += blockSize; + } + } + + slice_collection_t result; + result.nbSlices = nbSlices; + result.slicePtrs = sliceTable; + result.capacities = capacities; + return result; +} + + +static size_t sliceCollection_totalCapacity(slice_collection_t sc) +{ + size_t totalSize = 0; + for (size_t n=0; n<sc.nbSlices; n++) + totalSize += sc.capacities[n]; + return totalSize; +} + + +/* --- buffer collection --- */ + +typedef struct { + buffer_t buffer; + slice_collection_t slices; +} buffer_collection_t; + + +static void freeBufferCollection(buffer_collection_t bc) +{ + freeBuffer(bc.buffer); + freeSliceCollection(bc.slices); +} + + +static buffer_collection_t +createBufferCollection_fromSliceCollectionSizes(slice_collection_t sc) +{ + size_t const bufferSize = sliceCollection_totalCapacity(sc); + + buffer_t buffer = createBuffer(bufferSize); + CONTROL(buffer.ptr != NULL); + + size_t const nbSlices = sc.nbSlices; + void** const slices = (void**)malloc(nbSlices * sizeof(*slices)); + CONTROL(slices != NULL); + + size_t* const capacities = (size_t*)malloc(nbSlices * sizeof(*capacities)); + CONTROL(capacities != NULL); + + char* const ptr = (char*)buffer.ptr; + size_t pos = 0; + for (size_t n=0; n < nbSlices; n++) { + capacities[n] = sc.capacities[n]; + slices[n] = ptr + pos; + pos += capacities[n]; + } + + buffer_collection_t result; + result.buffer = buffer; + result.slices.nbSlices = nbSlices; + result.slices.capacities = capacities; + result.slices.slicePtrs = slices; + return result; +} + +static buffer_collection_t +createBufferCollection_fromSliceCollection(slice_collection_t sc) +{ + size_t const bufferSize = sliceCollection_totalCapacity(sc); + + buffer_t buffer = createBuffer(bufferSize); + CONTROL(buffer.ptr != NULL); + + size_t const nbSlices = sc.nbSlices; + void** const slices = (void**)malloc(nbSlices * sizeof(*slices)); + CONTROL(slices != NULL); + + size_t* const capacities = (size_t*)malloc(nbSlices * sizeof(*capacities)); + CONTROL(capacities != NULL); + + char* const ptr = (char*)buffer.ptr; + size_t pos = 0; + for (size_t n=0; n < nbSlices; n++) { + capacities[n] = sc.capacities[n]; + slices[n] = ptr + pos; + pos += capacities[n]; + } + + for (size_t i = 0; i < nbSlices; i++) { + memcpy(slices[i], sc.slicePtrs[i], sc.capacities[i]); + capacities[i] = sc.capacities[i]; + } + + buffer_collection_t result; + result.buffer = buffer; + result.slices.nbSlices = nbSlices; + result.slices.capacities = capacities; + result.slices.slicePtrs = slices; + + return result; +} + +/* @return : kBuffNull if any error */ +static buffer_collection_t +createBufferCollection_fromFiles(const char* const * fileNamesTable, unsigned nbFiles) +{ + U64 const totalSizeToLoad = UTIL_getTotalFileSize(fileNamesTable, nbFiles); + assert(totalSizeToLoad != UTIL_FILESIZE_UNKNOWN); + assert(totalSizeToLoad <= BENCH_SIZE_MAX); + size_t const loadedSize = (size_t)totalSizeToLoad; + assert(loadedSize > 0); + void* const srcBuffer = malloc(loadedSize); + assert(srcBuffer != NULL); + + assert(nbFiles > 0); + size_t* const fileSizes = (size_t*)calloc(nbFiles, sizeof(*fileSizes)); + assert(fileSizes != NULL); + + /* Load input buffer */ + int const errorCode = loadFiles(srcBuffer, loadedSize, + fileSizes, + fileNamesTable, nbFiles); + assert(errorCode == 0); + + void** sliceTable = (void**)malloc(nbFiles * sizeof(*sliceTable)); + assert(sliceTable != NULL); + + char* const ptr = (char*)srcBuffer; + size_t pos = 0; + unsigned fileNb = 0; + for ( ; (pos < loadedSize) && (fileNb < nbFiles); fileNb++) { + sliceTable[fileNb] = ptr + pos; + pos += fileSizes[fileNb]; + } + assert(pos == loadedSize); + assert(fileNb == nbFiles); + + + buffer_t buffer; + buffer.ptr = srcBuffer; + buffer.capacity = loadedSize; + buffer.size = loadedSize; + + slice_collection_t slices; + slices.slicePtrs = sliceTable; + slices.capacities = fileSizes; + slices.nbSlices = nbFiles; + + buffer_collection_t bc; + bc.buffer = buffer; + bc.slices = slices; + return bc; +} + + + + +/*--- ddict_collection_t ---*/ + +typedef struct { + ZSTD_DDict** ddicts; + size_t nbDDict; +} ddict_collection_t; + +typedef struct { + ZSTD_CDict** cdicts; + size_t nbCDict; +} cdict_collection_t; + +static const cdict_collection_t kNullCDictCollection = { NULL, 0 }; + +static void freeCDictCollection(cdict_collection_t cdictc) +{ + for (size_t dictNb=0; dictNb < cdictc.nbCDict; dictNb++) { + ZSTD_freeCDict(cdictc.cdicts[dictNb]); + } + free(cdictc.cdicts); +} + +/* returns .buffers=NULL if operation fails */ +static cdict_collection_t createCDictCollection(const void* dictBuffer, size_t dictSize, size_t nbCDict, ZSTD_dictContentType_e dictContentType, ZSTD_CCtx_params* cctxParams) +{ + ZSTD_CDict** const cdicts = malloc(nbCDict * sizeof(ZSTD_CDict*)); + if (cdicts==NULL) return kNullCDictCollection; + for (size_t dictNb=0; dictNb < nbCDict; dictNb++) { + cdicts[dictNb] = ZSTD_createCDict_advanced2(dictBuffer, dictSize, DICT_LOAD_METHOD, dictContentType, cctxParams, ZSTD_defaultCMem); + CONTROL(cdicts[dictNb] != NULL); + } + cdict_collection_t cdictc; + cdictc.cdicts = cdicts; + cdictc.nbCDict = nbCDict; + return cdictc; +} + +static const ddict_collection_t kNullDDictCollection = { NULL, 0 }; + +static void freeDDictCollection(ddict_collection_t ddictc) +{ + for (size_t dictNb=0; dictNb < ddictc.nbDDict; dictNb++) { + ZSTD_freeDDict(ddictc.ddicts[dictNb]); + } + free(ddictc.ddicts); +} + +/* returns .buffers=NULL if operation fails */ +static ddict_collection_t createDDictCollection(const void* dictBuffer, size_t dictSize, size_t nbDDict) +{ + ZSTD_DDict** const ddicts = malloc(nbDDict * sizeof(ZSTD_DDict*)); + assert(ddicts != NULL); + if (ddicts==NULL) return kNullDDictCollection; + for (size_t dictNb=0; dictNb < nbDDict; dictNb++) { + ddicts[dictNb] = ZSTD_createDDict(dictBuffer, dictSize); + assert(ddicts[dictNb] != NULL); + } + ddict_collection_t ddictc; + ddictc.ddicts = ddicts; + ddictc.nbDDict = nbDDict; + return ddictc; +} + + +/* mess with addresses, so that linear scanning dictionaries != linear address scanning */ +void shuffleCDictionaries(cdict_collection_t dicts) +{ + size_t const nbDicts = dicts.nbCDict; + for (size_t r=0; r<nbDicts; r++) { + size_t const d = (size_t)rand() % nbDicts; + ZSTD_CDict* tmpd = dicts.cdicts[d]; + dicts.cdicts[d] = dicts.cdicts[r]; + dicts.cdicts[r] = tmpd; + } + for (size_t r=0; r<nbDicts; r++) { + size_t const d1 = (size_t)rand() % nbDicts; + size_t const d2 = (size_t)rand() % nbDicts; + ZSTD_CDict* tmpd = dicts.cdicts[d1]; + dicts.cdicts[d1] = dicts.cdicts[d2]; + dicts.cdicts[d2] = tmpd; + } +} + +/* mess with addresses, so that linear scanning dictionaries != linear address scanning */ +void shuffleDDictionaries(ddict_collection_t dicts) +{ + size_t const nbDicts = dicts.nbDDict; + for (size_t r=0; r<nbDicts; r++) { + size_t const d = (size_t)rand() % nbDicts; + ZSTD_DDict* tmpd = dicts.ddicts[d]; + dicts.ddicts[d] = dicts.ddicts[r]; + dicts.ddicts[r] = tmpd; + } + for (size_t r=0; r<nbDicts; r++) { + size_t const d1 = (size_t)rand() % nbDicts; + size_t const d2 = (size_t)rand() % nbDicts; + ZSTD_DDict* tmpd = dicts.ddicts[d1]; + dicts.ddicts[d1] = dicts.ddicts[d2]; + dicts.ddicts[d2] = tmpd; + } +} + + +/* --- Compression --- */ + +/* compressBlocks() : + * @return : total compressed size of all blocks, + * or 0 if error. + */ +static size_t compressBlocks(size_t* cSizes, /* optional (can be NULL). If present, must contain at least nbBlocks fields */ + slice_collection_t dstBlockBuffers, + slice_collection_t srcBlockBuffers, + ZSTD_CDict* cdict, int cLevel) +{ + size_t const nbBlocks = srcBlockBuffers.nbSlices; + assert(dstBlockBuffers.nbSlices == srcBlockBuffers.nbSlices); + + ZSTD_CCtx* const cctx = ZSTD_createCCtx(); + assert(cctx != NULL); + + size_t totalCSize = 0; + for (size_t blockNb=0; blockNb < nbBlocks; blockNb++) { + size_t cBlockSize; + if (cdict == NULL) { + cBlockSize = ZSTD_compressCCtx(cctx, + dstBlockBuffers.slicePtrs[blockNb], dstBlockBuffers.capacities[blockNb], + srcBlockBuffers.slicePtrs[blockNb], srcBlockBuffers.capacities[blockNb], + cLevel); + } else { + cBlockSize = ZSTD_compress_usingCDict(cctx, + dstBlockBuffers.slicePtrs[blockNb], dstBlockBuffers.capacities[blockNb], + srcBlockBuffers.slicePtrs[blockNb], srcBlockBuffers.capacities[blockNb], + cdict); + } + CONTROL(!ZSTD_isError(cBlockSize)); + if (cSizes) cSizes[blockNb] = cBlockSize; + totalCSize += cBlockSize; + } + return totalCSize; +} + + +/* --- Benchmark --- */ + +typedef struct { + ZSTD_CCtx* cctx; + size_t nbDicts; + size_t dictNb; + cdict_collection_t dictionaries; +} compressInstructions; + +compressInstructions createCompressInstructions(cdict_collection_t dictionaries, ZSTD_CCtx_params* cctxParams) +{ + compressInstructions ci; + ci.cctx = ZSTD_createCCtx(); + CONTROL(ci.cctx != NULL); + if (cctxParams) + ZSTD_CCtx_setParametersUsingCCtxParams(ci.cctx, cctxParams); + ci.nbDicts = dictionaries.nbCDict; + ci.dictNb = 0; + ci.dictionaries = dictionaries; + return ci; +} + +void freeCompressInstructions(compressInstructions ci) +{ + ZSTD_freeCCtx(ci.cctx); +} + +typedef struct { + ZSTD_DCtx* dctx; + size_t nbDicts; + size_t dictNb; + ddict_collection_t dictionaries; +} decompressInstructions; + +decompressInstructions createDecompressInstructions(ddict_collection_t dictionaries) +{ + decompressInstructions di; + di.dctx = ZSTD_createDCtx(); + assert(di.dctx != NULL); + di.nbDicts = dictionaries.nbDDict; + di.dictNb = 0; + di.dictionaries = dictionaries; + return di; +} + +void freeDecompressInstructions(decompressInstructions di) +{ + ZSTD_freeDCtx(di.dctx); +} + +/* benched function */ +size_t compress(const void* src, size_t srcSize, void* dst, size_t dstCapacity, void* payload) +{ + compressInstructions* const ci = (compressInstructions*) payload; + (void)dstCapacity; + + ZSTD_CCtx_refCDict(ci->cctx, ci->dictionaries.cdicts[ci->dictNb]); + ZSTD_compress2(ci->cctx, + dst, srcSize, + src, srcSize); + + ci->dictNb = ci->dictNb + 1; + if (ci->dictNb >= ci->nbDicts) ci->dictNb = 0; + + return srcSize; +} + +/* benched function */ +size_t decompress(const void* src, size_t srcSize, void* dst, size_t dstCapacity, void* payload) +{ + decompressInstructions* const di = (decompressInstructions*) payload; + + size_t const result = ZSTD_decompress_usingDDict(di->dctx, + dst, dstCapacity, + src, srcSize, + di->dictionaries.ddicts[di->dictNb]); + + di->dictNb = di->dictNb + 1; + if (di->dictNb >= di->nbDicts) di->dictNb = 0; + + return result; +} + +typedef enum { + fastest = 0, + median = 1, +} metricAggregatePref_e; + +/* compareFunction() : + * Sort input in decreasing order when used with qsort() */ +int compareFunction(const void *a, const void *b) +{ + double x = *(const double *)a; + double y = *(const double *)b; + if (x < y) + return 1; + else if (x > y) + return -1; + return 0; +} + +double aggregateData(double *data, size_t size, + metricAggregatePref_e metricAggregatePref) +{ + qsort(data, size, sizeof(*data), compareFunction); + if (metricAggregatePref == fastest) + return data[0]; + else /* median */ + return (data[(size - 1) / 2] + data[size / 2]) / 2; +} + +static int benchMem(slice_collection_t dstBlocks, slice_collection_t srcBlocks, + ddict_collection_t ddictionaries, + cdict_collection_t cdictionaries, unsigned nbRounds, + int benchCompression, const char *exeName, + ZSTD_CCtx_params *cctxParams, + metricAggregatePref_e metricAggregatePref) +{ + assert(dstBlocks.nbSlices == srcBlocks.nbSlices); + if (benchCompression) assert(cctxParams); + + unsigned const ms_per_round = RUN_TIME_DEFAULT_MS; + unsigned const total_time_ms = nbRounds * ms_per_round; + + double *const speedPerRound = (double *)malloc(nbRounds * sizeof(double)); + + BMK_timedFnState_t* const benchState = + BMK_createTimedFnState(total_time_ms, ms_per_round); + + decompressInstructions di = createDecompressInstructions(ddictionaries); + compressInstructions ci = + createCompressInstructions(cdictionaries, cctxParams); + void* payload = benchCompression ? (void*)&ci : (void*)&di; + BMK_benchParams_t const bp = { + .benchFn = benchCompression ? compress : decompress, + .benchPayload = payload, + .initFn = NULL, + .initPayload = NULL, + .errorFn = ZSTD_isError, + .blockCount = dstBlocks.nbSlices, + .srcBuffers = (const void* const*) srcBlocks.slicePtrs, + .srcSizes = srcBlocks.capacities, + .dstBuffers = dstBlocks.slicePtrs, + .dstCapacities = dstBlocks.capacities, + .blockResults = NULL + }; + + size_t roundNb = 0; + for (;;) { + BMK_runOutcome_t const outcome = BMK_benchTimedFn(benchState, bp); + CONTROL(BMK_isSuccessful_runOutcome(outcome)); + + BMK_runTime_t const result = BMK_extract_runTime(outcome); + double const dTime_ns = result.nanoSecPerRun; + double const dTime_sec = (double)dTime_ns / 1000000000; + size_t const srcSize = result.sumOfReturn; + double const speed_MBps = (double)srcSize / dTime_sec / (1 MB); + speedPerRound[roundNb] = speed_MBps; + if (benchCompression) + DISPLAY("Compression Speed : %.1f MB/s \r", speed_MBps); + else + DISPLAY("Decompression Speed : %.1f MB/s \r", speed_MBps); + + fflush(stdout); + if (BMK_isCompleted_TimedFn(benchState)) break; + roundNb++; + } + DISPLAY("\n"); + /* BMK_benchTimedFn may not run exactly nbRounds iterations */ + double speedAggregated = + aggregateData(speedPerRound, roundNb + 1, metricAggregatePref); + if (metricAggregatePref == fastest) + DISPLAY("Fastest Speed : %.1f MB/s \n", speedAggregated); + else + DISPLAY("Median Speed : %.1f MB/s \n", speedAggregated); + + char* csvFileName = malloc(strlen(exeName) + 5); + strcpy(csvFileName, exeName); + strcat(csvFileName, ".csv"); + FILE* csvFile = fopen(csvFileName, "r"); + if (!csvFile) { + csvFile = fopen(csvFileName, "wt"); + assert(csvFile); + fprintf(csvFile, "%s\n", exeName); + /* Print table headers */ + fprintf( + csvFile, + "Compression/Decompression,Level,nbDicts,dictAttachPref,metricAggregatePref,Speed\n"); + } else { + fclose(csvFile); + csvFile = fopen(csvFileName, "at"); + assert(csvFile); + } + + int cLevel = -1; + int dictAttachPref = -1; + if (benchCompression) { + ZSTD_CCtxParams_getParameter(cctxParams, ZSTD_c_compressionLevel, + &cLevel); + ZSTD_CCtxParams_getParameter(cctxParams, ZSTD_c_forceAttachDict, + &dictAttachPref); + } + fprintf(csvFile, "%s,%d,%ld,%d,%d,%.1f\n", + benchCompression ? "Compression" : "Decompression", cLevel, + benchCompression ? ci.nbDicts : di.nbDicts, dictAttachPref, + metricAggregatePref, speedAggregated); + fclose(csvFile); + free(csvFileName); + + freeDecompressInstructions(di); + freeCompressInstructions(ci); + BMK_freeTimedFnState(benchState); + + return 0; /* success */ +} + + +/*! bench() : + * fileName : file to load for benchmarking purpose + * dictionary : optional (can be NULL), file to load as dictionary, + * if none provided : will be calculated on the fly by the program. + * @return : 0 is success, 1+ otherwise */ +int bench(const char **fileNameTable, unsigned nbFiles, const char *dictionary, + size_t blockSize, int clevel, unsigned nbDictMax, unsigned nbBlocks, + unsigned nbRounds, int benchCompression, + ZSTD_dictContentType_e dictContentType, ZSTD_CCtx_params *cctxParams, + const char *exeName, metricAggregatePref_e metricAggregatePref) +{ + int result = 0; + + DISPLAYLEVEL(3, "loading %u files... \n", nbFiles); + buffer_collection_t const srcs = createBufferCollection_fromFiles(fileNameTable, nbFiles); + CONTROL(srcs.buffer.ptr != NULL); + buffer_t srcBuffer = srcs.buffer; + size_t const srcSize = srcBuffer.size; + DISPLAYLEVEL(3, "created src buffer of size %.1f MB \n", + (double)srcSize / (1 MB)); + + slice_collection_t const srcSlices = splitSlices(srcs.slices, blockSize, nbBlocks); + nbBlocks = (unsigned)(srcSlices.nbSlices); + DISPLAYLEVEL(3, "split input into %u blocks ", nbBlocks); + if (blockSize) + DISPLAYLEVEL(3, "of max size %u bytes ", (unsigned)blockSize); + DISPLAYLEVEL(3, "\n"); + size_t const totalSrcSlicesSize = sliceCollection_totalCapacity(srcSlices); + + + size_t* const dstCapacities = malloc(nbBlocks * sizeof(*dstCapacities)); + CONTROL(dstCapacities != NULL); + size_t dstBufferCapacity = 0; + for (size_t bnb=0; bnb<nbBlocks; bnb++) { + dstCapacities[bnb] = ZSTD_compressBound(srcSlices.capacities[bnb]); + dstBufferCapacity += dstCapacities[bnb]; + } + + buffer_t dstBuffer = createBuffer(dstBufferCapacity); + CONTROL(dstBuffer.ptr != NULL); + + void** const sliceTable = malloc(nbBlocks * sizeof(*sliceTable)); + CONTROL(sliceTable != NULL); + + { char* const ptr = dstBuffer.ptr; + size_t pos = 0; + for (size_t snb=0; snb < nbBlocks; snb++) { + sliceTable[snb] = ptr + pos; + pos += dstCapacities[snb]; + } } + + slice_collection_t dstSlices; + dstSlices.capacities = dstCapacities; + dstSlices.slicePtrs = sliceTable; + dstSlices.nbSlices = nbBlocks; + + + /* dictionary determination */ + buffer_t const dictBuffer = createDictionaryBuffer(dictionary, + srcs.buffer.ptr, + srcSlices.capacities, srcSlices.nbSlices, + DICTSIZE); + CONTROL(dictBuffer.ptr != NULL); + + ZSTD_CDict* const cdict = ZSTD_createCDict_advanced2(dictBuffer.ptr, dictBuffer.size, DICT_LOAD_METHOD, dictContentType, cctxParams, ZSTD_defaultCMem); + CONTROL(cdict != NULL); + + size_t const cTotalSizeNoDict = compressBlocks(NULL, dstSlices, srcSlices, NULL, clevel); + CONTROL(cTotalSizeNoDict != 0); + DISPLAYLEVEL(3, "compressing at level %u without dictionary : Ratio=%.2f (%u bytes) \n", + clevel, + (double)totalSrcSlicesSize / (double)cTotalSizeNoDict, (unsigned)cTotalSizeNoDict); + + size_t* const cSizes = malloc(nbBlocks * sizeof(size_t)); + CONTROL(cSizes != NULL); + + size_t const cTotalSize = compressBlocks(cSizes, dstSlices, srcSlices, cdict, clevel); + CONTROL(cTotalSize != 0); + DISPLAYLEVEL(3, "compressed using a %u bytes dictionary : Ratio=%.2f (%u bytes) \n", + (unsigned)dictBuffer.size, + (double)totalSrcSlicesSize / (double)cTotalSize, (unsigned)cTotalSize); + + /* now dstSlices contain the real compressed size of each block, instead of the maximum capacity */ + shrinkSizes(dstSlices, cSizes); + + unsigned const nbDicts = nbDictMax ? nbDictMax : nbBlocks; + + cdict_collection_t const cdictionaries = createCDictCollection(dictBuffer.ptr, dictBuffer.size, nbDicts, dictContentType, cctxParams); + CONTROL(cdictionaries.cdicts != NULL); + + ddict_collection_t const ddictionaries = createDDictCollection(dictBuffer.ptr, dictBuffer.size, nbDicts); + CONTROL(ddictionaries.ddicts != NULL); + + if (benchCompression) { + size_t const dictMem = ZSTD_sizeof_CDict(cdictionaries.cdicts[0]); + size_t const allDictMem = dictMem * nbDicts; + DISPLAYLEVEL(3, "generating %u dictionaries, using %.1f MB of memory \n", + nbDicts, (double)allDictMem / (1 MB)); + + shuffleCDictionaries(cdictionaries); + + buffer_collection_t resultCollection = createBufferCollection_fromSliceCollection(srcSlices); + CONTROL(resultCollection.buffer.ptr != NULL); + + result = benchMem(dstSlices, resultCollection.slices, ddictionaries, + cdictionaries, nbRounds, benchCompression, exeName, + cctxParams, metricAggregatePref); + + freeBufferCollection(resultCollection); + } else { + size_t const dictMem = ZSTD_estimateDDictSize(dictBuffer.size, DICT_LOAD_METHOD); + size_t const allDictMem = dictMem * nbDicts; + DISPLAYLEVEL(3, "generating %u dictionaries, using %.1f MB of memory \n", + nbDicts, (double)allDictMem / (1 MB)); + + shuffleDDictionaries(ddictionaries); + + buffer_collection_t resultCollection = createBufferCollection_fromSliceCollectionSizes(srcSlices); + CONTROL(resultCollection.buffer.ptr != NULL); + + result = benchMem(resultCollection.slices, dstSlices, ddictionaries, + cdictionaries, nbRounds, benchCompression, exeName, + NULL, metricAggregatePref); + + freeBufferCollection(resultCollection); + } + + /* free all heap objects in reverse order */ + freeCDictCollection(cdictionaries); + freeDDictCollection(ddictionaries); + free(cSizes); + ZSTD_freeCDict(cdict); + freeBuffer(dictBuffer); + freeSliceCollection(dstSlices); + freeBuffer(dstBuffer); + freeSliceCollection(srcSlices); + freeBufferCollection(srcs); + + return result; +} + + + +/* --- Command Line --- */ + +/*! readU32FromChar() : + * @return : unsigned integer value read from input in `char` format. + * allows and interprets K, KB, KiB, M, MB and MiB suffix. + * Will also modify `*stringPtr`, advancing it to position where it stopped reading. + * Note : function will exit() program if digit sequence overflows */ +static unsigned readU32FromChar(const char** stringPtr) +{ + unsigned result = 0; + while ((**stringPtr >='0') && (**stringPtr <='9')) { + unsigned const max = (((unsigned)(-1)) / 10) - 1; + assert(result <= max); /* check overflow */ + result *= 10, result += (unsigned)**stringPtr - '0', (*stringPtr)++ ; + } + if ((**stringPtr=='K') || (**stringPtr=='M')) { + unsigned const maxK = ((unsigned)(-1)) >> 10; + assert(result <= maxK); /* check overflow */ + result <<= 10; + if (**stringPtr=='M') { + assert(result <= maxK); /* check overflow */ + result <<= 10; + } + (*stringPtr)++; /* skip `K` or `M` */ + if (**stringPtr=='i') (*stringPtr)++; + if (**stringPtr=='B') (*stringPtr)++; + } + return result; +} + +/** longCommandWArg() : + * check if *stringPtr is the same as longCommand. + * If yes, @return 1 and advances *stringPtr to the position which immediately follows longCommand. + * @return 0 and doesn't modify *stringPtr otherwise. + */ +static int longCommandWArg(const char** stringPtr, const char* longCommand) +{ + size_t const comSize = strlen(longCommand); + int const result = !strncmp(*stringPtr, longCommand, comSize); + if (result) *stringPtr += comSize; + return result; +} + + +int usage(const char* exeName) +{ + DISPLAY (" \n"); + DISPLAY (" %s [Options] filename(s) \n", exeName); + DISPLAY (" \n"); + DISPLAY ("Options : \n"); + DISPLAY ("-z : benchmark compression (default) \n"); + DISPLAY ("-d : benchmark decompression \n"); + DISPLAY ("-r : recursively load all files in subdirectories (default: off) \n"); + DISPLAY ("-B# : split input into blocks of size # (default: no split) \n"); + DISPLAY ("-# : use compression level # (default: %u) \n", CLEVEL_DEFAULT); + DISPLAY ("-D # : use # as a dictionary (default: create one) \n"); + DISPLAY ("-i# : nb benchmark rounds (default: %u) \n", BENCH_TIME_DEFAULT_S); + DISPLAY ("-p# : print speed for all rounds 0=fastest 1=median (default: 0) \n"); + DISPLAY ("--nbBlocks=#: use # blocks for bench (default: one per file) \n"); + DISPLAY ("--nbDicts=# : create # dictionaries for bench (default: one per block) \n"); + DISPLAY ("-h : help (this text) \n"); + DISPLAY (" \n"); + DISPLAY ("Advanced Options (see zstd.h for documentation) : \n"); + DISPLAY ("--dedicated-dict-search\n"); + DISPLAY ("--dict-content-type=#\n"); + DISPLAY ("--dict-attach-pref=#\n"); + return 0; +} + +int bad_usage(const char* exeName) +{ + DISPLAY (" bad usage : \n"); + usage(exeName); + return 1; +} + +int main (int argc, const char** argv) +{ + int recursiveMode = 0; + int benchCompression = 1; + int dedicatedDictSearch = 0; + unsigned nbRounds = BENCH_TIME_DEFAULT_S; + const char* const exeName = argv[0]; + + if (argc < 2) return bad_usage(exeName); + + const char** nameTable = (const char**)malloc((size_t)argc * sizeof(const char*)); + assert(nameTable != NULL); + unsigned nameIdx = 0; + + const char* dictionary = NULL; + int cLevel = CLEVEL_DEFAULT; + size_t blockSize = BLOCKSIZE_DEFAULT; + unsigned nbDicts = 0; /* determine nbDicts automatically: 1 dictionary per block */ + unsigned nbBlocks = 0; /* determine nbBlocks automatically, from source and blockSize */ + ZSTD_dictContentType_e dictContentType = ZSTD_dct_auto; + ZSTD_dictAttachPref_e dictAttachPref = ZSTD_dictDefaultAttach; + ZSTD_paramSwitch_e prefetchCDictTables = ZSTD_ps_auto; + metricAggregatePref_e metricAggregatePref = fastest; + + for (int argNb = 1; argNb < argc ; argNb++) { + const char* argument = argv[argNb]; + if (!strcmp(argument, "-h")) { free(nameTable); return usage(exeName); } + if (!strcmp(argument, "-d")) { benchCompression = 0; continue; } + if (!strcmp(argument, "-z")) { benchCompression = 1; continue; } + if (!strcmp(argument, "-r")) { recursiveMode = 1; continue; } + if (!strcmp(argument, "-D")) { argNb++; assert(argNb < argc); dictionary = argv[argNb]; continue; } + if (longCommandWArg(&argument, "-i")) { nbRounds = readU32FromChar(&argument); continue; } + if (longCommandWArg(&argument, "-p")) { metricAggregatePref = (int)readU32FromChar(&argument); continue;} + if (longCommandWArg(&argument, "--dictionary=")) { dictionary = argument; continue; } + if (longCommandWArg(&argument, "-B")) { blockSize = readU32FromChar(&argument); continue; } + if (longCommandWArg(&argument, "--blockSize=")) { blockSize = readU32FromChar(&argument); continue; } + if (longCommandWArg(&argument, "--nbDicts=")) { nbDicts = readU32FromChar(&argument); continue; } + if (longCommandWArg(&argument, "--nbBlocks=")) { nbBlocks = readU32FromChar(&argument); continue; } + if (longCommandWArg(&argument, "--clevel=")) { cLevel = (int)readU32FromChar(&argument); continue; } + if (longCommandWArg(&argument, "--dedicated-dict-search")) { dedicatedDictSearch = 1; continue; } + if (longCommandWArg(&argument, "--dict-content-type=")) { dictContentType = (int)readU32FromChar(&argument); continue; } + if (longCommandWArg(&argument, "--dict-attach-pref=")) { dictAttachPref = (int)readU32FromChar(&argument); continue; } + if (longCommandWArg(&argument, "--prefetch-cdict-tables=")) { prefetchCDictTables = (int)readU32FromChar(&argument); continue; } + if (longCommandWArg(&argument, "-")) { cLevel = (int)readU32FromChar(&argument); continue; } + /* anything that's not a command is a filename */ + nameTable[nameIdx++] = argument; + } + + FileNamesTable* filenameTable; + + if (recursiveMode) { +#ifndef UTIL_HAS_CREATEFILELIST + assert(0); /* missing capability, do not run */ +#endif + filenameTable = UTIL_createExpandedFNT(nameTable, nameIdx, 1 /* follow_links */); + } else { + filenameTable = UTIL_assembleFileNamesTable(nameTable, nameIdx, NULL); + nameTable = NULL; /* UTIL_createFileNamesTable() takes ownership of nameTable */ + } + + ZSTD_CCtx_params* cctxParams = ZSTD_createCCtxParams(); + ZSTD_CCtxParams_init(cctxParams, cLevel); + ZSTD_CCtxParams_setParameter(cctxParams, ZSTD_c_enableDedicatedDictSearch, dedicatedDictSearch); + ZSTD_CCtxParams_setParameter(cctxParams, ZSTD_c_nbWorkers, 0); + ZSTD_CCtxParams_setParameter(cctxParams, ZSTD_c_forceAttachDict, dictAttachPref); + ZSTD_CCtxParams_setParameter(cctxParams, ZSTD_c_prefetchCDictTables, prefetchCDictTables); + + int result = + bench(filenameTable->fileNames, (unsigned)filenameTable->tableSize, + dictionary, blockSize, cLevel, nbDicts, nbBlocks, nbRounds, + benchCompression, dictContentType, cctxParams, exeName, + metricAggregatePref); + + UTIL_freeFileNamesTable(filenameTable); + free(nameTable); + ZSTD_freeCCtxParams(cctxParams); + + return result; +} diff --git a/3rdparty/zstd/contrib/linux-kernel/.gitignore b/3rdparty/zstd/contrib/linux-kernel/.gitignore new file mode 100644 index 00000000000..d8dfeef21e5 --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/.gitignore @@ -0,0 +1,4 @@ +!lib/zstd +!lib/zstd/* +*.o +*.a diff --git a/3rdparty/zstd/contrib/linux-kernel/Makefile b/3rdparty/zstd/contrib/linux-kernel/Makefile new file mode 100644 index 00000000000..63dd15d958f --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/Makefile @@ -0,0 +1,108 @@ +# ################################################################ +# Copyright (c) Meta Platforms, Inc. and affiliates. +# All rights reserved. +# +# This source code is licensed under both the BSD-style license (found in the +# LICENSE file in the root directory of this source tree) and the GPLv2 (found +# in the COPYING file in the root directory of this source tree). +# You may select, at your option, one of the above-listed licenses. +# ################################################################ + +.PHONY: libzstd +libzstd: + rm -rf linux + mkdir -p linux + mkdir -p linux/include/linux + mkdir -p linux/lib/zstd + ../freestanding_lib/freestanding.py \ + --source-lib ../../lib \ + --output-lib linux/lib/zstd \ + --xxhash '<linux/xxhash.h>' \ + --xxh64-state 'struct xxh64_state' \ + --xxh64-prefix 'xxh64' \ + --rewrite-include '<limits\.h>=<linux/limits.h>' \ + --rewrite-include '<stddef\.h>=<linux/types.h>' \ + --rewrite-include '"\.\./zstd.h"=<linux/zstd.h>' \ + --rewrite-include '"(\.\./)?zstd_errors.h"=<linux/zstd_errors.h>' \ + --sed 's,/\*\*\*,/* *,g' \ + --sed 's,/\*\*,/*,g' \ + --spdx \ + -DZSTD_NO_INTRINSICS \ + -DZSTD_NO_UNUSED_FUNCTIONS \ + -DZSTD_LEGACY_SUPPORT=0 \ + -DZSTD_STATIC_LINKING_ONLY \ + -DFSE_STATIC_LINKING_ONLY \ + -DXXH_STATIC_LINKING_ONLY \ + -D__GNUC__ \ + -D__linux__=1 \ + -DSTATIC_BMI2=0 \ + -DZSTD_ADDRESS_SANITIZER=0 \ + -DZSTD_MEMORY_SANITIZER=0 \ + -DZSTD_DATAFLOW_SANITIZER=0 \ + -DZSTD_COMPRESS_HEAPMODE=1 \ + -UNO_PREFETCH \ + -U__cplusplus \ + -UZSTD_DLL_EXPORT \ + -UZSTD_DLL_IMPORT \ + -U__ICCARM__ \ + -UZSTD_MULTITHREAD \ + -U_MSC_VER \ + -U_WIN32 \ + -RZSTDLIB_VISIBLE= \ + -RZSTDERRORLIB_VISIBLE= \ + -RZSTD_FALLTHROUGH=fallthrough \ + -DZSTD_HAVE_WEAK_SYMBOLS=0 \ + -DZSTD_TRACE=0 \ + -DZSTD_NO_TRACE \ + -DZSTD_DISABLE_ASM \ + -DZSTD_LINUX_KERNEL + rm linux/lib/zstd/decompress/huf_decompress_amd64.S + mv linux/lib/zstd/zstd.h linux/include/linux/zstd_lib.h + mv linux/lib/zstd/zstd_errors.h linux/include/linux/ + cp linux_zstd.h linux/include/linux/zstd.h + cp zstd_common_module.c linux/lib/zstd + cp zstd_compress_module.c linux/lib/zstd + cp zstd_decompress_module.c linux/lib/zstd + cp decompress_sources.h linux/lib/zstd + cp linux.mk linux/lib/zstd/Makefile + +LINUX ?= $(HOME)/repos/linux + +.PHONY: import +import: libzstd + rm -f $(LINUX)/include/linux/zstd.h + rm -f $(LINUX)/include/linux/zstd_errors.h + rm -rf $(LINUX)/lib/zstd + cp linux/include/linux/zstd.h $(LINUX)/include/linux + cp linux/include/linux/zstd_lib.h $(LINUX)/include/linux + cp linux/include/linux/zstd_errors.h $(LINUX)/include/linux + cp -r linux/lib/zstd $(LINUX)/lib + +import-upstream: + rm -rf $(LINUX)/lib/zstd + mkdir $(LINUX)/lib/zstd + cp ../../lib/zstd.h $(LINUX)/include/linux/zstd_lib.h + cp -r ../../lib/common $(LINUX)/lib/zstd + cp -r ../../lib/compress $(LINUX)/lib/zstd + cp -r ../../lib/decompress $(LINUX)/lib/zstd + mv $(LINUX)/lib/zstd/zstd_errors.h $(LINUX)/include/linux + rm $(LINUX)/lib/zstd/common/threading.* + rm $(LINUX)/lib/zstd/common/pool.* + rm $(LINUX)/lib/zstd/common/xxhash.* + rm $(LINUX)/lib/zstd/compress/zstdmt_* + +DEBUGFLAGS= -Wall -Wextra -Wcast-qual -Wcast-align -Wshadow \ + -Wstrict-aliasing=1 -Wswitch-enum -Wdeclaration-after-statement \ + -Wstrict-prototypes -Wundef -Wpointer-arith \ + -Wvla -Wformat=2 -Winit-self -Wfloat-equal -Wwrite-strings \ + -Wredundant-decls -Wmissing-prototypes -Wc++-compat \ + -Wimplicit-fallthrough + +.PHONY: test +test: libzstd + $(MAKE) -C test run-test CFLAGS="-O3 $(CFLAGS) $(DEBUGFLAGS) -Werror" -j + +.PHONY: clean +clean: + $(RM) -rf linux + $(MAKE) -C test clean diff --git a/3rdparty/zstd/contrib/linux-kernel/README.md b/3rdparty/zstd/contrib/linux-kernel/README.md new file mode 100644 index 00000000000..bfa070d1794 --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/README.md @@ -0,0 +1,14 @@ +# Zstd in the Linux Kernel + +This directory contains the scripts needed to transform upstream zstd into the version imported into the kernel. All the transforms are automated and tested by our continuous integration. + +## Upgrading Zstd in the Linux Kernel + +1. `cd` into this directory. +2. Run `make libzstd` and read the output. Make sure that all the diffs printed and changes made by the script are correct. +3. Run `make test` and ensure that it passes. +4. Import zstd into the Linux Kernel `make import LINUX=/path/to/linux/repo` +5. Inspect the diff for sanity. +6. Check the Linux Kernel history for zstd. If any patches were made to the kernel version of zstd, but not to upstream zstd, then port them upstream if necessary. +7. Test the diff. Benchmark if necessary. Make sure to test multiple architectures: At least x86, i386, and arm. +8. Submit the patch to the LKML. diff --git a/3rdparty/zstd/contrib/linux-kernel/btrfs-benchmark.sh b/3rdparty/zstd/contrib/linux-kernel/btrfs-benchmark.sh new file mode 100755 index 00000000000..5e28da9c6db --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/btrfs-benchmark.sh @@ -0,0 +1,104 @@ +# !/bin/sh +set -e + +# Benchmarks run on a Ubuntu 14.04 VM with 2 cores and 4 GiB of RAM. +# The VM is running on a Macbook Pro with a 3.1 GHz Intel Core i7 processor and +# 16 GB of RAM and an SSD. + +# silesia is a directory that can be downloaded from +# http://mattmahoney.net/dc/silesia.html +# ls -l silesia/ +# total 203M +# -rwxr-xr-x 1 terrelln 9.8M Apr 12 2002 dickens +# -rwxr-xr-x 1 terrelln 49M May 31 2002 mozilla +# -rwxr-xr-x 1 terrelln 9.6M Mar 20 2003 mr +# -rwxr-xr-x 1 terrelln 32M Apr 2 2002 nci +# -rwxr-xr-x 1 terrelln 5.9M Jul 4 2002 ooffice +# -rwxr-xr-x 1 terrelln 9.7M Apr 11 2002 osdb +# -rwxr-xr-x 1 terrelln 6.4M Apr 2 2002 reymont +# -rwxr-xr-x 1 terrelln 21M Mar 25 2002 samba +# -rwxr-xr-x 1 terrelln 7.0M Mar 24 2002 sao +# -rwxr-xr-x 1 terrelln 40M Mar 25 2002 webster +# -rwxr-xr-x 1 terrelln 8.1M Apr 4 2002 x-ray +# -rwxr-xr-x 1 terrelln 5.1M Nov 30 2000 xml + +# $HOME is on a ext4 filesystem +BENCHMARK_DIR="$HOME/silesia/" +N=10 + +# Normalize the environment +sudo umount /mnt/btrfs 2> /dev/null > /dev/null || true +sudo mount -t btrfs $@ /dev/sda3 /mnt/btrfs +sudo rm -rf /mnt/btrfs/* +sync +sudo umount /mnt/btrfs +sudo mount -t btrfs $@ /dev/sda3 /mnt/btrfs + +# Run the benchmark +echo "Compression" +time sh -c "for i in \$(seq $N); do sudo cp -r $BENCHMARK_DIR /mnt/btrfs/\$i; done; sync" + +echo "Approximate compression ratio" +printf "%d / %d\n" \ + $(df /mnt/btrfs --output=used -B 1 | tail -n 1) \ + $(sudo du /mnt/btrfs -b -d 0 | tr '\t' '\n' | head -n 1); + +# Unmount and remount to avoid any caching +sudo umount /mnt/btrfs +sudo mount -t btrfs $@ /dev/sda3 /mnt/btrfs + +echo "Decompression" +time sudo tar -c /mnt/btrfs 2> /dev/null | wc -c > /dev/null + +sudo rm -rf /mnt/btrfs/* +sudo umount /mnt/btrfs + +# Run for each of -o compress-force={none, lzo, zlib, zstd} 5 times and take the +# min time and ratio. +# Ran zstd with compression levels {1, 3, 6, 9, 12, 15}. +# Original size: 2119415342 B (using du /mnt/btrfs) + +# none +# compress: 4.205 s +# decompress: 3.090 s +# ratio: 0.99 + +# lzo +# compress: 5.328 s +# decompress: 4.793 s +# ratio: 1.66 + +# zlib +# compress: 32.588 s +# decompress: 8.791 s +# ratio : 2.58 + +# zstd 1 +# compress: 8.147 s +# decompress: 5.527 s +# ratio : 2.57 + +# zstd 3 +# compress: 12.207 s +# decompress: 5.195 s +# ratio : 2.71 + +# zstd 6 +# compress: 30.253 s +# decompress: 5.324 s +# ratio : 2.87 + +# zstd 9 +# compress: 49.659 s +# decompress: 5.220 s +# ratio : 2.92 + +# zstd 12 +# compress: 99.245 s +# decompress: 5.193 s +# ratio : 2.93 + +# zstd 15 +# compress: 196.997 s +# decompress: 5.992 s +# ratio : 3.01 diff --git a/3rdparty/zstd/contrib/linux-kernel/btrfs-extract-benchmark.sh b/3rdparty/zstd/contrib/linux-kernel/btrfs-extract-benchmark.sh new file mode 100755 index 00000000000..69721d0935d --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/btrfs-extract-benchmark.sh @@ -0,0 +1,99 @@ +# !/bin/sh +set -e + +# Benchmarks run on a Ubuntu 14.04 VM with 2 cores and 4 GiB of RAM. +# The VM is running on a Macbook Pro with a 3.1 GHz Intel Core i7 processor and +# 16 GB of RAM and an SSD. + +# silesia is a directory that can be downloaded from +# http://mattmahoney.net/dc/silesia.html +# ls -l silesia/ +# total 203M +# -rwxr-xr-x 1 terrelln 9.8M Apr 12 2002 dickens +# -rwxr-xr-x 1 terrelln 49M May 31 2002 mozilla +# -rwxr-xr-x 1 terrelln 9.6M Mar 20 2003 mr +# -rwxr-xr-x 1 terrelln 32M Apr 2 2002 nci +# -rwxr-xr-x 1 terrelln 5.9M Jul 4 2002 ooffice +# -rwxr-xr-x 1 terrelln 9.7M Apr 11 2002 osdb +# -rwxr-xr-x 1 terrelln 6.4M Apr 2 2002 reymont +# -rwxr-xr-x 1 terrelln 21M Mar 25 2002 samba +# -rwxr-xr-x 1 terrelln 7.0M Mar 24 2002 sao +# -rwxr-xr-x 1 terrelln 40M Mar 25 2002 webster +# -rwxr-xr-x 1 terrelln 8.1M Apr 4 2002 x-ray +# -rwxr-xr-x 1 terrelln 5.1M Nov 30 2000 xml + +# $HOME is on a ext4 filesystem +BENCHMARK_FILE="linux-4.11.6.tar" +BENCHMARK_DIR="$HOME/$BENCHMARK_FILE" + +# Normalize the environment +sudo umount /mnt/btrfs 2> /dev/null > /dev/null || true +sudo mount -t btrfs $@ /dev/sda3 /mnt/btrfs +sudo rm -rf /mnt/btrfs/* +sync +sudo umount /mnt/btrfs +sudo mount -t btrfs $@ /dev/sda3 /mnt/btrfs + +# Run the benchmark +echo "Copy" +time sh -c "sudo cp -r $BENCHMARK_DIR /mnt/btrfs/$BENCHMARK_FILE && sync" + +echo "Approximate tarred compression ratio" +printf "%d / %d\n" \ + $(df /mnt/btrfs --output=used -B 1 | tail -n 1) \ + $(sudo du /mnt/btrfs -b -d 0 | tr '\t' '\n' | head -n 1); + +# Unmount and remount to avoid any caching +sudo umount /mnt/btrfs +sudo mount -t btrfs $@ /dev/sda3 /mnt/btrfs + +echo "Extract" +time sh -c "sudo tar -C /mnt/btrfs -xf /mnt/btrfs/$BENCHMARK_FILE && sync" + +# Remove the tarball, leaving only the extracted data +sudo rm /mnt/btrfs/$BENCHMARK_FILE +# Unmount and remount to avoid any caching +sudo umount /mnt/btrfs +sudo mount -t btrfs $@ /dev/sda3 /mnt/btrfs + +echo "Approximate extracted compression ratio" +printf "%d / %d\n" \ + $(df /mnt/btrfs --output=used -B 1 | tail -n 1) \ + $(sudo du /mnt/btrfs -b -d 0 | tr '\t' '\n' | head -n 1); + +echo "Read" +time sudo tar -c /mnt/btrfs 2> /dev/null | wc -c > /dev/null + +sudo rm -rf /mnt/btrfs/* +sudo umount /mnt/btrfs + +# Run for each of -o compress-force={none, lzo, zlib, zstd} 5 times and take the +# min time and ratio. + +# none +# copy: 0.981 s +# extract: 5.501 s +# read: 8.807 s +# tarball ratio: 0.97 +# extracted ratio: 0.78 + +# lzo +# copy: 1.631 s +# extract: 8.458 s +# read: 8.585 s +# tarball ratio: 2.06 +# extracted ratio: 1.38 + +# zlib +# copy: 7.750 s +# extract: 21.544 s +# read: 11.744 s +# tarball ratio : 3.40 +# extracted ratio: 1.86 + +# zstd 1 +# copy: 2.579 s +# extract: 11.479 s +# read: 9.389 s +# tarball ratio : 3.57 +# extracted ratio: 1.85 diff --git a/3rdparty/zstd/contrib/linux-kernel/decompress_sources.h b/3rdparty/zstd/contrib/linux-kernel/decompress_sources.h new file mode 100644 index 00000000000..8a47eb2a451 --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/decompress_sources.h @@ -0,0 +1,34 @@ +/* SPDX-License-Identifier: GPL-2.0+ OR BSD-3-Clause */ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ + +/* + * This file includes every .c file needed for decompression. + * It is used by lib/decompress_unzstd.c to include the decompression + * source into the translation-unit, so it can be used for kernel + * decompression. + */ + +/* + * Disable the ASM Huffman implementation because we need to + * include all the sources. + */ +#define ZSTD_DISABLE_ASM 1 + +#include "common/debug.c" +#include "common/entropy_common.c" +#include "common/error_private.c" +#include "common/fse_decompress.c" +#include "common/zstd_common.c" +#include "decompress/huf_decompress.c" +#include "decompress/zstd_ddict.c" +#include "decompress/zstd_decompress.c" +#include "decompress/zstd_decompress_block.c" +#include "zstd_decompress_module.c" diff --git a/3rdparty/zstd/contrib/linux-kernel/linux.mk b/3rdparty/zstd/contrib/linux-kernel/linux.mk new file mode 100644 index 00000000000..464c410b276 --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/linux.mk @@ -0,0 +1,43 @@ +# SPDX-License-Identifier: GPL-2.0+ OR BSD-3-Clause +# ################################################################ +# Copyright (c) Meta Platforms, Inc. and affiliates. +# All rights reserved. +# +# This source code is licensed under both the BSD-style license (found in the +# LICENSE file in the root directory of this source tree) and the GPLv2 (found +# in the COPYING file in the root directory of this source tree). +# You may select, at your option, one of the above-listed licenses. +# ################################################################ +obj-$(CONFIG_ZSTD_COMPRESS) += zstd_compress.o +obj-$(CONFIG_ZSTD_DECOMPRESS) += zstd_decompress.o +obj-$(CONFIG_ZSTD_COMMON) += zstd_common.o + +zstd_compress-y := \ + zstd_compress_module.o \ + compress/fse_compress.o \ + compress/hist.o \ + compress/huf_compress.o \ + compress/zstd_compress.o \ + compress/zstd_compress_literals.o \ + compress/zstd_compress_sequences.o \ + compress/zstd_compress_superblock.o \ + compress/zstd_double_fast.o \ + compress/zstd_fast.o \ + compress/zstd_lazy.o \ + compress/zstd_ldm.o \ + compress/zstd_opt.o \ + +zstd_decompress-y := \ + zstd_decompress_module.o \ + decompress/huf_decompress.o \ + decompress/zstd_ddict.o \ + decompress/zstd_decompress.o \ + decompress/zstd_decompress_block.o \ + +zstd_common-y := \ + zstd_common_module.o \ + common/debug.o \ + common/entropy_common.o \ + common/error_private.o \ + common/fse_decompress.o \ + common/zstd_common.o \ diff --git a/3rdparty/zstd/contrib/linux-kernel/linux_zstd.h b/3rdparty/zstd/contrib/linux-kernel/linux_zstd.h new file mode 100644 index 00000000000..f109d49f43f --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/linux_zstd.h @@ -0,0 +1,447 @@ +/* SPDX-License-Identifier: GPL-2.0+ OR BSD-3-Clause */ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of https://github.com/facebook/zstd) and + * the GPLv2 (found in the COPYING file in the root directory of + * https://github.com/facebook/zstd). You may select, at your option, one of the + * above-listed licenses. + */ + +#ifndef LINUX_ZSTD_H +#define LINUX_ZSTD_H + +/** + * This is a kernel-style API that wraps the upstream zstd API, which cannot be + * used directly because the symbols aren't exported. It exposes the minimal + * functionality which is currently required by users of zstd in the kernel. + * Expose extra functions from lib/zstd/zstd.h as needed. + */ + +/* ====== Dependency ====== */ +#include <linux/types.h> +#include <linux/zstd_errors.h> +#include <linux/zstd_lib.h> + +/* ====== Helper Functions ====== */ +/** + * zstd_compress_bound() - maximum compressed size in worst case scenario + * @src_size: The size of the data to compress. + * + * Return: The maximum compressed size in the worst case scenario. + */ +size_t zstd_compress_bound(size_t src_size); + +/** + * zstd_is_error() - tells if a size_t function result is an error code + * @code: The function result to check for error. + * + * Return: Non-zero iff the code is an error. + */ +unsigned int zstd_is_error(size_t code); + +/** + * enum zstd_error_code - zstd error codes + */ +typedef ZSTD_ErrorCode zstd_error_code; + +/** + * zstd_get_error_code() - translates an error function result to an error code + * @code: The function result for which zstd_is_error(code) is true. + * + * Return: A unique error code for this error. + */ +zstd_error_code zstd_get_error_code(size_t code); + +/** + * zstd_get_error_name() - translates an error function result to a string + * @code: The function result for which zstd_is_error(code) is true. + * + * Return: An error string corresponding to the error code. + */ +const char *zstd_get_error_name(size_t code); + +/** + * zstd_min_clevel() - minimum allowed compression level + * + * Return: The minimum allowed compression level. + */ +int zstd_min_clevel(void); + +/** + * zstd_max_clevel() - maximum allowed compression level + * + * Return: The maximum allowed compression level. + */ +int zstd_max_clevel(void); + +/* ====== Parameter Selection ====== */ + +/** + * enum zstd_strategy - zstd compression search strategy + * + * From faster to stronger. See zstd_lib.h. + */ +typedef ZSTD_strategy zstd_strategy; + +/** + * struct zstd_compression_parameters - zstd compression parameters + * @windowLog: Log of the largest match distance. Larger means more + * compression, and more memory needed during decompression. + * @chainLog: Fully searched segment. Larger means more compression, + * slower, and more memory (useless for fast). + * @hashLog: Dispatch table. Larger means more compression, + * slower, and more memory. + * @searchLog: Number of searches. Larger means more compression and slower. + * @searchLength: Match length searched. Larger means faster decompression, + * sometimes less compression. + * @targetLength: Acceptable match size for optimal parser (only). Larger means + * more compression, and slower. + * @strategy: The zstd compression strategy. + * + * See zstd_lib.h. + */ +typedef ZSTD_compressionParameters zstd_compression_parameters; + +/** + * struct zstd_frame_parameters - zstd frame parameters + * @contentSizeFlag: Controls whether content size will be present in the + * frame header (when known). + * @checksumFlag: Controls whether a 32-bit checksum is generated at the + * end of the frame for error detection. + * @noDictIDFlag: Controls whether dictID will be saved into the frame + * header when using dictionary compression. + * + * The default value is all fields set to 0. See zstd_lib.h. + */ +typedef ZSTD_frameParameters zstd_frame_parameters; + +/** + * struct zstd_parameters - zstd parameters + * @cParams: The compression parameters. + * @fParams: The frame parameters. + */ +typedef ZSTD_parameters zstd_parameters; + +/** + * zstd_get_params() - returns zstd_parameters for selected level + * @level: The compression level + * @estimated_src_size: The estimated source size to compress or 0 + * if unknown. + * + * Return: The selected zstd_parameters. + */ +zstd_parameters zstd_get_params(int level, + unsigned long long estimated_src_size); + +/* ====== Single-pass Compression ====== */ + +typedef ZSTD_CCtx zstd_cctx; + +/** + * zstd_cctx_workspace_bound() - max memory needed to initialize a zstd_cctx + * @parameters: The compression parameters to be used. + * + * If multiple compression parameters might be used, the caller must call + * zstd_cctx_workspace_bound() for each set of parameters and use the maximum + * size. + * + * Return: A lower bound on the size of the workspace that is passed to + * zstd_init_cctx(). + */ +size_t zstd_cctx_workspace_bound(const zstd_compression_parameters *parameters); + +/** + * zstd_init_cctx() - initialize a zstd compression context + * @workspace: The workspace to emplace the context into. It must outlive + * the returned context. + * @workspace_size: The size of workspace. Use zstd_cctx_workspace_bound() to + * determine how large the workspace must be. + * + * Return: A zstd compression context or NULL on error. + */ +zstd_cctx *zstd_init_cctx(void *workspace, size_t workspace_size); + +/** + * zstd_compress_cctx() - compress src into dst with the initialized parameters + * @cctx: The context. Must have been initialized with zstd_init_cctx(). + * @dst: The buffer to compress src into. + * @dst_capacity: The size of the destination buffer. May be any size, but + * ZSTD_compressBound(srcSize) is guaranteed to be large enough. + * @src: The data to compress. + * @src_size: The size of the data to compress. + * @parameters: The compression parameters to be used. + * + * Return: The compressed size or an error, which can be checked using + * zstd_is_error(). + */ +size_t zstd_compress_cctx(zstd_cctx *cctx, void *dst, size_t dst_capacity, + const void *src, size_t src_size, const zstd_parameters *parameters); + +/* ====== Single-pass Decompression ====== */ + +typedef ZSTD_DCtx zstd_dctx; + +/** + * zstd_dctx_workspace_bound() - max memory needed to initialize a zstd_dctx + * + * Return: A lower bound on the size of the workspace that is passed to + * zstd_init_dctx(). + */ +size_t zstd_dctx_workspace_bound(void); + +/** + * zstd_init_dctx() - initialize a zstd decompression context + * @workspace: The workspace to emplace the context into. It must outlive + * the returned context. + * @workspace_size: The size of workspace. Use zstd_dctx_workspace_bound() to + * determine how large the workspace must be. + * + * Return: A zstd decompression context or NULL on error. + */ +zstd_dctx *zstd_init_dctx(void *workspace, size_t workspace_size); + +/** + * zstd_decompress_dctx() - decompress zstd compressed src into dst + * @dctx: The decompression context. + * @dst: The buffer to decompress src into. + * @dst_capacity: The size of the destination buffer. Must be at least as large + * as the decompressed size. If the caller cannot upper bound the + * decompressed size, then it's better to use the streaming API. + * @src: The zstd compressed data to decompress. Multiple concatenated + * frames and skippable frames are allowed. + * @src_size: The exact size of the data to decompress. + * + * Return: The decompressed size or an error, which can be checked using + * zstd_is_error(). + */ +size_t zstd_decompress_dctx(zstd_dctx *dctx, void *dst, size_t dst_capacity, + const void *src, size_t src_size); + +/* ====== Streaming Buffers ====== */ + +/** + * struct zstd_in_buffer - input buffer for streaming + * @src: Start of the input buffer. + * @size: Size of the input buffer. + * @pos: Position where reading stopped. Will be updated. + * Necessarily 0 <= pos <= size. + * + * See zstd_lib.h. + */ +typedef ZSTD_inBuffer zstd_in_buffer; + +/** + * struct zstd_out_buffer - output buffer for streaming + * @dst: Start of the output buffer. + * @size: Size of the output buffer. + * @pos: Position where writing stopped. Will be updated. + * Necessarily 0 <= pos <= size. + * + * See zstd_lib.h. + */ +typedef ZSTD_outBuffer zstd_out_buffer; + +/* ====== Streaming Compression ====== */ + +typedef ZSTD_CStream zstd_cstream; + +/** + * zstd_cstream_workspace_bound() - memory needed to initialize a zstd_cstream + * @cparams: The compression parameters to be used for compression. + * + * Return: A lower bound on the size of the workspace that is passed to + * zstd_init_cstream(). + */ +size_t zstd_cstream_workspace_bound(const zstd_compression_parameters *cparams); + +/** + * zstd_init_cstream() - initialize a zstd streaming compression context + * @parameters The zstd parameters to use for compression. + * @pledged_src_size: If params.fParams.contentSizeFlag == 1 then the caller + * must pass the source size (zero means empty source). + * Otherwise, the caller may optionally pass the source + * size, or zero if unknown. + * @workspace: The workspace to emplace the context into. It must outlive + * the returned context. + * @workspace_size: The size of workspace. + * Use zstd_cstream_workspace_bound(params->cparams) to + * determine how large the workspace must be. + * + * Return: The zstd streaming compression context or NULL on error. + */ +zstd_cstream *zstd_init_cstream(const zstd_parameters *parameters, + unsigned long long pledged_src_size, void *workspace, size_t workspace_size); + +/** + * zstd_reset_cstream() - reset the context using parameters from creation + * @cstream: The zstd streaming compression context to reset. + * @pledged_src_size: Optionally the source size, or zero if unknown. + * + * Resets the context using the parameters from creation. Skips dictionary + * loading, since it can be reused. If `pledged_src_size` is non-zero the frame + * content size is always written into the frame header. + * + * Return: Zero or an error, which can be checked using + * zstd_is_error(). + */ +size_t zstd_reset_cstream(zstd_cstream *cstream, + unsigned long long pledged_src_size); + +/** + * zstd_compress_stream() - streaming compress some of input into output + * @cstream: The zstd streaming compression context. + * @output: Destination buffer. `output->pos` is updated to indicate how much + * compressed data was written. + * @input: Source buffer. `input->pos` is updated to indicate how much data + * was read. Note that it may not consume the entire input, in which + * case `input->pos < input->size`, and it's up to the caller to + * present remaining data again. + * + * The `input` and `output` buffers may be any size. Guaranteed to make some + * forward progress if `input` and `output` are not empty. + * + * Return: A hint for the number of bytes to use as the input for the next + * function call or an error, which can be checked using + * zstd_is_error(). + */ +size_t zstd_compress_stream(zstd_cstream *cstream, zstd_out_buffer *output, + zstd_in_buffer *input); + +/** + * zstd_flush_stream() - flush internal buffers into output + * @cstream: The zstd streaming compression context. + * @output: Destination buffer. `output->pos` is updated to indicate how much + * compressed data was written. + * + * zstd_flush_stream() must be called until it returns 0, meaning all the data + * has been flushed. Since zstd_flush_stream() causes a block to be ended, + * calling it too often will degrade the compression ratio. + * + * Return: The number of bytes still present within internal buffers or an + * error, which can be checked using zstd_is_error(). + */ +size_t zstd_flush_stream(zstd_cstream *cstream, zstd_out_buffer *output); + +/** + * zstd_end_stream() - flush internal buffers into output and end the frame + * @cstream: The zstd streaming compression context. + * @output: Destination buffer. `output->pos` is updated to indicate how much + * compressed data was written. + * + * zstd_end_stream() must be called until it returns 0, meaning all the data has + * been flushed and the frame epilogue has been written. + * + * Return: The number of bytes still present within internal buffers or an + * error, which can be checked using zstd_is_error(). + */ +size_t zstd_end_stream(zstd_cstream *cstream, zstd_out_buffer *output); + +/* ====== Streaming Decompression ====== */ + +typedef ZSTD_DStream zstd_dstream; + +/** + * zstd_dstream_workspace_bound() - memory needed to initialize a zstd_dstream + * @max_window_size: The maximum window size allowed for compressed frames. + * + * Return: A lower bound on the size of the workspace that is passed + * to zstd_init_dstream(). + */ +size_t zstd_dstream_workspace_bound(size_t max_window_size); + +/** + * zstd_init_dstream() - initialize a zstd streaming decompression context + * @max_window_size: The maximum window size allowed for compressed frames. + * @workspace: The workspace to emplace the context into. It must outlive + * the returned context. + * @workspaceSize: The size of workspace. + * Use zstd_dstream_workspace_bound(max_window_size) to + * determine how large the workspace must be. + * + * Return: The zstd streaming decompression context. + */ +zstd_dstream *zstd_init_dstream(size_t max_window_size, void *workspace, + size_t workspace_size); + +/** + * zstd_reset_dstream() - reset the context using parameters from creation + * @dstream: The zstd streaming decompression context to reset. + * + * Resets the context using the parameters from creation. Skips dictionary + * loading, since it can be reused. + * + * Return: Zero or an error, which can be checked using zstd_is_error(). + */ +size_t zstd_reset_dstream(zstd_dstream *dstream); + +/** + * zstd_decompress_stream() - streaming decompress some of input into output + * @dstream: The zstd streaming decompression context. + * @output: Destination buffer. `output.pos` is updated to indicate how much + * decompressed data was written. + * @input: Source buffer. `input.pos` is updated to indicate how much data was + * read. Note that it may not consume the entire input, in which case + * `input.pos < input.size`, and it's up to the caller to present + * remaining data again. + * + * The `input` and `output` buffers may be any size. Guaranteed to make some + * forward progress if `input` and `output` are not empty. + * zstd_decompress_stream() will not consume the last byte of the frame until + * the entire frame is flushed. + * + * Return: Returns 0 iff a frame is completely decoded and fully flushed. + * Otherwise returns a hint for the number of bytes to use as the + * input for the next function call or an error, which can be checked + * using zstd_is_error(). The size hint will never load more than the + * frame. + */ +size_t zstd_decompress_stream(zstd_dstream *dstream, zstd_out_buffer *output, + zstd_in_buffer *input); + +/* ====== Frame Inspection Functions ====== */ + +/** + * zstd_find_frame_compressed_size() - returns the size of a compressed frame + * @src: Source buffer. It should point to the start of a zstd encoded + * frame or a skippable frame. + * @src_size: The size of the source buffer. It must be at least as large as the + * size of the frame. + * + * Return: The compressed size of the frame pointed to by `src` or an error, + * which can be check with zstd_is_error(). + * Suitable to pass to ZSTD_decompress() or similar functions. + */ +size_t zstd_find_frame_compressed_size(const void *src, size_t src_size); + +/** + * struct zstd_frame_params - zstd frame parameters stored in the frame header + * @frameContentSize: The frame content size, or ZSTD_CONTENTSIZE_UNKNOWN if not + * present. + * @windowSize: The window size, or 0 if the frame is a skippable frame. + * @blockSizeMax: The maximum block size. + * @frameType: The frame type (zstd or skippable) + * @headerSize: The size of the frame header. + * @dictID: The dictionary id, or 0 if not present. + * @checksumFlag: Whether a checksum was used. + * + * See zstd_lib.h. + */ +typedef ZSTD_frameHeader zstd_frame_header; + +/** + * zstd_get_frame_header() - extracts parameters from a zstd or skippable frame + * @params: On success the frame parameters are written here. + * @src: The source buffer. It must point to a zstd or skippable frame. + * @src_size: The size of the source buffer. + * + * Return: 0 on success. If more data is required it returns how many bytes + * must be provided to make forward progress. Otherwise it returns + * an error, which can be checked using zstd_is_error(). + */ +size_t zstd_get_frame_header(zstd_frame_header *params, const void *src, + size_t src_size); + +#endif /* LINUX_ZSTD_H */ diff --git a/3rdparty/zstd/contrib/linux-kernel/mem.h b/3rdparty/zstd/contrib/linux-kernel/mem.h new file mode 100644 index 00000000000..a7231822b6e --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/mem.h @@ -0,0 +1,261 @@ +/* SPDX-License-Identifier: GPL-2.0+ OR BSD-3-Clause */ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ + +#ifndef MEM_H_MODULE +#define MEM_H_MODULE + +/*-**************************************** +* Dependencies +******************************************/ +#include <asm/unaligned.h> /* get_unaligned, put_unaligned* */ +#include <linux/compiler.h> /* inline */ +#include <linux/swab.h> /* swab32, swab64 */ +#include <linux/types.h> /* size_t, ptrdiff_t */ +#include "debug.h" /* DEBUG_STATIC_ASSERT */ + +/*-**************************************** +* Compiler specifics +******************************************/ +#define MEM_STATIC static inline + +/*-************************************************************** +* Basic Types +*****************************************************************/ +typedef uint8_t BYTE; +typedef uint8_t U8; +typedef int8_t S8; +typedef uint16_t U16; +typedef int16_t S16; +typedef uint32_t U32; +typedef int32_t S32; +typedef uint64_t U64; +typedef int64_t S64; + +/*-************************************************************** +* Memory I/O API +*****************************************************************/ +/*=== Static platform detection ===*/ +MEM_STATIC unsigned MEM_32bits(void); +MEM_STATIC unsigned MEM_64bits(void); +MEM_STATIC unsigned MEM_isLittleEndian(void); + +/*=== Native unaligned read/write ===*/ +MEM_STATIC U16 MEM_read16(const void* memPtr); +MEM_STATIC U32 MEM_read32(const void* memPtr); +MEM_STATIC U64 MEM_read64(const void* memPtr); +MEM_STATIC size_t MEM_readST(const void* memPtr); + +MEM_STATIC void MEM_write16(void* memPtr, U16 value); +MEM_STATIC void MEM_write32(void* memPtr, U32 value); +MEM_STATIC void MEM_write64(void* memPtr, U64 value); + +/*=== Little endian unaligned read/write ===*/ +MEM_STATIC U16 MEM_readLE16(const void* memPtr); +MEM_STATIC U32 MEM_readLE24(const void* memPtr); +MEM_STATIC U32 MEM_readLE32(const void* memPtr); +MEM_STATIC U64 MEM_readLE64(const void* memPtr); +MEM_STATIC size_t MEM_readLEST(const void* memPtr); + +MEM_STATIC void MEM_writeLE16(void* memPtr, U16 val); +MEM_STATIC void MEM_writeLE24(void* memPtr, U32 val); +MEM_STATIC void MEM_writeLE32(void* memPtr, U32 val32); +MEM_STATIC void MEM_writeLE64(void* memPtr, U64 val64); +MEM_STATIC void MEM_writeLEST(void* memPtr, size_t val); + +/*=== Big endian unaligned read/write ===*/ +MEM_STATIC U32 MEM_readBE32(const void* memPtr); +MEM_STATIC U64 MEM_readBE64(const void* memPtr); +MEM_STATIC size_t MEM_readBEST(const void* memPtr); + +MEM_STATIC void MEM_writeBE32(void* memPtr, U32 val32); +MEM_STATIC void MEM_writeBE64(void* memPtr, U64 val64); +MEM_STATIC void MEM_writeBEST(void* memPtr, size_t val); + +/*=== Byteswap ===*/ +MEM_STATIC U32 MEM_swap32(U32 in); +MEM_STATIC U64 MEM_swap64(U64 in); +MEM_STATIC size_t MEM_swapST(size_t in); + +/*-************************************************************** +* Memory I/O Implementation +*****************************************************************/ +MEM_STATIC unsigned MEM_32bits(void) +{ + return sizeof(size_t) == 4; +} + +MEM_STATIC unsigned MEM_64bits(void) +{ + return sizeof(size_t) == 8; +} + +#if defined(__LITTLE_ENDIAN) +#define MEM_LITTLE_ENDIAN 1 +#else +#define MEM_LITTLE_ENDIAN 0 +#endif + +MEM_STATIC unsigned MEM_isLittleEndian(void) +{ + return MEM_LITTLE_ENDIAN; +} + +MEM_STATIC U16 MEM_read16(const void *memPtr) +{ + return get_unaligned((const U16 *)memPtr); +} + +MEM_STATIC U32 MEM_read32(const void *memPtr) +{ + return get_unaligned((const U32 *)memPtr); +} + +MEM_STATIC U64 MEM_read64(const void *memPtr) +{ + return get_unaligned((const U64 *)memPtr); +} + +MEM_STATIC size_t MEM_readST(const void *memPtr) +{ + return get_unaligned((const size_t *)memPtr); +} + +MEM_STATIC void MEM_write16(void *memPtr, U16 value) +{ + put_unaligned(value, (U16 *)memPtr); +} + +MEM_STATIC void MEM_write32(void *memPtr, U32 value) +{ + put_unaligned(value, (U32 *)memPtr); +} + +MEM_STATIC void MEM_write64(void *memPtr, U64 value) +{ + put_unaligned(value, (U64 *)memPtr); +} + +/*=== Little endian r/w ===*/ + +MEM_STATIC U16 MEM_readLE16(const void *memPtr) +{ + return get_unaligned_le16(memPtr); +} + +MEM_STATIC void MEM_writeLE16(void *memPtr, U16 val) +{ + put_unaligned_le16(val, memPtr); +} + +MEM_STATIC U32 MEM_readLE24(const void *memPtr) +{ + return MEM_readLE16(memPtr) + (((const BYTE *)memPtr)[2] << 16); +} + +MEM_STATIC void MEM_writeLE24(void *memPtr, U32 val) +{ + MEM_writeLE16(memPtr, (U16)val); + ((BYTE *)memPtr)[2] = (BYTE)(val >> 16); +} + +MEM_STATIC U32 MEM_readLE32(const void *memPtr) +{ + return get_unaligned_le32(memPtr); +} + +MEM_STATIC void MEM_writeLE32(void *memPtr, U32 val32) +{ + put_unaligned_le32(val32, memPtr); +} + +MEM_STATIC U64 MEM_readLE64(const void *memPtr) +{ + return get_unaligned_le64(memPtr); +} + +MEM_STATIC void MEM_writeLE64(void *memPtr, U64 val64) +{ + put_unaligned_le64(val64, memPtr); +} + +MEM_STATIC size_t MEM_readLEST(const void *memPtr) +{ + if (MEM_32bits()) + return (size_t)MEM_readLE32(memPtr); + else + return (size_t)MEM_readLE64(memPtr); +} + +MEM_STATIC void MEM_writeLEST(void *memPtr, size_t val) +{ + if (MEM_32bits()) + MEM_writeLE32(memPtr, (U32)val); + else + MEM_writeLE64(memPtr, (U64)val); +} + +/*=== Big endian r/w ===*/ + +MEM_STATIC U32 MEM_readBE32(const void *memPtr) +{ + return get_unaligned_be32(memPtr); +} + +MEM_STATIC void MEM_writeBE32(void *memPtr, U32 val32) +{ + put_unaligned_be32(val32, memPtr); +} + +MEM_STATIC U64 MEM_readBE64(const void *memPtr) +{ + return get_unaligned_be64(memPtr); +} + +MEM_STATIC void MEM_writeBE64(void *memPtr, U64 val64) +{ + put_unaligned_be64(val64, memPtr); +} + +MEM_STATIC size_t MEM_readBEST(const void *memPtr) +{ + if (MEM_32bits()) + return (size_t)MEM_readBE32(memPtr); + else + return (size_t)MEM_readBE64(memPtr); +} + +MEM_STATIC void MEM_writeBEST(void *memPtr, size_t val) +{ + if (MEM_32bits()) + MEM_writeBE32(memPtr, (U32)val); + else + MEM_writeBE64(memPtr, (U64)val); +} + +MEM_STATIC U32 MEM_swap32(U32 in) +{ + return swab32(in); +} + +MEM_STATIC U64 MEM_swap64(U64 in) +{ + return swab64(in); +} + +MEM_STATIC size_t MEM_swapST(size_t in) +{ + if (MEM_32bits()) + return (size_t)MEM_swap32((U32)in); + else + return (size_t)MEM_swap64((U64)in); +} + +#endif /* MEM_H_MODULE */ diff --git a/3rdparty/zstd/contrib/linux-kernel/squashfs-benchmark.sh b/3rdparty/zstd/contrib/linux-kernel/squashfs-benchmark.sh new file mode 100755 index 00000000000..02dfd7325bd --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/squashfs-benchmark.sh @@ -0,0 +1,39 @@ +# !/bin/sh +set -e + +# Benchmarks run on a Ubuntu 14.04 VM with 2 cores and 4 GiB of RAM. +# The VM is running on a Macbook Pro with a 3.1 GHz Intel Core i7 processor and +# 16 GB of RAM and an SSD. + +# $BENCHMARK_DIR is generated with the following commands, from the Ubuntu image +# ubuntu-16.10-desktop-amd64.iso. +# > mkdir mnt +# > sudo mount -o loop ubuntu-16.10-desktop-amd64.iso mnt +# > cp mnt/casper/filesystem.squashfs . +# > sudo unsquashfs filesystem.squashfs + +# $HOME is on a ext4 filesystem +BENCHMARK_DIR="$HOME/squashfs-root/" +BENCHMARK_FS="$HOME/filesystem.squashfs" + +# Normalize the environment +sudo rm -f $BENCHMARK_FS 2> /dev/null > /dev/null || true +sudo umount /mnt/squashfs 2> /dev/null > /dev/null || true + +# Run the benchmark +echo "Compression" +echo "sudo mksquashfs $BENCHMARK_DIR $BENCHMARK_FS $@" +time sudo mksquashfs $BENCHMARK_DIR $BENCHMARK_FS $@ 2> /dev/null > /dev/null + +echo "Approximate compression ratio" +printf "%d / %d\n" \ + $(sudo du -sx --block-size=1 $BENCHMARK_DIR | cut -f1) \ + $(sudo du -sx --block-size=1 $BENCHMARK_FS | cut -f1); + +# Mount the filesystem +sudo mount -t squashfs $BENCHMARK_FS /mnt/squashfs + +echo "Decompression" +time sudo tar -c /mnt/squashfs 2> /dev/null | wc -c > /dev/null + +sudo umount /mnt/squashfs diff --git a/3rdparty/zstd/contrib/linux-kernel/test/Makefile b/3rdparty/zstd/contrib/linux-kernel/test/Makefile new file mode 100644 index 00000000000..67b55e66504 --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/test/Makefile @@ -0,0 +1,49 @@ +# ################################################################ +# Copyright (c) Meta Platforms, Inc. and affiliates. +# All rights reserved. +# +# This source code is licensed under both the BSD-style license (found in the +# LICENSE file in the root directory of this source tree) and the GPLv2 (found +# in the COPYING file in the root directory of this source tree). +# You may select, at your option, one of the above-listed licenses. +# ################################################################ + +LINUX := ../linux +LINUX_ZSTDLIB := $(LINUX)/lib/zstd + +CPPFLAGS += -I$(LINUX)/include -I$(LINUX_ZSTDLIB) -Iinclude -DNDEBUG -Wno-deprecated-declarations +# Don't poison the workspace, it currently doesn't work with static allocation and workspace reuse +CPPFLAGS += -DZSTD_ASAN_DONT_POISON_WORKSPACE + +LINUX_ZSTD_MODULE := $(wildcard $(LINUX_ZSTDLIB)/*.c) +LINUX_ZSTD_COMMON := $(wildcard $(LINUX_ZSTDLIB)/common/*.c) +LINUX_ZSTD_COMPRESS := $(wildcard $(LINUX_ZSTDLIB)/compress/*.c) +LINUX_ZSTD_DECOMPRESS := $(wildcard $(LINUX_ZSTDLIB)/decompress/*.c $(LINUX_ZSTDLIB)/decompress/*.S) +LINUX_ZSTD_FILES := $(LINUX_ZSTD_MODULE) $(LINUX_ZSTD_COMMON) $(LINUX_ZSTD_COMPRESS) $(LINUX_ZSTD_DECOMPRESS) +LINUX_ZSTD_OBJECTS0 := $(LINUX_ZSTD_FILES:.c=.o) +LINUX_ZSTD_OBJECTS := $(LINUX_ZSTD_OBJECTS0:.S=.o) + +%.o: %.S + $(COMPILE.S) $(OUTPUT_OPTION) $< + +liblinuxzstd.a: $(LINUX_ZSTD_OBJECTS) + $(AR) $(ARFLAGS) $@ $^ + +test: test.c liblinuxzstd.a + $(CC) $(LDFLAGS) $(CPPFLAGS) $(CFLAGS) $^ -o $@ + +static_test: static_test.c + $(CC) $(LDFLAGS) $(CPPFLAGS) $(CFLAGS) $^ -o $@ + +run-test: test static_test + ./macro-test.sh + ./test + ./static_test + +.PHONY: +clean: + $(RM) -f $(LINUX_ZSTDLIB)/*.o + $(RM) -f $(LINUX_ZSTDLIB)/**/*.o + $(RM) -f *.o *.a + $(RM) -f static_test + $(RM) -f test diff --git a/3rdparty/zstd/contrib/linux-kernel/test/include/asm/unaligned.h b/3rdparty/zstd/contrib/linux-kernel/test/include/asm/unaligned.h new file mode 100644 index 00000000000..86ec4ca3899 --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/test/include/asm/unaligned.h @@ -0,0 +1,187 @@ +#ifndef ASM_UNALIGNED_H +#define ASM_UNALIGNED_H + +#include <assert.h> +#include <linux/types.h> + +#ifndef __LITTLE_ENDIAN +# if (__BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__) || defined(__LITTLE_ENDIAN__) +# define __LITTLE_ENDIAN 1 +# endif +#endif + +#ifdef __LITTLE_ENDIAN +# define _IS_LITTLE_ENDIAN 1 +#else +# define _IS_LITTLE_ENDIAN 0 +#endif + +static unsigned _isLittleEndian(void) +{ + const union { uint32_t u; uint8_t c[4]; } one = { 1 }; + assert(_IS_LITTLE_ENDIAN == one.c[0]); + (void)one; + return _IS_LITTLE_ENDIAN; +} + +static uint16_t _swap16(uint16_t in) +{ + return ((in & 0xF) << 8) + ((in & 0xF0) >> 8); +} + +static uint32_t _swap32(uint32_t in) +{ + return __builtin_bswap32(in); +} + +static uint64_t _swap64(uint64_t in) +{ + return __builtin_bswap64(in); +} + +/* Little endian */ +static uint16_t get_unaligned_le16(const void* memPtr) +{ + uint16_t val; + __builtin_memcpy(&val, memPtr, sizeof(val)); + if (!_isLittleEndian()) _swap16(val); + return val; +} + +static uint32_t get_unaligned_le32(const void* memPtr) +{ + uint32_t val; + __builtin_memcpy(&val, memPtr, sizeof(val)); + if (!_isLittleEndian()) _swap32(val); + return val; +} + +static uint64_t get_unaligned_le64(const void* memPtr) +{ + uint64_t val; + __builtin_memcpy(&val, memPtr, sizeof(val)); + if (!_isLittleEndian()) _swap64(val); + return val; +} + +static void put_unaligned_le16(uint16_t value, void* memPtr) +{ + if (!_isLittleEndian()) value = _swap16(value); + __builtin_memcpy(memPtr, &value, sizeof(value)); +} + +static void put_unaligned_le32(uint32_t value, void* memPtr) +{ + if (!_isLittleEndian()) value = _swap32(value); + __builtin_memcpy(memPtr, &value, sizeof(value)); +} + +static void put_unaligned_le64(uint64_t value, void* memPtr) +{ + if (!_isLittleEndian()) value = _swap64(value); + __builtin_memcpy(memPtr, &value, sizeof(value)); +} + +/* big endian */ +static uint32_t get_unaligned_be32(const void* memPtr) +{ + uint32_t val; + __builtin_memcpy(&val, memPtr, sizeof(val)); + if (_isLittleEndian()) _swap32(val); + return val; +} + +static uint64_t get_unaligned_be64(const void* memPtr) +{ + uint64_t val; + __builtin_memcpy(&val, memPtr, sizeof(val)); + if (_isLittleEndian()) _swap64(val); + return val; +} + +static void put_unaligned_be32(uint32_t value, void* memPtr) +{ + if (_isLittleEndian()) value = _swap32(value); + __builtin_memcpy(memPtr, &value, sizeof(value)); +} + +static void put_unaligned_be64(uint64_t value, void* memPtr) +{ + if (_isLittleEndian()) value = _swap64(value); + __builtin_memcpy(memPtr, &value, sizeof(value)); +} + +/* generic */ +extern void __bad_unaligned_access_size(void); + +#define __get_unaligned_le(ptr) ((typeof(*(ptr)))({ \ + __builtin_choose_expr(sizeof(*(ptr)) == 1, *(ptr), \ + __builtin_choose_expr(sizeof(*(ptr)) == 2, get_unaligned_le16((ptr)), \ + __builtin_choose_expr(sizeof(*(ptr)) == 4, get_unaligned_le32((ptr)), \ + __builtin_choose_expr(sizeof(*(ptr)) == 8, get_unaligned_le64((ptr)), \ + __bad_unaligned_access_size())))); \ + })) + +#define __get_unaligned_be(ptr) ((typeof(*(ptr)))({ \ + __builtin_choose_expr(sizeof(*(ptr)) == 1, *(ptr), \ + __builtin_choose_expr(sizeof(*(ptr)) == 2, get_unaligned_be16((ptr)), \ + __builtin_choose_expr(sizeof(*(ptr)) == 4, get_unaligned_be32((ptr)), \ + __builtin_choose_expr(sizeof(*(ptr)) == 8, get_unaligned_be64((ptr)), \ + __bad_unaligned_access_size())))); \ + })) + +#define __put_unaligned_le(val, ptr) \ + ({ \ + void *__gu_p = (ptr); \ + switch (sizeof(*(ptr))) { \ + case 1: \ + *(uint8_t *)__gu_p = (uint8_t)(val); \ + break; \ + case 2: \ + put_unaligned_le16((uint16_t)(val), __gu_p); \ + break; \ + case 4: \ + put_unaligned_le32((uint32_t)(val), __gu_p); \ + break; \ + case 8: \ + put_unaligned_le64((uint64_t)(val), __gu_p); \ + break; \ + default: \ + __bad_unaligned_access_size(); \ + break; \ + } \ + (void)0; \ + }) + +#define __put_unaligned_be(val, ptr) \ + ({ \ + void *__gu_p = (ptr); \ + switch (sizeof(*(ptr))) { \ + case 1: \ + *(uint8_t *)__gu_p = (uint8_t)(val); \ + break; \ + case 2: \ + put_unaligned_be16((uint16_t)(val), __gu_p); \ + break; \ + case 4: \ + put_unaligned_be32((uint32_t)(val), __gu_p); \ + break; \ + case 8: \ + put_unaligned_be64((uint64_t)(val), __gu_p); \ + break; \ + default: \ + __bad_unaligned_access_size(); \ + break; \ + } \ + (void)0; \ + }) + +#if _IS_LITTLE_ENDIAN +# define get_unaligned __get_unaligned_le +# define put_unaligned __put_unaligned_le +#else +# define get_unaligned __get_unaligned_be +# define put_unaligned __put_unaligned_be +#endif + +#endif // ASM_UNALIGNED_H diff --git a/3rdparty/zstd/contrib/linux-kernel/test/include/linux/compiler.h b/3rdparty/zstd/contrib/linux-kernel/test/include/linux/compiler.h new file mode 100644 index 00000000000..988ce4a20ff --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/test/include/linux/compiler.h @@ -0,0 +1,23 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ +#ifndef LINUX_COMPILER_H +#define LINUX_COMPILER_H + +#ifndef inline +#define inline __inline __attribute__((unused)) +#endif + +#ifndef noinline +#define noinline __attribute__((noinline)) +#endif + +#define fallthrough __attribute__((__fallthrough__)) + +#endif diff --git a/3rdparty/zstd/contrib/linux-kernel/test/include/linux/errno.h b/3rdparty/zstd/contrib/linux-kernel/test/include/linux/errno.h new file mode 100644 index 00000000000..b4bdcba0ea6 --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/test/include/linux/errno.h @@ -0,0 +1,15 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ +#ifndef LINUX_ERRNO_H +#define LINUX_ERRNO_H + +#define EINVAL 22 + +#endif diff --git a/3rdparty/zstd/contrib/linux-kernel/test/include/linux/kernel.h b/3rdparty/zstd/contrib/linux-kernel/test/include/linux/kernel.h new file mode 100644 index 00000000000..a4d791cd1d5 --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/test/include/linux/kernel.h @@ -0,0 +1,19 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ +#ifndef LINUX_KERNEL_H +#define LINUX_KERNEL_H + +#define WARN_ON(x) + +#define PTR_ALIGN(p, a) (typeof(p))ALIGN((unsigned long long)(p), (a)) +#define ALIGN(x, a) ALIGN_MASK((x), (a) - 1) +#define ALIGN_MASK(x, mask) (((x) + (mask)) & ~(mask)) + +#endif diff --git a/3rdparty/zstd/contrib/linux-kernel/test/include/linux/limits.h b/3rdparty/zstd/contrib/linux-kernel/test/include/linux/limits.h new file mode 100644 index 00000000000..574aa7b343f --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/test/include/linux/limits.h @@ -0,0 +1,15 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ +#ifndef LINUX_LIMITS_H +#define LINUX_LIMITS_H + +#include <limits.h> + +#endif diff --git a/3rdparty/zstd/contrib/linux-kernel/test/include/linux/math64.h b/3rdparty/zstd/contrib/linux-kernel/test/include/linux/math64.h new file mode 100644 index 00000000000..7f6713e73bc --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/test/include/linux/math64.h @@ -0,0 +1,15 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ +#ifndef LINUX_MATH64_H +#define LINUX_MATH64_H + +#define div_u64(dividend, divisor) ((dividend) / (divisor)) + +#endif diff --git a/3rdparty/zstd/contrib/linux-kernel/test/include/linux/module.h b/3rdparty/zstd/contrib/linux-kernel/test/include/linux/module.h new file mode 100644 index 00000000000..06ef56f9ebe --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/test/include/linux/module.h @@ -0,0 +1,20 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ +#ifndef LINUX_MODULE_H +#define LINUX_MODULE_H + +#define EXPORT_SYMBOL(symbol) \ + void* __##symbol = symbol +#define EXPORT_SYMBOL_GPL(symbol) \ + void* __##symbol = symbol +#define MODULE_LICENSE(license) +#define MODULE_DESCRIPTION(description) + +#endif diff --git a/3rdparty/zstd/contrib/linux-kernel/test/include/linux/printk.h b/3rdparty/zstd/contrib/linux-kernel/test/include/linux/printk.h new file mode 100644 index 00000000000..92a25278e71 --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/test/include/linux/printk.h @@ -0,0 +1,15 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ +#ifndef LINUX_PRINTK_H +#define LINUX_PRINTK_H + +#define pr_debug(...) + +#endif diff --git a/3rdparty/zstd/contrib/linux-kernel/test/include/linux/stddef.h b/3rdparty/zstd/contrib/linux-kernel/test/include/linux/stddef.h new file mode 100644 index 00000000000..15c7408fcdf --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/test/include/linux/stddef.h @@ -0,0 +1,15 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ +#ifndef LINUX_STDDEF_H +#define LINUX_STDDEF_H + +#include <stddef.h> + +#endif diff --git a/3rdparty/zstd/contrib/linux-kernel/test/include/linux/swab.h b/3rdparty/zstd/contrib/linux-kernel/test/include/linux/swab.h new file mode 100644 index 00000000000..2b48b434c97 --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/test/include/linux/swab.h @@ -0,0 +1,16 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ +#ifndef LINUX_SWAB_H +#define LINUX_SWAB_H + +#define swab32(x) __builtin_bswap32((x)) +#define swab64(x) __builtin_bswap64((x)) + +#endif diff --git a/3rdparty/zstd/contrib/linux-kernel/test/include/linux/types.h b/3rdparty/zstd/contrib/linux-kernel/test/include/linux/types.h new file mode 100644 index 00000000000..b413db6f983 --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/test/include/linux/types.h @@ -0,0 +1,16 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ +#ifndef LINUX_TYPES_H +#define LINUX_TYPES_H + +#include <stddef.h> +#include <stdint.h> + +#endif diff --git a/3rdparty/zstd/contrib/linux-kernel/test/include/linux/xxhash.h b/3rdparty/zstd/contrib/linux-kernel/test/include/linux/xxhash.h new file mode 100644 index 00000000000..d41cbd93318 --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/test/include/linux/xxhash.h @@ -0,0 +1,745 @@ +/* + * xxHash - Extremely Fast Hash algorithm + * Copyright (C) 2012-2016, Yann Collet. + * + * BSD 2-Clause License (https://opensource.org/licenses/bsd-license.php) + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions are + * met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following disclaimer + * in the documentation and/or other materials provided with the + * distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * This program is free software; you can redistribute it and/or modify it under + * the terms of the GNU General Public License version 2 as published by the + * Free Software Foundation. This program is dual-licensed; you may select + * either version 2 of the GNU General Public License ("GPL") or BSD license + * ("BSD"). + * + * You can contact the author at: + * - xxHash homepage: https://cyan4973.github.io/xxHash/ + * - xxHash source repository: https://github.com/Cyan4973/xxHash + */ + +/* + * Notice extracted from xxHash homepage: + * + * xxHash is an extremely fast Hash algorithm, running at RAM speed limits. + * It also successfully passes all tests from the SMHasher suite. + * + * Comparison (single thread, Windows Seven 32 bits, using SMHasher on a Core 2 + * Duo @3GHz) + * + * Name Speed Q.Score Author + * xxHash 5.4 GB/s 10 + * CrapWow 3.2 GB/s 2 Andrew + * MumurHash 3a 2.7 GB/s 10 Austin Appleby + * SpookyHash 2.0 GB/s 10 Bob Jenkins + * SBox 1.4 GB/s 9 Bret Mulvey + * Lookup3 1.2 GB/s 9 Bob Jenkins + * SuperFastHash 1.2 GB/s 1 Paul Hsieh + * CityHash64 1.05 GB/s 10 Pike & Alakuijala + * FNV 0.55 GB/s 5 Fowler, Noll, Vo + * CRC32 0.43 GB/s 9 + * MD5-32 0.33 GB/s 10 Ronald L. Rivest + * SHA1-32 0.28 GB/s 10 + * + * Q.Score is a measure of quality of the hash function. + * It depends on successfully passing SMHasher test set. + * 10 is a perfect score. + * + * A 64-bits version, named xxh64 offers much better speed, + * but for 64-bits applications only. + * Name Speed on 64 bits Speed on 32 bits + * xxh64 13.8 GB/s 1.9 GB/s + * xxh32 6.8 GB/s 6.0 GB/s + */ + +#ifndef XXHASH_H +#define XXHASH_H + +#include <linux/types.h> + +#define XXH_API static inline __attribute__((unused)) +/*-**************************** + * Simple Hash Functions + *****************************/ + +/** + * xxh32() - calculate the 32-bit hash of the input with a given seed. + * + * @input: The data to hash. + * @length: The length of the data to hash. + * @seed: The seed can be used to alter the result predictably. + * + * Speed on Core 2 Duo @ 3 GHz (single thread, SMHasher benchmark) : 5.4 GB/s + * + * Return: The 32-bit hash of the data. + */ +XXH_API uint32_t xxh32(const void *input, size_t length, uint32_t seed); + +/** + * xxh64() - calculate the 64-bit hash of the input with a given seed. + * + * @input: The data to hash. + * @length: The length of the data to hash. + * @seed: The seed can be used to alter the result predictably. + * + * This function runs 2x faster on 64-bit systems, but slower on 32-bit systems. + * + * Return: The 64-bit hash of the data. + */ +XXH_API uint64_t xxh64(const void *input, size_t length, uint64_t seed); + +/** + * xxhash() - calculate wordsize hash of the input with a given seed + * @input: The data to hash. + * @length: The length of the data to hash. + * @seed: The seed can be used to alter the result predictably. + * + * If the hash does not need to be comparable between machines with + * different word sizes, this function will call whichever of xxh32() + * or xxh64() is faster. + * + * Return: wordsize hash of the data. + */ + +static inline unsigned long xxhash(const void *input, size_t length, + uint64_t seed) +{ + if (sizeof(size_t) == 8) + return xxh64(input, length, seed); + else + return xxh32(input, length, seed); +} + +/*-**************************** + * Streaming Hash Functions + *****************************/ + +/* + * These definitions are only meant to allow allocation of XXH state + * statically, on stack, or in a struct for example. + * Do not use members directly. + */ + +/** + * struct xxh32_state - private xxh32 state, do not use members directly + */ +struct xxh32_state { + uint32_t total_len_32; + uint32_t large_len; + uint32_t v1; + uint32_t v2; + uint32_t v3; + uint32_t v4; + uint32_t mem32[4]; + uint32_t memsize; +}; + +/** + * struct xxh32_state - private xxh64 state, do not use members directly + */ +struct xxh64_state { + uint64_t total_len; + uint64_t v1; + uint64_t v2; + uint64_t v3; + uint64_t v4; + uint64_t mem64[4]; + uint32_t memsize; +}; + +/** + * xxh32_reset() - reset the xxh32 state to start a new hashing operation + * + * @state: The xxh32 state to reset. + * @seed: Initialize the hash state with this seed. + * + * Call this function on any xxh32_state to prepare for a new hashing operation. + */ +XXH_API void xxh32_reset(struct xxh32_state *state, uint32_t seed); + +/** + * xxh32_update() - hash the data given and update the xxh32 state + * + * @state: The xxh32 state to update. + * @input: The data to hash. + * @length: The length of the data to hash. + * + * After calling xxh32_reset() call xxh32_update() as many times as necessary. + * + * Return: Zero on success, otherwise an error code. + */ +XXH_API int xxh32_update(struct xxh32_state *state, const void *input, size_t length); + +/** + * xxh32_digest() - produce the current xxh32 hash + * + * @state: Produce the current xxh32 hash of this state. + * + * A hash value can be produced at any time. It is still possible to continue + * inserting input into the hash state after a call to xxh32_digest(), and + * generate new hashes later on, by calling xxh32_digest() again. + * + * Return: The xxh32 hash stored in the state. + */ +XXH_API uint32_t xxh32_digest(const struct xxh32_state *state); + +/** + * xxh64_reset() - reset the xxh64 state to start a new hashing operation + * + * @state: The xxh64 state to reset. + * @seed: Initialize the hash state with this seed. + */ +XXH_API void xxh64_reset(struct xxh64_state *state, uint64_t seed); + +/** + * xxh64_update() - hash the data given and update the xxh64 state + * @state: The xxh64 state to update. + * @input: The data to hash. + * @length: The length of the data to hash. + * + * After calling xxh64_reset() call xxh64_update() as many times as necessary. + * + * Return: Zero on success, otherwise an error code. + */ +XXH_API int xxh64_update(struct xxh64_state *state, const void *input, size_t length); + +/** + * xxh64_digest() - produce the current xxh64 hash + * + * @state: Produce the current xxh64 hash of this state. + * + * A hash value can be produced at any time. It is still possible to continue + * inserting input into the hash state after a call to xxh64_digest(), and + * generate new hashes later on, by calling xxh64_digest() again. + * + * Return: The xxh64 hash stored in the state. + */ +XXH_API uint64_t xxh64_digest(const struct xxh64_state *state); + +/*-************************** + * Utils + ***************************/ + +/** + * xxh32_copy_state() - copy the source state into the destination state + * + * @src: The source xxh32 state. + * @dst: The destination xxh32 state. + */ +XXH_API void xxh32_copy_state(struct xxh32_state *dst, const struct xxh32_state *src); + +/** + * xxh64_copy_state() - copy the source state into the destination state + * + * @src: The source xxh64 state. + * @dst: The destination xxh64 state. + */ +XXH_API void xxh64_copy_state(struct xxh64_state *dst, const struct xxh64_state *src); + +/* + * xxHash - Extremely Fast Hash algorithm + * Copyright (C) 2012-2016, Yann Collet. + * + * BSD 2-Clause License (https://opensource.org/licenses/bsd-license.php) + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions are + * met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following disclaimer + * in the documentation and/or other materials provided with the + * distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * This program is free software; you can redistribute it and/or modify it under + * the terms of the GNU General Public License version 2 as published by the + * Free Software Foundation. This program is dual-licensed; you may select + * either version 2 of the GNU General Public License ("GPL") or BSD license + * ("BSD"). + * + * You can contact the author at: + * - xxHash homepage: https://cyan4973.github.io/xxHash/ + * - xxHash source repository: https://github.com/Cyan4973/xxHash + */ + +#include <asm/unaligned.h> +#include <linux/errno.h> +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/xxhash.h> + +/*-************************************* + * Macros + **************************************/ +#define xxh_rotl32(x, r) ((x << r) | (x >> (32 - r))) +#define xxh_rotl64(x, r) ((x << r) | (x >> (64 - r))) + +#ifdef __LITTLE_ENDIAN +# define XXH_CPU_LITTLE_ENDIAN 1 +#else +# define XXH_CPU_LITTLE_ENDIAN 0 +#endif + +/*-************************************* + * Constants + **************************************/ +static const uint32_t PRIME32_1 = 2654435761U; +static const uint32_t PRIME32_2 = 2246822519U; +static const uint32_t PRIME32_3 = 3266489917U; +static const uint32_t PRIME32_4 = 668265263U; +static const uint32_t PRIME32_5 = 374761393U; + +static const uint64_t PRIME64_1 = 11400714785074694791ULL; +static const uint64_t PRIME64_2 = 14029467366897019727ULL; +static const uint64_t PRIME64_3 = 1609587929392839161ULL; +static const uint64_t PRIME64_4 = 9650029242287828579ULL; +static const uint64_t PRIME64_5 = 2870177450012600261ULL; + +/*-************************** + * Utils + ***************************/ +XXH_API void xxh32_copy_state(struct xxh32_state *dst, const struct xxh32_state *src) +{ + __builtin_memcpy(dst, src, sizeof(*dst)); +} + +XXH_API void xxh64_copy_state(struct xxh64_state *dst, const struct xxh64_state *src) +{ + __builtin_memcpy(dst, src, sizeof(*dst)); +} + +/*-*************************** + * Simple Hash Functions + ****************************/ +static uint32_t xxh32_round(uint32_t seed, const uint32_t input) +{ + seed += input * PRIME32_2; + seed = xxh_rotl32(seed, 13); + seed *= PRIME32_1; + return seed; +} + +XXH_API uint32_t xxh32(const void *input, const size_t len, const uint32_t seed) +{ + const uint8_t *p = (const uint8_t *)input; + const uint8_t *b_end = p + len; + uint32_t h32; + + if (len >= 16) { + const uint8_t *const limit = b_end - 16; + uint32_t v1 = seed + PRIME32_1 + PRIME32_2; + uint32_t v2 = seed + PRIME32_2; + uint32_t v3 = seed + 0; + uint32_t v4 = seed - PRIME32_1; + + do { + v1 = xxh32_round(v1, get_unaligned_le32(p)); + p += 4; + v2 = xxh32_round(v2, get_unaligned_le32(p)); + p += 4; + v3 = xxh32_round(v3, get_unaligned_le32(p)); + p += 4; + v4 = xxh32_round(v4, get_unaligned_le32(p)); + p += 4; + } while (p <= limit); + + h32 = xxh_rotl32(v1, 1) + xxh_rotl32(v2, 7) + + xxh_rotl32(v3, 12) + xxh_rotl32(v4, 18); + } else { + h32 = seed + PRIME32_5; + } + + h32 += (uint32_t)len; + + while (p + 4 <= b_end) { + h32 += get_unaligned_le32(p) * PRIME32_3; + h32 = xxh_rotl32(h32, 17) * PRIME32_4; + p += 4; + } + + while (p < b_end) { + h32 += (*p) * PRIME32_5; + h32 = xxh_rotl32(h32, 11) * PRIME32_1; + p++; + } + + h32 ^= h32 >> 15; + h32 *= PRIME32_2; + h32 ^= h32 >> 13; + h32 *= PRIME32_3; + h32 ^= h32 >> 16; + + return h32; +} + +static uint64_t xxh64_round(uint64_t acc, const uint64_t input) +{ + acc += input * PRIME64_2; + acc = xxh_rotl64(acc, 31); + acc *= PRIME64_1; + return acc; +} + +static uint64_t xxh64_merge_round(uint64_t acc, uint64_t val) +{ + val = xxh64_round(0, val); + acc ^= val; + acc = acc * PRIME64_1 + PRIME64_4; + return acc; +} + +XXH_API uint64_t xxh64(const void *input, const size_t len, const uint64_t seed) +{ + const uint8_t *p = (const uint8_t *)input; + const uint8_t *const b_end = p + len; + uint64_t h64; + + if (len >= 32) { + const uint8_t *const limit = b_end - 32; + uint64_t v1 = seed + PRIME64_1 + PRIME64_2; + uint64_t v2 = seed + PRIME64_2; + uint64_t v3 = seed + 0; + uint64_t v4 = seed - PRIME64_1; + + do { + v1 = xxh64_round(v1, get_unaligned_le64(p)); + p += 8; + v2 = xxh64_round(v2, get_unaligned_le64(p)); + p += 8; + v3 = xxh64_round(v3, get_unaligned_le64(p)); + p += 8; + v4 = xxh64_round(v4, get_unaligned_le64(p)); + p += 8; + } while (p <= limit); + + h64 = xxh_rotl64(v1, 1) + xxh_rotl64(v2, 7) + + xxh_rotl64(v3, 12) + xxh_rotl64(v4, 18); + h64 = xxh64_merge_round(h64, v1); + h64 = xxh64_merge_round(h64, v2); + h64 = xxh64_merge_round(h64, v3); + h64 = xxh64_merge_round(h64, v4); + + } else { + h64 = seed + PRIME64_5; + } + + h64 += (uint64_t)len; + + while (p + 8 <= b_end) { + const uint64_t k1 = xxh64_round(0, get_unaligned_le64(p)); + + h64 ^= k1; + h64 = xxh_rotl64(h64, 27) * PRIME64_1 + PRIME64_4; + p += 8; + } + + if (p + 4 <= b_end) { + h64 ^= (uint64_t)(get_unaligned_le32(p)) * PRIME64_1; + h64 = xxh_rotl64(h64, 23) * PRIME64_2 + PRIME64_3; + p += 4; + } + + while (p < b_end) { + h64 ^= (*p) * PRIME64_5; + h64 = xxh_rotl64(h64, 11) * PRIME64_1; + p++; + } + + h64 ^= h64 >> 33; + h64 *= PRIME64_2; + h64 ^= h64 >> 29; + h64 *= PRIME64_3; + h64 ^= h64 >> 32; + + return h64; +} + +/*-************************************************** + * Advanced Hash Functions + ***************************************************/ +XXH_API void xxh32_reset(struct xxh32_state *statePtr, const uint32_t seed) +{ + /* use a local state for memcpy() to avoid strict-aliasing warnings */ + struct xxh32_state state; + + __builtin_memset(&state, 0, sizeof(state)); + state.v1 = seed + PRIME32_1 + PRIME32_2; + state.v2 = seed + PRIME32_2; + state.v3 = seed + 0; + state.v4 = seed - PRIME32_1; + __builtin_memcpy(statePtr, &state, sizeof(state)); +} + +XXH_API void xxh64_reset(struct xxh64_state *statePtr, const uint64_t seed) +{ + /* use a local state for memcpy() to avoid strict-aliasing warnings */ + struct xxh64_state state; + + __builtin_memset(&state, 0, sizeof(state)); + state.v1 = seed + PRIME64_1 + PRIME64_2; + state.v2 = seed + PRIME64_2; + state.v3 = seed + 0; + state.v4 = seed - PRIME64_1; + __builtin_memcpy(statePtr, &state, sizeof(state)); +} + +XXH_API int xxh32_update(struct xxh32_state *state, const void *input, const size_t len) +{ + const uint8_t *p = (const uint8_t *)input; + const uint8_t *const b_end = p + len; + + if (input == NULL) + return -EINVAL; + + state->total_len_32 += (uint32_t)len; + state->large_len |= (len >= 16) | (state->total_len_32 >= 16); + + if (state->memsize + len < 16) { /* fill in tmp buffer */ + __builtin_memcpy((uint8_t *)(state->mem32) + state->memsize, input, len); + state->memsize += (uint32_t)len; + return 0; + } + + if (state->memsize) { /* some data left from previous update */ + const uint32_t *p32 = state->mem32; + + __builtin_memcpy((uint8_t *)(state->mem32) + state->memsize, input, + 16 - state->memsize); + + state->v1 = xxh32_round(state->v1, get_unaligned_le32(p32)); + p32++; + state->v2 = xxh32_round(state->v2, get_unaligned_le32(p32)); + p32++; + state->v3 = xxh32_round(state->v3, get_unaligned_le32(p32)); + p32++; + state->v4 = xxh32_round(state->v4, get_unaligned_le32(p32)); + p32++; + + p += 16-state->memsize; + state->memsize = 0; + } + + if (p <= b_end - 16) { + const uint8_t *const limit = b_end - 16; + uint32_t v1 = state->v1; + uint32_t v2 = state->v2; + uint32_t v3 = state->v3; + uint32_t v4 = state->v4; + + do { + v1 = xxh32_round(v1, get_unaligned_le32(p)); + p += 4; + v2 = xxh32_round(v2, get_unaligned_le32(p)); + p += 4; + v3 = xxh32_round(v3, get_unaligned_le32(p)); + p += 4; + v4 = xxh32_round(v4, get_unaligned_le32(p)); + p += 4; + } while (p <= limit); + + state->v1 = v1; + state->v2 = v2; + state->v3 = v3; + state->v4 = v4; + } + + if (p < b_end) { + __builtin_memcpy(state->mem32, p, (size_t)(b_end-p)); + state->memsize = (uint32_t)(b_end-p); + } + + return 0; +} + +XXH_API uint32_t xxh32_digest(const struct xxh32_state *state) +{ + const uint8_t *p = (const uint8_t *)state->mem32; + const uint8_t *const b_end = (const uint8_t *)(state->mem32) + + state->memsize; + uint32_t h32; + + if (state->large_len) { + h32 = xxh_rotl32(state->v1, 1) + xxh_rotl32(state->v2, 7) + + xxh_rotl32(state->v3, 12) + xxh_rotl32(state->v4, 18); + } else { + h32 = state->v3 /* == seed */ + PRIME32_5; + } + + h32 += state->total_len_32; + + while (p + 4 <= b_end) { + h32 += get_unaligned_le32(p) * PRIME32_3; + h32 = xxh_rotl32(h32, 17) * PRIME32_4; + p += 4; + } + + while (p < b_end) { + h32 += (*p) * PRIME32_5; + h32 = xxh_rotl32(h32, 11) * PRIME32_1; + p++; + } + + h32 ^= h32 >> 15; + h32 *= PRIME32_2; + h32 ^= h32 >> 13; + h32 *= PRIME32_3; + h32 ^= h32 >> 16; + + return h32; +} + +XXH_API int xxh64_update(struct xxh64_state *state, const void *input, const size_t len) +{ + const uint8_t *p = (const uint8_t *)input; + const uint8_t *const b_end = p + len; + + if (input == NULL) + return -EINVAL; + + state->total_len += len; + + if (state->memsize + len < 32) { /* fill in tmp buffer */ + __builtin_memcpy(((uint8_t *)state->mem64) + state->memsize, input, len); + state->memsize += (uint32_t)len; + return 0; + } + + if (state->memsize) { /* tmp buffer is full */ + uint64_t *p64 = state->mem64; + + __builtin_memcpy(((uint8_t *)p64) + state->memsize, input, + 32 - state->memsize); + + state->v1 = xxh64_round(state->v1, get_unaligned_le64(p64)); + p64++; + state->v2 = xxh64_round(state->v2, get_unaligned_le64(p64)); + p64++; + state->v3 = xxh64_round(state->v3, get_unaligned_le64(p64)); + p64++; + state->v4 = xxh64_round(state->v4, get_unaligned_le64(p64)); + + p += 32 - state->memsize; + state->memsize = 0; + } + + if (p + 32 <= b_end) { + const uint8_t *const limit = b_end - 32; + uint64_t v1 = state->v1; + uint64_t v2 = state->v2; + uint64_t v3 = state->v3; + uint64_t v4 = state->v4; + + do { + v1 = xxh64_round(v1, get_unaligned_le64(p)); + p += 8; + v2 = xxh64_round(v2, get_unaligned_le64(p)); + p += 8; + v3 = xxh64_round(v3, get_unaligned_le64(p)); + p += 8; + v4 = xxh64_round(v4, get_unaligned_le64(p)); + p += 8; + } while (p <= limit); + + state->v1 = v1; + state->v2 = v2; + state->v3 = v3; + state->v4 = v4; + } + + if (p < b_end) { + __builtin_memcpy(state->mem64, p, (size_t)(b_end-p)); + state->memsize = (uint32_t)(b_end - p); + } + + return 0; +} + +XXH_API uint64_t xxh64_digest(const struct xxh64_state *state) +{ + const uint8_t *p = (const uint8_t *)state->mem64; + const uint8_t *const b_end = (const uint8_t *)state->mem64 + + state->memsize; + uint64_t h64; + + if (state->total_len >= 32) { + const uint64_t v1 = state->v1; + const uint64_t v2 = state->v2; + const uint64_t v3 = state->v3; + const uint64_t v4 = state->v4; + + h64 = xxh_rotl64(v1, 1) + xxh_rotl64(v2, 7) + + xxh_rotl64(v3, 12) + xxh_rotl64(v4, 18); + h64 = xxh64_merge_round(h64, v1); + h64 = xxh64_merge_round(h64, v2); + h64 = xxh64_merge_round(h64, v3); + h64 = xxh64_merge_round(h64, v4); + } else { + h64 = state->v3 + PRIME64_5; + } + + h64 += (uint64_t)state->total_len; + + while (p + 8 <= b_end) { + const uint64_t k1 = xxh64_round(0, get_unaligned_le64(p)); + + h64 ^= k1; + h64 = xxh_rotl64(h64, 27) * PRIME64_1 + PRIME64_4; + p += 8; + } + + if (p + 4 <= b_end) { + h64 ^= (uint64_t)(get_unaligned_le32(p)) * PRIME64_1; + h64 = xxh_rotl64(h64, 23) * PRIME64_2 + PRIME64_3; + p += 4; + } + + while (p < b_end) { + h64 ^= (*p) * PRIME64_5; + h64 = xxh_rotl64(h64, 11) * PRIME64_1; + p++; + } + + h64 ^= h64 >> 33; + h64 *= PRIME64_2; + h64 ^= h64 >> 29; + h64 *= PRIME64_3; + h64 ^= h64 >> 32; + + return h64; +} + +#endif /* XXHASH_H */ diff --git a/3rdparty/zstd/contrib/linux-kernel/test/macro-test.sh b/3rdparty/zstd/contrib/linux-kernel/test/macro-test.sh new file mode 100755 index 00000000000..9ea84aa66b1 --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/test/macro-test.sh @@ -0,0 +1,44 @@ +#!/usr/bin/env sh + +set -e + +SCRIPT_DIR=$(cd "$(dirname "$0")" && pwd) +INCLUDE_DIR="$SCRIPT_DIR/../linux/include" +LIB_DIR="$SCRIPT_DIR/../linux/lib" + + +print() { + printf '%b' "${*}" +} + +println() { + printf '%b\n' "${*}" +} + +die() { + println "$@" 1>&2 + exit 1 +} + +test_not_present() { + print "Testing that '$1' is not present... " + grep -r $1 "$INCLUDE_DIR" "$LIB_DIR" && die "Fail!" + println "Okay" +} + +println "This test checks that the macro removal process worked as expected" +println "If this test fails, then freestanding.py wasn't able to remove one of these" +println "macros from the source code completely. You'll either need to rewrite the check" +println "or improve freestanding.py." +println "" + +test_not_present "ZSTD_NO_INTRINSICS" +test_not_present "ZSTD_NO_UNUSED_FUNCTIONS" +test_not_present "ZSTD_LEGACY_SUPPORT" +test_not_present "STATIC_BMI2" +test_not_present "ZSTD_DLL_EXPORT" +test_not_present "ZSTD_DLL_IMPORT" +test_not_present "__ICCARM__" +test_not_present "_MSC_VER" +test_not_present "_WIN32" +test_not_present "__linux__" diff --git a/3rdparty/zstd/contrib/linux-kernel/test/static_test.c b/3rdparty/zstd/contrib/linux-kernel/test/static_test.c new file mode 100644 index 00000000000..ba4a420d417 --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/test/static_test.c @@ -0,0 +1,52 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ +#include <stddef.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> + +#include "decompress_sources.h" +#include <linux/zstd.h> + +#define CONTROL(x) \ + do { \ + if (!(x)) { \ + fprintf(stderr, "%s:%u: %s failed!\n", __FUNCTION__, __LINE__, #x); \ + abort(); \ + } \ + } while (0) + + +static const char kEmptyZstdFrame[] = { + 0x28, 0xb5, 0x2f, 0xfd, 0x24, 0x00, 0x01, 0x00, 0x00, 0x99, 0xe9, 0xd8, 0x51 +}; + +static void test_decompress_unzstd(void) { + fprintf(stderr, "Testing decompress unzstd... "); + { + size_t const wkspSize = zstd_dctx_workspace_bound(); + void* wksp = malloc(wkspSize); + ZSTD_DCtx* dctx = zstd_init_dctx(wksp, wkspSize); + CONTROL(wksp != NULL); + CONTROL(dctx != NULL); + { + size_t const dSize = zstd_decompress_dctx(dctx, NULL, 0, kEmptyZstdFrame, sizeof(kEmptyZstdFrame)); + CONTROL(!zstd_is_error(dSize)); + CONTROL(dSize == 0); + } + free(wksp); + } + fprintf(stderr, "Ok\n"); +} + +int main(void) { + test_decompress_unzstd(); + return 0; +} diff --git a/3rdparty/zstd/contrib/linux-kernel/test/test.c b/3rdparty/zstd/contrib/linux-kernel/test/test.c new file mode 100644 index 00000000000..0f4ba3f45ef --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/test/test.c @@ -0,0 +1,229 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ +#include <stddef.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> + +#include <linux/zstd.h> + +#define CONTROL(x) \ + do { \ + if (!(x)) { \ + fprintf(stderr, "%s:%u: %s failed!\n", __FUNCTION__, __LINE__, #x); \ + abort(); \ + } \ + } while (0) + +typedef struct { + char *data; + char *data2; + size_t dataSize; + char *comp; + size_t compSize; +} test_data_t; + +static test_data_t create_test_data(void) { + test_data_t data; + data.dataSize = 128 * 1024; + data.data = (char*)malloc(data.dataSize); + CONTROL(data.data != NULL); + data.data2 = (char*)malloc(data.dataSize); + CONTROL(data.data2 != NULL); + data.compSize = zstd_compress_bound(data.dataSize); + data.comp = (char*)malloc(data.compSize); + CONTROL(data.comp != NULL); + memset(data.data, 0, data.dataSize); + return data; +} + +static void free_test_data(test_data_t const *data) { + free(data->data); + free(data->data2); + free(data->comp); +} + +#define MIN(a, b) ((a) < (b) ? (a) : (b)) +#define MAX(a, b) ((a) > (b) ? (a) : (b)) + +static void test_btrfs(test_data_t const *data) { + size_t const size = MIN(data->dataSize, 128 * 1024); + fprintf(stderr, "testing btrfs use cases... "); + for (int level = -1; level < 16; ++level) { + zstd_parameters params = zstd_get_params(level, size); + size_t const workspaceSize = + MAX(zstd_cstream_workspace_bound(¶ms.cParams), + zstd_dstream_workspace_bound(size)); + void *workspace = malloc(workspaceSize); + + char const *ip = data->data; + char const *iend = ip + size; + char *op = data->comp; + char *oend = op + data->compSize; + + CONTROL(params.cParams.windowLog <= 17); + CONTROL(workspace != NULL); + { + zstd_cstream *cctx = zstd_init_cstream(¶ms, size, workspace, workspaceSize); + zstd_out_buffer out = {NULL, 0, 0}; + zstd_in_buffer in = {NULL, 0, 0}; + CONTROL(cctx != NULL); + for (;;) { + if (in.pos == in.size) { + in.src = ip; + in.size = MIN(4096, iend - ip); + in.pos = 0; + ip += in.size; + } + + if (out.pos == out.size) { + out.dst = op; + out.size = MIN(4096, oend - op); + out.pos = 0; + op += out.size; + } + + if (ip != iend || in.pos < in.size) { + CONTROL(!zstd_is_error(zstd_compress_stream(cctx, &out, &in))); + } else { + size_t const ret = zstd_end_stream(cctx, &out); + CONTROL(!zstd_is_error(ret)); + if (ret == 0) { + break; + } + } + } + op += out.pos; + } + + ip = data->comp; + iend = op; + op = data->data2; + oend = op + size; + { + zstd_dstream *dctx = zstd_init_dstream(1ULL << params.cParams.windowLog, workspace, workspaceSize); + zstd_out_buffer out = {NULL, 0, 0}; + zstd_in_buffer in = {NULL, 0, 0}; + CONTROL(dctx != NULL); + for (;;) { + if (in.pos == in.size) { + in.src = ip; + in.size = MIN(4096, iend - ip); + in.pos = 0; + ip += in.size; + } + + if (out.pos == out.size) { + out.dst = op; + out.size = MIN(4096, oend - op); + out.pos = 0; + op += out.size; + } + { + size_t const ret = zstd_decompress_stream(dctx, &out, &in); + CONTROL(!zstd_is_error(ret)); + if (ret == 0) { + break; + } + } + } + } + CONTROL((size_t)(op - data->data2) == data->dataSize); + CONTROL(!memcmp(data->data, data->data2, data->dataSize)); + free(workspace); + } + fprintf(stderr, "Ok\n"); +} + +static void test_decompress_unzstd(test_data_t const *data) { + size_t cSize; + fprintf(stderr, "Testing decompress unzstd... "); + { + zstd_parameters params = zstd_get_params(19, 0); + size_t const wkspSize = zstd_cctx_workspace_bound(¶ms.cParams); + void* wksp = malloc(wkspSize); + zstd_cctx* cctx = zstd_init_cctx(wksp, wkspSize); + CONTROL(wksp != NULL); + CONTROL(cctx != NULL); + cSize = zstd_compress_cctx(cctx, data->comp, data->compSize, data->data, data->dataSize, ¶ms); + CONTROL(!zstd_is_error(cSize)); + free(wksp); + } + { + size_t const wkspSize = zstd_dctx_workspace_bound(); + void* wksp = malloc(wkspSize); + zstd_dctx* dctx = zstd_init_dctx(wksp, wkspSize); + CONTROL(wksp != NULL); + CONTROL(dctx != NULL); + { + size_t const dSize = zstd_decompress_dctx(dctx, data->data2, data->dataSize, data->comp, cSize); + CONTROL(!zstd_is_error(dSize)); + CONTROL(dSize == data->dataSize); + } + CONTROL(!memcmp(data->data, data->data2, data->dataSize)); + free(wksp); + } + fprintf(stderr, "Ok\n"); +} + +static void test_f2fs(void) { + fprintf(stderr, "testing f2fs uses... "); + CONTROL(zstd_min_clevel() < 0); + CONTROL(zstd_max_clevel() == 22); + fprintf(stderr, "Ok\n"); +} + +static char *g_stack = NULL; + +static void __attribute__((noinline)) use(void *x) { + asm volatile("" : "+r"(x)); +} + +static void __attribute__((noinline)) fill_stack(void) { + memset(g_stack, 0x33, 8192); +} + +static void __attribute__((noinline)) set_stack(void) { + + char stack[8192]; + g_stack = stack; + use(g_stack); +} + +static void __attribute__((noinline)) check_stack(void) { + size_t cleanStack = 0; + while (cleanStack < 8192 && g_stack[cleanStack] == 0x33) { + ++cleanStack; + } + { + size_t const stackSize = 8192 - cleanStack; + fprintf(stderr, "Maximum stack size: %zu\n", stackSize); + CONTROL(stackSize <= 2048 + 512); + } +} + +static void test_stack_usage(test_data_t const *data) { + set_stack(); + fill_stack(); + test_f2fs(); + test_btrfs(data); + test_decompress_unzstd(data); + check_stack(); +} + +int main(void) { + test_data_t data = create_test_data(); + test_f2fs(); + test_btrfs(&data); + test_decompress_unzstd(&data); + test_stack_usage(&data); + free_test_data(&data); + return 0; +} diff --git a/3rdparty/zstd/contrib/linux-kernel/zstd_common_module.c b/3rdparty/zstd/contrib/linux-kernel/zstd_common_module.c new file mode 100644 index 00000000000..466828e3575 --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/zstd_common_module.c @@ -0,0 +1,29 @@ +// SPDX-License-Identifier: GPL-2.0+ OR BSD-3-Clause +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ + +#include <linux/module.h> + +#include "common/huf.h" +#include "common/fse.h" +#include "common/zstd_internal.h" + +// Export symbols shared by compress and decompress into a common module + +#undef ZSTD_isError /* defined within zstd_internal.h */ +EXPORT_SYMBOL_GPL(FSE_readNCount); +EXPORT_SYMBOL_GPL(HUF_readStats); +EXPORT_SYMBOL_GPL(HUF_readStats_wksp); +EXPORT_SYMBOL_GPL(ZSTD_isError); +EXPORT_SYMBOL_GPL(ZSTD_getErrorName); +EXPORT_SYMBOL_GPL(ZSTD_getErrorCode); + +MODULE_LICENSE("Dual BSD/GPL"); +MODULE_DESCRIPTION("Zstd Common"); diff --git a/3rdparty/zstd/contrib/linux-kernel/zstd_compress_module.c b/3rdparty/zstd/contrib/linux-kernel/zstd_compress_module.c new file mode 100644 index 00000000000..8ecf43226af --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/zstd_compress_module.c @@ -0,0 +1,164 @@ +// SPDX-License-Identifier: GPL-2.0+ OR BSD-3-Clause +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ + +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/string.h> +#include <linux/zstd.h> + +#include "common/zstd_deps.h" +#include "common/zstd_internal.h" + +#define ZSTD_FORWARD_IF_ERR(ret) \ + do { \ + size_t const __ret = (ret); \ + if (ZSTD_isError(__ret)) \ + return __ret; \ + } while (0) + +static size_t zstd_cctx_init(zstd_cctx *cctx, const zstd_parameters *parameters, + unsigned long long pledged_src_size) +{ + ZSTD_FORWARD_IF_ERR(ZSTD_CCtx_reset( + cctx, ZSTD_reset_session_and_parameters)); + ZSTD_FORWARD_IF_ERR(ZSTD_CCtx_setPledgedSrcSize( + cctx, pledged_src_size)); + ZSTD_FORWARD_IF_ERR(ZSTD_CCtx_setParameter( + cctx, ZSTD_c_windowLog, parameters->cParams.windowLog)); + ZSTD_FORWARD_IF_ERR(ZSTD_CCtx_setParameter( + cctx, ZSTD_c_hashLog, parameters->cParams.hashLog)); + ZSTD_FORWARD_IF_ERR(ZSTD_CCtx_setParameter( + cctx, ZSTD_c_chainLog, parameters->cParams.chainLog)); + ZSTD_FORWARD_IF_ERR(ZSTD_CCtx_setParameter( + cctx, ZSTD_c_searchLog, parameters->cParams.searchLog)); + ZSTD_FORWARD_IF_ERR(ZSTD_CCtx_setParameter( + cctx, ZSTD_c_minMatch, parameters->cParams.minMatch)); + ZSTD_FORWARD_IF_ERR(ZSTD_CCtx_setParameter( + cctx, ZSTD_c_targetLength, parameters->cParams.targetLength)); + ZSTD_FORWARD_IF_ERR(ZSTD_CCtx_setParameter( + cctx, ZSTD_c_strategy, parameters->cParams.strategy)); + ZSTD_FORWARD_IF_ERR(ZSTD_CCtx_setParameter( + cctx, ZSTD_c_contentSizeFlag, parameters->fParams.contentSizeFlag)); + ZSTD_FORWARD_IF_ERR(ZSTD_CCtx_setParameter( + cctx, ZSTD_c_checksumFlag, parameters->fParams.checksumFlag)); + ZSTD_FORWARD_IF_ERR(ZSTD_CCtx_setParameter( + cctx, ZSTD_c_dictIDFlag, !parameters->fParams.noDictIDFlag)); + return 0; +} + +int zstd_min_clevel(void) +{ + return ZSTD_minCLevel(); +} +EXPORT_SYMBOL(zstd_min_clevel); + +int zstd_max_clevel(void) +{ + return ZSTD_maxCLevel(); +} +EXPORT_SYMBOL(zstd_max_clevel); + +size_t zstd_compress_bound(size_t src_size) +{ + return ZSTD_compressBound(src_size); +} +EXPORT_SYMBOL(zstd_compress_bound); + +zstd_parameters zstd_get_params(int level, + unsigned long long estimated_src_size) +{ + return ZSTD_getParams(level, estimated_src_size, 0); +} +EXPORT_SYMBOL(zstd_get_params); + +size_t zstd_cctx_workspace_bound(const zstd_compression_parameters *cparams) +{ + return ZSTD_estimateCCtxSize_usingCParams(*cparams); +} +EXPORT_SYMBOL(zstd_cctx_workspace_bound); + +zstd_cctx *zstd_init_cctx(void *workspace, size_t workspace_size) +{ + if (workspace == NULL) + return NULL; + return ZSTD_initStaticCCtx(workspace, workspace_size); +} +EXPORT_SYMBOL(zstd_init_cctx); + +size_t zstd_compress_cctx(zstd_cctx *cctx, void *dst, size_t dst_capacity, + const void *src, size_t src_size, const zstd_parameters *parameters) +{ + ZSTD_FORWARD_IF_ERR(zstd_cctx_init(cctx, parameters, src_size)); + return ZSTD_compress2(cctx, dst, dst_capacity, src, src_size); +} +EXPORT_SYMBOL(zstd_compress_cctx); + +size_t zstd_cstream_workspace_bound(const zstd_compression_parameters *cparams) +{ + return ZSTD_estimateCStreamSize_usingCParams(*cparams); +} +EXPORT_SYMBOL(zstd_cstream_workspace_bound); + +zstd_cstream *zstd_init_cstream(const zstd_parameters *parameters, + unsigned long long pledged_src_size, void *workspace, size_t workspace_size) +{ + zstd_cstream *cstream; + + if (workspace == NULL) + return NULL; + + cstream = ZSTD_initStaticCStream(workspace, workspace_size); + if (cstream == NULL) + return NULL; + + /* 0 means unknown in linux zstd API but means 0 in new zstd API */ + if (pledged_src_size == 0) + pledged_src_size = ZSTD_CONTENTSIZE_UNKNOWN; + + if (ZSTD_isError(zstd_cctx_init(cstream, parameters, pledged_src_size))) + return NULL; + + return cstream; +} +EXPORT_SYMBOL(zstd_init_cstream); + +size_t zstd_reset_cstream(zstd_cstream *cstream, + unsigned long long pledged_src_size) +{ + if (pledged_src_size == 0) + pledged_src_size = ZSTD_CONTENTSIZE_UNKNOWN; + ZSTD_FORWARD_IF_ERR( ZSTD_CCtx_reset(cstream, ZSTD_reset_session_only) ); + ZSTD_FORWARD_IF_ERR( ZSTD_CCtx_setPledgedSrcSize(cstream, pledged_src_size) ); + return 0; +} +EXPORT_SYMBOL(zstd_reset_cstream); + +size_t zstd_compress_stream(zstd_cstream *cstream, zstd_out_buffer *output, + zstd_in_buffer *input) +{ + return ZSTD_compressStream(cstream, output, input); +} +EXPORT_SYMBOL(zstd_compress_stream); + +size_t zstd_flush_stream(zstd_cstream *cstream, zstd_out_buffer *output) +{ + return ZSTD_flushStream(cstream, output); +} +EXPORT_SYMBOL(zstd_flush_stream); + +size_t zstd_end_stream(zstd_cstream *cstream, zstd_out_buffer *output) +{ + return ZSTD_endStream(cstream, output); +} +EXPORT_SYMBOL(zstd_end_stream); + +MODULE_LICENSE("Dual BSD/GPL"); +MODULE_DESCRIPTION("Zstd Compressor"); diff --git a/3rdparty/zstd/contrib/linux-kernel/zstd_decompress_module.c b/3rdparty/zstd/contrib/linux-kernel/zstd_decompress_module.c new file mode 100644 index 00000000000..eb1c49e6972 --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/zstd_decompress_module.c @@ -0,0 +1,105 @@ +// SPDX-License-Identifier: GPL-2.0+ OR BSD-3-Clause +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ + +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/string.h> +#include <linux/zstd.h> + +#include "common/zstd_deps.h" + +/* Common symbols. zstd_compress must depend on zstd_decompress. */ + +unsigned int zstd_is_error(size_t code) +{ + return ZSTD_isError(code); +} +EXPORT_SYMBOL(zstd_is_error); + +zstd_error_code zstd_get_error_code(size_t code) +{ + return ZSTD_getErrorCode(code); +} +EXPORT_SYMBOL(zstd_get_error_code); + +const char *zstd_get_error_name(size_t code) +{ + return ZSTD_getErrorName(code); +} +EXPORT_SYMBOL(zstd_get_error_name); + +/* Decompression symbols. */ + +size_t zstd_dctx_workspace_bound(void) +{ + return ZSTD_estimateDCtxSize(); +} +EXPORT_SYMBOL(zstd_dctx_workspace_bound); + +zstd_dctx *zstd_init_dctx(void *workspace, size_t workspace_size) +{ + if (workspace == NULL) + return NULL; + return ZSTD_initStaticDCtx(workspace, workspace_size); +} +EXPORT_SYMBOL(zstd_init_dctx); + +size_t zstd_decompress_dctx(zstd_dctx *dctx, void *dst, size_t dst_capacity, + const void *src, size_t src_size) +{ + return ZSTD_decompressDCtx(dctx, dst, dst_capacity, src, src_size); +} +EXPORT_SYMBOL(zstd_decompress_dctx); + +size_t zstd_dstream_workspace_bound(size_t max_window_size) +{ + return ZSTD_estimateDStreamSize(max_window_size); +} +EXPORT_SYMBOL(zstd_dstream_workspace_bound); + +zstd_dstream *zstd_init_dstream(size_t max_window_size, void *workspace, + size_t workspace_size) +{ + if (workspace == NULL) + return NULL; + (void)max_window_size; + return ZSTD_initStaticDStream(workspace, workspace_size); +} +EXPORT_SYMBOL(zstd_init_dstream); + +size_t zstd_reset_dstream(zstd_dstream *dstream) +{ + return ZSTD_resetDStream(dstream); +} +EXPORT_SYMBOL(zstd_reset_dstream); + +size_t zstd_decompress_stream(zstd_dstream *dstream, zstd_out_buffer *output, + zstd_in_buffer *input) +{ + return ZSTD_decompressStream(dstream, output, input); +} +EXPORT_SYMBOL(zstd_decompress_stream); + +size_t zstd_find_frame_compressed_size(const void *src, size_t src_size) +{ + return ZSTD_findFrameCompressedSize(src, src_size); +} +EXPORT_SYMBOL(zstd_find_frame_compressed_size); + +size_t zstd_get_frame_header(zstd_frame_header *header, const void *src, + size_t src_size) +{ + return ZSTD_getFrameHeader(header, src, src_size); +} +EXPORT_SYMBOL(zstd_get_frame_header); + +MODULE_LICENSE("Dual BSD/GPL"); +MODULE_DESCRIPTION("Zstd Decompressor"); diff --git a/3rdparty/zstd/contrib/linux-kernel/zstd_deps.h b/3rdparty/zstd/contrib/linux-kernel/zstd_deps.h new file mode 100644 index 00000000000..670c5fa2a95 --- /dev/null +++ b/3rdparty/zstd/contrib/linux-kernel/zstd_deps.h @@ -0,0 +1,125 @@ +/* SPDX-License-Identifier: GPL-2.0+ OR BSD-3-Clause */ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ + +/* + * This file provides common libc dependencies that zstd requires. + * The purpose is to allow replacing this file with a custom implementation + * to compile zstd without libc support. + */ + +/* Need: + * NULL + * INT_MAX + * UINT_MAX + * ZSTD_memcpy() + * ZSTD_memset() + * ZSTD_memmove() + */ +#ifndef ZSTD_DEPS_COMMON +#define ZSTD_DEPS_COMMON + +#include <linux/limits.h> +#include <linux/stddef.h> + +#define ZSTD_memcpy(d,s,n) __builtin_memcpy((d),(s),(n)) +#define ZSTD_memmove(d,s,n) __builtin_memmove((d),(s),(n)) +#define ZSTD_memset(d,s,n) __builtin_memset((d),(s),(n)) + +#endif /* ZSTD_DEPS_COMMON */ + +/* + * Define malloc as always failing. That means the user must + * either use ZSTD_customMem or statically allocate memory. + * Need: + * ZSTD_malloc() + * ZSTD_free() + * ZSTD_calloc() + */ +#ifdef ZSTD_DEPS_NEED_MALLOC +#ifndef ZSTD_DEPS_MALLOC +#define ZSTD_DEPS_MALLOC + +#define ZSTD_malloc(s) ({ (void)(s); NULL; }) +#define ZSTD_free(p) ((void)(p)) +#define ZSTD_calloc(n,s) ({ (void)(n); (void)(s); NULL; }) + +#endif /* ZSTD_DEPS_MALLOC */ +#endif /* ZSTD_DEPS_NEED_MALLOC */ + +/* + * Provides 64-bit math support. + * Need: + * U64 ZSTD_div64(U64 dividend, U32 divisor) + */ +#ifdef ZSTD_DEPS_NEED_MATH64 +#ifndef ZSTD_DEPS_MATH64 +#define ZSTD_DEPS_MATH64 + +#include <linux/math64.h> + +static uint64_t ZSTD_div64(uint64_t dividend, uint32_t divisor) { + return div_u64(dividend, divisor); +} + +#endif /* ZSTD_DEPS_MATH64 */ +#endif /* ZSTD_DEPS_NEED_MATH64 */ + +/* + * This is only requested when DEBUGLEVEL >= 1, meaning + * it is disabled in production. + * Need: + * assert() + */ +#ifdef ZSTD_DEPS_NEED_ASSERT +#ifndef ZSTD_DEPS_ASSERT +#define ZSTD_DEPS_ASSERT + +#include <linux/kernel.h> + +#define assert(x) WARN_ON(!(x)) + +#endif /* ZSTD_DEPS_ASSERT */ +#endif /* ZSTD_DEPS_NEED_ASSERT */ + +/* + * This is only requested when DEBUGLEVEL >= 2, meaning + * it is disabled in production. + * Need: + * ZSTD_DEBUG_PRINT() + */ +#ifdef ZSTD_DEPS_NEED_IO +#ifndef ZSTD_DEPS_IO +#define ZSTD_DEPS_IO + +#include <linux/printk.h> + +#define ZSTD_DEBUG_PRINT(...) pr_debug(__VA_ARGS__) + +#endif /* ZSTD_DEPS_IO */ +#endif /* ZSTD_DEPS_NEED_IO */ + +/* + * Only requested when MSAN is enabled. + * Need: + * intptr_t + */ +#ifdef ZSTD_DEPS_NEED_STDINT +#ifndef ZSTD_DEPS_STDINT +#define ZSTD_DEPS_STDINT + +/* + * The Linux Kernel doesn't provide intptr_t, only uintptr_t, which + * is an unsigned long. + */ +typedef long intptr_t; + +#endif /* ZSTD_DEPS_STDINT */ +#endif /* ZSTD_DEPS_NEED_STDINT */ diff --git a/3rdparty/zstd/contrib/match_finders/README.md b/3rdparty/zstd/contrib/match_finders/README.md new file mode 100644 index 00000000000..54055c37086 --- /dev/null +++ b/3rdparty/zstd/contrib/match_finders/README.md @@ -0,0 +1,42 @@ +## Edit Distance Match Finder + +``` +/* This match finder leverages techniques used in file comparison algorithms + * to find matches between a dictionary and a source file. + * + * The original motivation for studying this approach was to try and optimize + * Zstandard for the use case of patching: the most common scenario being + * updating an existing software package with the next version. When patching, + * the difference between the old version of the package and the new version + * is generally tiny (most of the new file will be identical to + * the old one). In more technical terms, the edit distance (the minimal number + * of changes required to take one sequence of bytes to another) between the + * files would be small relative to the size of the file. + * + * Various 'diffing' algorithms utilize this notion of edit distance and + * the corresponding concept of a minimal edit script between two + * sequences to identify the regions within two files where they differ. + * The core algorithm used in this match finder is described in: + * + * "An O(ND) Difference Algorithm and its Variations", Eugene W. Myers, + * Algorithmica Vol. 1, 1986, pp. 251-266, + * <https://doi.org/10.1007/BF01840446>. + * + * Additional algorithmic heuristics for speed improvement have also been included. + * These we inspired from implementations of various regular and binary diffing + * algorithms such as GNU diff, bsdiff, and Xdelta. + * + * Note: after some experimentation, this approach proved to not provide enough + * utility to justify the additional CPU used in finding matches. The one area + * where this approach consistently outperforms Zstandard even on level 19 is + * when compressing small files (<10 KB) using an equally small dictionary that + * is very similar to the source file. For the use case that this was intended, + * (large similar files) this approach by itself took 5-10X longer than zstd-19 and + * generally resulted in 2-3X larger files. The core advantage that zstd-19 has + * over this approach for match finding is the overlapping matches. This approach + * cannot find any. + * + * I'm leaving this in the contrib section in case this ever becomes interesting + * to explore again. + * */ +``` diff --git a/3rdparty/zstd/contrib/match_finders/zstd_edist.c b/3rdparty/zstd/contrib/match_finders/zstd_edist.c new file mode 100644 index 00000000000..d685cdd9e2f --- /dev/null +++ b/3rdparty/zstd/contrib/match_finders/zstd_edist.c @@ -0,0 +1,558 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ + +/*-************************************* +* Dependencies +***************************************/ + +/* Currently relies on qsort when combining contiguous matches. This can probably + * be avoided but would require changes to the algorithm. The qsort is far from + * the bottleneck in this algorithm even for medium sized files so it's probably + * not worth trying to address */ +#include <stdlib.h> +#include <assert.h> + +#include "zstd_edist.h" +#include "mem.h" + +/*-************************************* +* Constants +***************************************/ + +/* Just a sential for the entries of the diagonal matrix */ +#define ZSTD_EDIST_DIAG_MAX (S32)(1 << 30) + +/* How large should a snake be to be considered a 'big' snake. + * For an explanation of what a 'snake' is with respect to the + * edit distance matrix, see the linked paper in zstd_edist.h */ +#define ZSTD_EDIST_SNAKE_THRESH 20 + +/* After how many iterations should we start to use the heuristic + * based on 'big' snakes */ +#define ZSTD_EDIST_SNAKE_ITER_THRESH 200 + +/* After how many iterations should be just give up and take + * the best available edit script for this round */ +#define ZSTD_EDIST_EXPENSIVE_THRESH 1024 + +/*-************************************* +* Structures +***************************************/ + +typedef struct { + U32 dictIdx; + U32 srcIdx; + U32 matchLength; +} ZSTD_eDist_match; + +typedef struct { + const BYTE* dict; + const BYTE* src; + size_t dictSize; + size_t srcSize; + S32* forwardDiag; /* Entries of the forward diagonal stored here */ + S32* backwardDiag; /* Entries of the backward diagonal stored here. + * Note: this buffer and the 'forwardDiag' buffer + * are contiguous. See the ZSTD_eDist_genSequences */ + ZSTD_eDist_match* matches; /* Accumulate matches of length 1 in this buffer. + * In a subsequence post-processing step, we combine + * contiguous matches. */ + U32 nbMatches; +} ZSTD_eDist_state; + +typedef struct { + S32 dictMid; /* The mid diagonal for the dictionary */ + S32 srcMid; /* The mid diagonal for the source */ + int lowUseHeuristics; /* Should we use heuristics for the low part */ + int highUseHeuristics; /* Should we use heuristics for the high part */ +} ZSTD_eDist_partition; + +/*-************************************* +* Internal +***************************************/ + +static void ZSTD_eDist_diag(ZSTD_eDist_state* state, + ZSTD_eDist_partition* partition, + S32 dictLow, S32 dictHigh, S32 srcLow, + S32 srcHigh, int useHeuristics) +{ + S32* const forwardDiag = state->forwardDiag; + S32* const backwardDiag = state->backwardDiag; + const BYTE* const dict = state->dict; + const BYTE* const src = state->src; + + S32 const diagMin = dictLow - srcHigh; + S32 const diagMax = dictHigh - srcLow; + S32 const forwardMid = dictLow - srcLow; + S32 const backwardMid = dictHigh - srcHigh; + + S32 forwardMin = forwardMid; + S32 forwardMax = forwardMid; + S32 backwardMin = backwardMid; + S32 backwardMax = backwardMid; + int odd = (forwardMid - backwardMid) & 1; + U32 iterations; + + forwardDiag[forwardMid] = dictLow; + backwardDiag[backwardMid] = dictHigh; + + /* Main loop for updating diag entries. Unless useHeuristics is + * set to false, this loop will run until it finds the minimal + * edit script */ + for (iterations = 1;;iterations++) { + S32 diag; + int bigSnake = 0; + + if (forwardMin > diagMin) { + forwardMin--; + forwardDiag[forwardMin - 1] = -1; + } else { + forwardMin++; + } + + if (forwardMax < diagMax) { + forwardMax++; + forwardDiag[forwardMax + 1] = -1; + } else { + forwardMax--; + } + + for (diag = forwardMax; diag >= forwardMin; diag -= 2) { + S32 dictIdx; + S32 srcIdx; + S32 low = forwardDiag[diag - 1]; + S32 high = forwardDiag[diag + 1]; + S32 dictIdx0 = low < high ? high : low + 1; + + for (dictIdx = dictIdx0, srcIdx = dictIdx0 - diag; + dictIdx < dictHigh && srcIdx < srcHigh && dict[dictIdx] == src[srcIdx]; + dictIdx++, srcIdx++) continue; + + if (dictIdx - dictIdx0 > ZSTD_EDIST_SNAKE_THRESH) + bigSnake = 1; + + forwardDiag[diag] = dictIdx; + + if (odd && backwardMin <= diag && diag <= backwardMax && backwardDiag[diag] <= dictIdx) { + partition->dictMid = dictIdx; + partition->srcMid = srcIdx; + partition->lowUseHeuristics = 0; + partition->highUseHeuristics = 0; + return; + } + } + + if (backwardMin > diagMin) { + backwardMin--; + backwardDiag[backwardMin - 1] = ZSTD_EDIST_DIAG_MAX; + } else { + backwardMin++; + } + + if (backwardMax < diagMax) { + backwardMax++; + backwardDiag[backwardMax + 1] = ZSTD_EDIST_DIAG_MAX; + } else { + backwardMax--; + } + + + for (diag = backwardMax; diag >= backwardMin; diag -= 2) { + S32 dictIdx; + S32 srcIdx; + S32 low = backwardDiag[diag - 1]; + S32 high = backwardDiag[diag + 1]; + S32 dictIdx0 = low < high ? low : high - 1; + + for (dictIdx = dictIdx0, srcIdx = dictIdx0 - diag; + dictLow < dictIdx && srcLow < srcIdx && dict[dictIdx - 1] == src[srcIdx - 1]; + dictIdx--, srcIdx--) continue; + + if (dictIdx0 - dictIdx > ZSTD_EDIST_SNAKE_THRESH) + bigSnake = 1; + + backwardDiag[diag] = dictIdx; + + if (!odd && forwardMin <= diag && diag <= forwardMax && dictIdx <= forwardDiag[diag]) { + partition->dictMid = dictIdx; + partition->srcMid = srcIdx; + partition->lowUseHeuristics = 0; + partition->highUseHeuristics = 0; + return; + } + } + + if (!useHeuristics) + continue; + + /* Everything under this point is a heuristic. Using these will + * substantially speed up the match finding. In some cases, taking + * the total match finding time from several minutes to seconds. + * Of course, the caveat is that the edit script found may no longer + * be optimal */ + + /* Big snake heuristic */ + if (iterations > ZSTD_EDIST_SNAKE_ITER_THRESH && bigSnake) { + { + S32 best = 0; + + for (diag = forwardMax; diag >= forwardMin; diag -= 2) { + S32 diagDiag = diag - forwardMid; + S32 dictIdx = forwardDiag[diag]; + S32 srcIdx = dictIdx - diag; + S32 v = (dictIdx - dictLow) * 2 - diagDiag; + + if (v > 12 * (iterations + (diagDiag < 0 ? -diagDiag : diagDiag))) { + if (v > best + && dictLow + ZSTD_EDIST_SNAKE_THRESH <= dictIdx && dictIdx <= dictHigh + && srcLow + ZSTD_EDIST_SNAKE_THRESH <= srcIdx && srcIdx <= srcHigh) { + S32 k; + for (k = 1; dict[dictIdx - k] == src[srcIdx - k]; k++) { + if (k == ZSTD_EDIST_SNAKE_THRESH) { + best = v; + partition->dictMid = dictIdx; + partition->srcMid = srcIdx; + break; + } + } + } + } + } + + if (best > 0) { + partition->lowUseHeuristics = 0; + partition->highUseHeuristics = 1; + return; + } + } + + { + S32 best = 0; + + for (diag = backwardMax; diag >= backwardMin; diag -= 2) { + S32 diagDiag = diag - backwardMid; + S32 dictIdx = backwardDiag[diag]; + S32 srcIdx = dictIdx - diag; + S32 v = (dictHigh - dictIdx) * 2 + diagDiag; + + if (v > 12 * (iterations + (diagDiag < 0 ? -diagDiag : diagDiag))) { + if (v > best + && dictLow < dictIdx && dictIdx <= dictHigh - ZSTD_EDIST_SNAKE_THRESH + && srcLow < srcIdx && srcIdx <= srcHigh - ZSTD_EDIST_SNAKE_THRESH) { + int k; + for (k = 0; dict[dictIdx + k] == src[srcIdx + k]; k++) { + if (k == ZSTD_EDIST_SNAKE_THRESH - 1) { + best = v; + partition->dictMid = dictIdx; + partition->srcMid = srcIdx; + break; + } + } + } + } + } + + if (best > 0) { + partition->lowUseHeuristics = 1; + partition->highUseHeuristics = 0; + return; + } + } + } + + /* More general 'too expensive' heuristic */ + if (iterations >= ZSTD_EDIST_EXPENSIVE_THRESH) { + S32 forwardDictSrcBest; + S32 forwardDictBest = 0; + S32 backwardDictSrcBest; + S32 backwardDictBest = 0; + + forwardDictSrcBest = -1; + for (diag = forwardMax; diag >= forwardMin; diag -= 2) { + S32 dictIdx = MIN(forwardDiag[diag], dictHigh); + S32 srcIdx = dictIdx - diag; + + if (srcHigh < srcIdx) { + dictIdx = srcHigh + diag; + srcIdx = srcHigh; + } + + if (forwardDictSrcBest < dictIdx + srcIdx) { + forwardDictSrcBest = dictIdx + srcIdx; + forwardDictBest = dictIdx; + } + } + + backwardDictSrcBest = ZSTD_EDIST_DIAG_MAX; + for (diag = backwardMax; diag >= backwardMin; diag -= 2) { + S32 dictIdx = MAX(dictLow, backwardDiag[diag]); + S32 srcIdx = dictIdx - diag; + + if (srcIdx < srcLow) { + dictIdx = srcLow + diag; + srcIdx = srcLow; + } + + if (dictIdx + srcIdx < backwardDictSrcBest) { + backwardDictSrcBest = dictIdx + srcIdx; + backwardDictBest = dictIdx; + } + } + + if ((dictHigh + srcHigh) - backwardDictSrcBest < forwardDictSrcBest - (dictLow + srcLow)) { + partition->dictMid = forwardDictBest; + partition->srcMid = forwardDictSrcBest - forwardDictBest; + partition->lowUseHeuristics = 0; + partition->highUseHeuristics = 1; + } else { + partition->dictMid = backwardDictBest; + partition->srcMid = backwardDictSrcBest - backwardDictBest; + partition->lowUseHeuristics = 1; + partition->highUseHeuristics = 0; + } + return; + } + } +} + +static void ZSTD_eDist_insertMatch(ZSTD_eDist_state* state, + S32 const dictIdx, S32 const srcIdx) +{ + state->matches[state->nbMatches].dictIdx = dictIdx; + state->matches[state->nbMatches].srcIdx = srcIdx; + state->matches[state->nbMatches].matchLength = 1; + state->nbMatches++; +} + +static int ZSTD_eDist_compare(ZSTD_eDist_state* state, + S32 dictLow, S32 dictHigh, S32 srcLow, + S32 srcHigh, int useHeuristics) +{ + const BYTE* const dict = state->dict; + const BYTE* const src = state->src; + + /* Found matches while traversing from the low end */ + while (dictLow < dictHigh && srcLow < srcHigh && dict[dictLow] == src[srcLow]) { + ZSTD_eDist_insertMatch(state, dictLow, srcLow); + dictLow++; + srcLow++; + } + + /* Found matches while traversing from the high end */ + while (dictLow < dictHigh && srcLow < srcHigh && dict[dictHigh - 1] == src[srcHigh - 1]) { + ZSTD_eDist_insertMatch(state, dictHigh - 1, srcHigh - 1); + dictHigh--; + srcHigh--; + } + + /* If the low and high end end up touching. If we wanted to make + * note of the differences like most diffing algorithms do, we would + * do so here. In our case, we're only concerned with matches + * Note: if you wanted to find the edit distance of the algorithm, + * you could just accumulate the cost for an insertion/deletion + * below. */ + if (dictLow == dictHigh) { + while (srcLow < srcHigh) { + /* Reaching this point means inserting src[srcLow] into + * the current position of dict */ + srcLow++; + } + } else if (srcLow == srcHigh) { + while (dictLow < dictHigh) { + /* Reaching this point means deleting dict[dictLow] from + * the current position of dict */ + dictLow++; + } + } else { + ZSTD_eDist_partition partition; + partition.dictMid = 0; + partition.srcMid = 0; + ZSTD_eDist_diag(state, &partition, dictLow, dictHigh, + srcLow, srcHigh, useHeuristics); + if (ZSTD_eDist_compare(state, dictLow, partition.dictMid, + srcLow, partition.srcMid, partition.lowUseHeuristics)) + return 1; + if (ZSTD_eDist_compare(state, partition.dictMid, dictHigh, + partition.srcMid, srcHigh, partition.highUseHeuristics)) + return 1; + } + + return 0; +} + +static int ZSTD_eDist_matchComp(const void* p, const void* q) +{ + S32 const l = ((ZSTD_eDist_match*)p)->srcIdx; + S32 const r = ((ZSTD_eDist_match*)q)->srcIdx; + return (l - r); +} + +/* The matches from the approach above will all be of the form + * (dictIdx, srcIdx, 1). This method combines contiguous matches + * of length MINMATCH or greater. Matches less than MINMATCH + * are discarded */ +static void ZSTD_eDist_combineMatches(ZSTD_eDist_state* state) +{ + /* Create a new buffer to put the combined matches into + * and memcpy to state->matches after */ + ZSTD_eDist_match* combinedMatches = + ZSTD_malloc(state->nbMatches * sizeof(ZSTD_eDist_match), + ZSTD_defaultCMem); + + U32 nbCombinedMatches = 1; + size_t i; + + /* Make sure that the srcIdx and dictIdx are in sorted order. + * The combination step won't work otherwise */ + qsort(state->matches, state->nbMatches, sizeof(ZSTD_eDist_match), ZSTD_eDist_matchComp); + + memcpy(combinedMatches, state->matches, sizeof(ZSTD_eDist_match)); + for (i = 1; i < state->nbMatches; i++) { + ZSTD_eDist_match const match = state->matches[i]; + ZSTD_eDist_match const combinedMatch = + combinedMatches[nbCombinedMatches - 1]; + if (combinedMatch.srcIdx + combinedMatch.matchLength == match.srcIdx && + combinedMatch.dictIdx + combinedMatch.matchLength == match.dictIdx) { + combinedMatches[nbCombinedMatches - 1].matchLength++; + } else { + /* Discard matches that are less than MINMATCH */ + if (combinedMatches[nbCombinedMatches - 1].matchLength < MINMATCH) { + nbCombinedMatches--; + } + + memcpy(combinedMatches + nbCombinedMatches, + state->matches + i, sizeof(ZSTD_eDist_match)); + nbCombinedMatches++; + } + } + memcpy(state->matches, combinedMatches, nbCombinedMatches * sizeof(ZSTD_eDist_match)); + state->nbMatches = nbCombinedMatches; + ZSTD_free(combinedMatches, ZSTD_defaultCMem); +} + +static size_t ZSTD_eDist_convertMatchesToSequences(ZSTD_Sequence* sequences, + ZSTD_eDist_state* state) +{ + const ZSTD_eDist_match* matches = state->matches; + size_t const nbMatches = state->nbMatches; + size_t const dictSize = state->dictSize; + size_t nbSequences = 0; + size_t i; + for (i = 0; i < nbMatches; i++) { + ZSTD_eDist_match const match = matches[i]; + U32 const litLength = !i ? match.srcIdx : + match.srcIdx - (matches[i - 1].srcIdx + matches[i - 1].matchLength); + U32 const offset = (match.srcIdx + dictSize) - match.dictIdx; + U32 const matchLength = match.matchLength; + sequences[nbSequences].offset = offset; + sequences[nbSequences].litLength = litLength; + sequences[nbSequences].matchLength = matchLength; + nbSequences++; + } + return nbSequences; +} + +/*-************************************* +* Internal utils +***************************************/ + +static size_t ZSTD_eDist_hamingDist(const BYTE* const a, + const BYTE* const b, size_t n) +{ + size_t i; + size_t dist = 0; + for (i = 0; i < n; i++) + dist += a[i] != b[i]; + return dist; +} + +/* This is a pretty naive recursive implementation that should only + * be used for quick tests obviously. Don't try and run this on a + * GB file or something. There are faster implementations. Use those + * if you need to run it for large files. */ +static size_t ZSTD_eDist_levenshteinDist(const BYTE* const s, + size_t const sn, const BYTE* const t, + size_t const tn) +{ + size_t a, b, c; + + if (!sn) + return tn; + if (!tn) + return sn; + + if (s[sn - 1] == t[tn - 1]) + return ZSTD_eDist_levenshteinDist( + s, sn - 1, t, tn - 1); + + a = ZSTD_eDist_levenshteinDist(s, sn - 1, t, tn - 1); + b = ZSTD_eDist_levenshteinDist(s, sn, t, tn - 1); + c = ZSTD_eDist_levenshteinDist(s, sn - 1, t, tn); + + if (a > b) + a = b; + if (a > c) + a = c; + + return a + 1; +} + +static void ZSTD_eDist_validateMatches(ZSTD_eDist_match* matches, + size_t const nbMatches, const BYTE* const dict, + size_t const dictSize, const BYTE* const src, + size_t const srcSize) +{ + size_t i; + for (i = 0; i < nbMatches; i++) { + ZSTD_eDist_match match = matches[i]; + U32 const dictIdx = match.dictIdx; + U32 const srcIdx = match.srcIdx; + U32 const matchLength = match.matchLength; + + assert(dictIdx + matchLength < dictSize); + assert(srcIdx + matchLength < srcSize); + assert(!memcmp(dict + dictIdx, src + srcIdx, matchLength)); + } +} + +/*-************************************* +* API +***************************************/ + +size_t ZSTD_eDist_genSequences(ZSTD_Sequence* sequences, + const void* dict, size_t dictSize, + const void* src, size_t srcSize, + int useHeuristics) +{ + size_t const nbDiags = dictSize + srcSize + 3; + S32* buffer = ZSTD_malloc(nbDiags * 2 * sizeof(S32), ZSTD_defaultCMem); + ZSTD_eDist_state state; + size_t nbSequences = 0; + + state.dict = (const BYTE*)dict; + state.src = (const BYTE*)src; + state.dictSize = dictSize; + state.srcSize = srcSize; + state.forwardDiag = buffer; + state.backwardDiag = buffer + nbDiags; + state.forwardDiag += srcSize + 1; + state.backwardDiag += srcSize + 1; + state.matches = ZSTD_malloc(srcSize * sizeof(ZSTD_eDist_match), ZSTD_defaultCMem); + state.nbMatches = 0; + + ZSTD_eDist_compare(&state, 0, dictSize, 0, srcSize, 1); + ZSTD_eDist_combineMatches(&state); + nbSequences = ZSTD_eDist_convertMatchesToSequences(sequences, &state); + + ZSTD_free(buffer, ZSTD_defaultCMem); + ZSTD_free(state.matches, ZSTD_defaultCMem); + + return nbSequences; +} diff --git a/3rdparty/zstd/contrib/match_finders/zstd_edist.h b/3rdparty/zstd/contrib/match_finders/zstd_edist.h new file mode 100644 index 00000000000..c739e2abc51 --- /dev/null +++ b/3rdparty/zstd/contrib/match_finders/zstd_edist.h @@ -0,0 +1,70 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ + +/* This match finder leverages techniques used in file comparison algorithms + * to find matches between a dictionary and a source file. + * + * The original motivation for studying this approach was to try and optimize + * Zstandard for the use case of patching: the most common scenario being + * updating an existing software package with the next version. When patching, + * the difference between the old version of the package and the new version + * is generally tiny (most of the new file will be identical to + * the old one). In more technical terms, the edit distance (the minimal number + * of changes required to take one sequence of bytes to another) between the + * files would be small relative to the size of the file. + * + * Various 'diffing' algorithms utilize this notion of edit distance and + * the corresponding concept of a minimal edit script between two + * sequences to identify the regions within two files where they differ. + * The core algorithm used in this match finder is described in: + * + * "An O(ND) Difference Algorithm and its Variations", Eugene W. Myers, + * Algorithmica Vol. 1, 1986, pp. 251-266, + * <https://doi.org/10.1007/BF01840446>. + * + * Additional algorithmic heuristics for speed improvement have also been included. + * These we inspired from implementations of various regular and binary diffing + * algorithms such as GNU diff, bsdiff, and Xdelta. + * + * Note: after some experimentation, this approach proved to not provide enough + * utility to justify the additional CPU used in finding matches. The one area + * where this approach consistently outperforms Zstandard even on level 19 is + * when compressing small files (<10 KB) using an equally small dictionary that + * is very similar to the source file. For the use case that this was intended, + * (large similar files) this approach by itself took 5-10X longer than zstd-19 and + * generally resulted in 2-3X larger files. The core advantage that zstd-19 has + * over this approach for match finding is the overlapping matches. This approach + * cannot find any. + * + * I'm leaving this in the contrib section in case this ever becomes interesting + * to explore again. + * */ + +#ifndef ZSTD_EDIST_H +#define ZSTD_EDIST_H + +/*-************************************* +* Dependencies +***************************************/ + +#include <stddef.h> +#include "zstd_internal.h" /* ZSTD_Sequence */ + +/*! ZSTD_eDist_genSequences() : + * Will populate the provided ZSTD_Sequence buffer with sequences + * based on the optimal or near-optimal (depending on 'useHeuristics') + * edit script between 'dict' and 'src.' + * @return : the number of sequences found */ +size_t ZSTD_eDist_genSequences(ZSTD_Sequence* sequences, + const void* dict, size_t dictSize, + const void* src, size_t srcSize, + int useHeuristics); + +#endif diff --git a/3rdparty/zstd/contrib/premake/premake4.lua b/3rdparty/zstd/contrib/premake/premake4.lua new file mode 100644 index 00000000000..6675e2e481c --- /dev/null +++ b/3rdparty/zstd/contrib/premake/premake4.lua @@ -0,0 +1,6 @@ +-- Include zstd.lua in your GENie or premake4 file, which exposes a project_zstd function +dofile('zstd.lua') + +solution 'example' + configurations { 'Debug', 'Release' } + project_zstd('../../lib/') diff --git a/3rdparty/zstd/contrib/premake/zstd.lua b/3rdparty/zstd/contrib/premake/zstd.lua new file mode 100644 index 00000000000..df1ace3ee8e --- /dev/null +++ b/3rdparty/zstd/contrib/premake/zstd.lua @@ -0,0 +1,80 @@ +-- This GENie/premake file copies the behavior of the Makefile in the lib folder. +-- Basic usage: project_zstd(ZSTD_DIR) + +function project_zstd(dir, compression, decompression, deprecated, dictbuilder, legacy) + if compression == nil then compression = true end + if decompression == nil then decompression = true end + if deprecated == nil then deprecated = false end + if dictbuilder == nil then dictbuilder = false end + + if legacy == nil then legacy = 0 end + + if not compression then + dictbuilder = false + deprecated = false + end + + if not decompression then + legacy = 0 + deprecated = false + end + + project 'zstd' + kind 'StaticLib' + language 'C' + + files { + dir .. 'zstd.h', + dir .. 'common/**.c', + dir .. 'common/**.h' + } + + if compression then + files { + dir .. 'compress/**.c', + dir .. 'compress/**.h' + } + end + + if decompression then + files { + dir .. 'decompress/**.c', + dir .. 'decompress/**.h' + } + end + + if dictbuilder then + files { + dir .. 'dictBuilder/**.c', + dir .. 'dictBuilder/**.h' + } + end + + if deprecated then + files { + dir .. 'deprecated/**.c', + dir .. 'deprecated/**.h' + } + end + + if legacy ~= 0 then + if legacy >= 8 then + files { + dir .. 'legacy/zstd_v0' .. (legacy - 7) .. '.*' + } + end + includedirs { + dir .. 'legacy' + } + end + + includedirs { + dir, + dir .. 'common' + } + + defines { + 'XXH_NAMESPACE=ZSTD_', + 'ZSTD_LEGACY_SUPPORT=' .. legacy + } +end diff --git a/3rdparty/zstd/contrib/pzstd/.gitignore b/3rdparty/zstd/contrib/pzstd/.gitignore new file mode 100644 index 00000000000..84e68fb07d5 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/.gitignore @@ -0,0 +1,2 @@ +# compilation result +pzstd diff --git a/3rdparty/zstd/contrib/pzstd/BUCK b/3rdparty/zstd/contrib/pzstd/BUCK new file mode 100644 index 00000000000..d04eeedd8a1 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/BUCK @@ -0,0 +1,72 @@ +cxx_library( + name='libpzstd', + visibility=['PUBLIC'], + header_namespace='', + exported_headers=[ + 'ErrorHolder.h', + 'Logging.h', + 'Pzstd.h', + ], + headers=[ + 'SkippableFrame.h', + ], + srcs=[ + 'Pzstd.cpp', + 'SkippableFrame.cpp', + ], + deps=[ + ':options', + '//contrib/pzstd/utils:utils', + '//lib:mem', + '//lib:zstd', + ], +) + +cxx_library( + name='options', + visibility=['PUBLIC'], + header_namespace='', + exported_headers=['Options.h'], + srcs=['Options.cpp'], + deps=[ + '//contrib/pzstd/utils:scope_guard', + '//lib:zstd', + '//programs:util', + ], +) + +cxx_binary( + name='pzstd', + visibility=['PUBLIC'], + srcs=['main.cpp'], + deps=[ + ':libpzstd', + ':options', + ], +) + +# Must run "make googletest" first +cxx_library( + name='gtest', + srcs=glob([ + 'googletest/googletest/src/gtest-all.cc', + 'googletest/googlemock/src/gmock-all.cc', + 'googletest/googlemock/src/gmock_main.cc', + ]), + header_namespace='', + exported_headers=subdir_glob([ + ('googletest/googletest/include', '**/*.h'), + ('googletest/googlemock/include', '**/*.h'), + ]), + headers=subdir_glob([ + ('googletest/googletest', 'src/*.cc'), + ('googletest/googletest', 'src/*.h'), + ('googletest/googlemock', 'src/*.cc'), + ('googletest/googlemock', 'src/*.h'), + ]), + platform_linker_flags=[ + ('android', []), + ('', ['-lpthread']), + ], + visibility=['PUBLIC'], +) diff --git a/3rdparty/zstd/contrib/pzstd/ErrorHolder.h b/3rdparty/zstd/contrib/pzstd/ErrorHolder.h new file mode 100644 index 00000000000..2c2797edea0 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/ErrorHolder.h @@ -0,0 +1,54 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +#pragma once + +#include <atomic> +#include <cassert> +#include <stdexcept> +#include <string> + +namespace pzstd { + +// Coordinates graceful shutdown of the pzstd pipeline +class ErrorHolder { + std::atomic<bool> error_; + std::string message_; + + public: + ErrorHolder() : error_(false) {} + + bool hasError() noexcept { + return error_.load(); + } + + void setError(std::string message) noexcept { + // Given multiple possibly concurrent calls, exactly one will ever succeed. + bool expected = false; + if (error_.compare_exchange_strong(expected, true)) { + message_ = std::move(message); + } + } + + bool check(bool predicate, std::string message) noexcept { + if (!predicate) { + setError(std::move(message)); + } + return !hasError(); + } + + std::string getError() noexcept { + error_.store(false); + return std::move(message_); + } + + ~ErrorHolder() { + assert(!hasError()); + } +}; +} diff --git a/3rdparty/zstd/contrib/pzstd/Logging.h b/3rdparty/zstd/contrib/pzstd/Logging.h new file mode 100644 index 00000000000..84a08d20080 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/Logging.h @@ -0,0 +1,72 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +#pragma once + +#include <cstdio> +#include <mutex> + +namespace pzstd { + +constexpr int kLogError = 1; +constexpr int kLogInfo = 2; +constexpr int kLogDebug = 3; +constexpr int kLogVerbose = 4; + +class Logger { + std::mutex mutex_; + FILE* out_; + const int level_; + + using Clock = std::chrono::system_clock; + Clock::time_point lastUpdate_; + std::chrono::milliseconds refreshRate_; + + public: + explicit Logger(int level, FILE* out = stderr) + : out_(out), level_(level), lastUpdate_(Clock::now()), + refreshRate_(150) {} + + + bool logsAt(int level) { + return level <= level_; + } + + template <typename... Args> + void operator()(int level, const char *fmt, Args... args) { + if (level > level_) { + return; + } + std::lock_guard<std::mutex> lock(mutex_); + std::fprintf(out_, fmt, args...); + } + + template <typename... Args> + void update(int level, const char *fmt, Args... args) { + if (level > level_) { + return; + } + std::lock_guard<std::mutex> lock(mutex_); + auto now = Clock::now(); + if (now - lastUpdate_ > refreshRate_) { + lastUpdate_ = now; + std::fprintf(out_, "\r"); + std::fprintf(out_, fmt, args...); + } + } + + void clear(int level) { + if (level > level_) { + return; + } + std::lock_guard<std::mutex> lock(mutex_); + std::fprintf(out_, "\r%79s\r", ""); + } +}; + +} diff --git a/3rdparty/zstd/contrib/pzstd/Makefile b/3rdparty/zstd/contrib/pzstd/Makefile new file mode 100644 index 00000000000..e62f8e873c5 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/Makefile @@ -0,0 +1,268 @@ +# ################################################################ +# Copyright (c) Meta Platforms, Inc. and affiliates. +# All rights reserved. +# +# This source code is licensed under both the BSD-style license (found in the +# LICENSE file in the root directory of this source tree) and the GPLv2 (found +# in the COPYING file in the root directory of this source tree). +# ################################################################ + +# Standard variables for installation +DESTDIR ?= +PREFIX ?= /usr/local +BINDIR := $(DESTDIR)$(PREFIX)/bin + +ZSTDDIR = ../../lib +PROGDIR = ../../programs + +# External program to use to run tests, e.g. qemu or valgrind +TESTPROG ?= +# Flags to pass to the tests +TESTFLAGS ?= + +# We use gcc/clang to generate the header dependencies of files +DEPFLAGS = -MMD -MP -MF $*.Td +POSTCOMPILE = mv -f $*.Td $*.d + +# CFLAGS, CXXFLAGS, CPPFLAGS, and LDFLAGS are for the users to override +CFLAGS ?= -O3 -Wall -Wextra +CXXFLAGS ?= -O3 -Wall -Wextra -pedantic +CPPFLAGS ?= +LDFLAGS ?= + +# PZstd uses legacy APIs +CFLAGS += -Wno-deprecated-declarations + +# Include flags +PZSTD_INC = -I$(ZSTDDIR) -I$(ZSTDDIR)/common -I$(PROGDIR) -I. +GTEST_INC = -isystem googletest/googletest/include + +# If default C++ version is older than C++11, explicitly set C++11, which is the +# minimum required by the code. +ifeq ($(shell echo "\043if __cplusplus < 201103L\n\043error\n\043endif" | $(CXX) -x c++ -Werror -c - -o /dev/null 2>/dev/null && echo 1 || echo 0),0) +PZSTD_CXX_STD := -std=c++11 +endif + +PZSTD_CPPFLAGS = $(PZSTD_INC) +PZSTD_CCXXFLAGS = +PZSTD_CFLAGS = $(PZSTD_CCXXFLAGS) +PZSTD_CXXFLAGS = $(PZSTD_CCXXFLAGS) $(PZSTD_CXX_STD) +PZSTD_LDFLAGS = +EXTRA_FLAGS = +ALL_CFLAGS = $(EXTRA_FLAGS) $(CPPFLAGS) $(PZSTD_CPPFLAGS) $(CFLAGS) $(PZSTD_CFLAGS) +ALL_CXXFLAGS = $(EXTRA_FLAGS) $(CPPFLAGS) $(PZSTD_CPPFLAGS) $(CXXFLAGS) $(PZSTD_CXXFLAGS) +ALL_LDFLAGS = $(EXTRA_FLAGS) $(CXXFLAGS) $(LDFLAGS) $(PZSTD_LDFLAGS) + + +# gtest libraries need to go before "-lpthread" because they depend on it. +GTEST_LIB = -L googletest/build/googlemock/gtest +LIBS = + +# Compilation commands +LD_COMMAND = $(CXX) $^ $(ALL_LDFLAGS) $(LIBS) -pthread -o $@ +CC_COMMAND = $(CC) $(DEPFLAGS) $(ALL_CFLAGS) -c $< -o $@ +CXX_COMMAND = $(CXX) $(DEPFLAGS) $(ALL_CXXFLAGS) -c $< -o $@ + +# List all the pzstd source files so we can determine their dependencies +PZSTD_SRCS := $(wildcard *.cpp) +PZSTD_TESTS := $(wildcard test/*.cpp) +UTILS_TESTS := $(wildcard utils/test/*.cpp) +ALL_SRCS := $(PZSTD_SRCS) $(PZSTD_TESTS) $(UTILS_TESTS) + + +# Define *.exe as extension for Windows systems +ifneq (,$(filter Windows%,$(OS))) +EXT =.exe +else +EXT = +endif + +# Standard targets +.PHONY: default +default: all + +.PHONY: test-pzstd +test-pzstd: TESTFLAGS=--gtest_filter=-*ExtremelyLarge* +test-pzstd: clean googletest pzstd tests check + +.PHONY: test-pzstd32 +test-pzstd32: clean googletest32 all32 check + +.PHONY: test-pzstd-tsan +test-pzstd-tsan: LDFLAGS=-fuse-ld=gold +test-pzstd-tsan: TESTFLAGS=--gtest_filter=-*ExtremelyLarge* +test-pzstd-tsan: clean googletest tsan check + +.PHONY: test-pzstd-asan +test-pzstd-asan: LDFLAGS=-fuse-ld=gold +test-pzstd-asan: TESTFLAGS=--gtest_filter=-*ExtremelyLarge* +test-pzstd-asan: clean asan check + +.PHONY: check +check: + $(TESTPROG) ./utils/test/BufferTest$(EXT) $(TESTFLAGS) + $(TESTPROG) ./utils/test/RangeTest$(EXT) $(TESTFLAGS) + $(TESTPROG) ./utils/test/ResourcePoolTest$(EXT) $(TESTFLAGS) + $(TESTPROG) ./utils/test/ScopeGuardTest$(EXT) $(TESTFLAGS) + $(TESTPROG) ./utils/test/ThreadPoolTest$(EXT) $(TESTFLAGS) + $(TESTPROG) ./utils/test/WorkQueueTest$(EXT) $(TESTFLAGS) + $(TESTPROG) ./test/OptionsTest$(EXT) $(TESTFLAGS) + $(TESTPROG) ./test/PzstdTest$(EXT) $(TESTFLAGS) + +.PHONY: install +install: PZSTD_CPPFLAGS += -DNDEBUG +install: pzstd$(EXT) + install -d -m 755 $(BINDIR)/ + install -m 755 pzstd$(EXT) $(BINDIR)/pzstd$(EXT) + +.PHONY: uninstall +uninstall: + $(RM) $(BINDIR)/pzstd$(EXT) + +# Targets for many different builds +.PHONY: all +all: PZSTD_CPPFLAGS += -DNDEBUG +all: pzstd$(EXT) + +.PHONY: debug +debug: EXTRA_FLAGS += -g +debug: pzstd$(EXT) tests roundtrip + +.PHONY: tsan +tsan: PZSTD_CCXXFLAGS += -fsanitize=thread -fPIC +tsan: PZSTD_LDFLAGS += -fsanitize=thread +tsan: debug + +.PHONY: asan +asan: EXTRA_FLAGS += -fsanitize=address +asan: debug + +.PHONY: ubsan +ubsan: EXTRA_FLAGS += -fsanitize=undefined +ubsan: debug + +.PHONY: all32 +all32: EXTRA_FLAGS += -m32 +all32: all tests roundtrip + +.PHONY: debug32 +debug32: EXTRA_FLAGS += -m32 +debug32: debug + +.PHONY: asan32 +asan32: EXTRA_FLAGS += -m32 +asan32: asan + +.PHONY: tsan32 +tsan32: EXTRA_FLAGS += -m32 +tsan32: tsan + +.PHONY: ubsan32 +ubsan32: EXTRA_FLAGS += -m32 +ubsan32: ubsan + +# Run long round trip tests +.PHONY: roundtripcheck +roundtripcheck: roundtrip check + $(TESTPROG) ./test/RoundTripTest$(EXT) $(TESTFLAGS) + +# Build the main binary +pzstd$(EXT): main.o $(PROGDIR)/util.o Options.o Pzstd.o SkippableFrame.o $(ZSTDDIR)/libzstd.a + $(LD_COMMAND) + +# Target that depends on all the tests +.PHONY: tests +tests: EXTRA_FLAGS += -Wno-deprecated-declarations +tests: $(patsubst %,%$(EXT),$(basename $(PZSTD_TESTS) $(UTILS_TESTS))) + +# Build the round trip tests +.PHONY: roundtrip +roundtrip: EXTRA_FLAGS += -Wno-deprecated-declarations +roundtrip: test/RoundTripTest$(EXT) + +# Use the static library that zstd builds for simplicity and +# so we get the compiler options correct +.PHONY: $(ZSTDDIR)/libzstd.a +$(ZSTDDIR)/libzstd.a: + CFLAGS="$(ALL_CFLAGS)" LDFLAGS="$(ALL_LDFLAGS)" $(MAKE) -C $(ZSTDDIR) libzstd.a + +# Rules to build the tests +test/RoundTripTest$(EXT): test/RoundTripTest.o $(PROGDIR)/datagen.o \ + $(PROGDIR)/util.o Options.o \ + Pzstd.o SkippableFrame.o $(ZSTDDIR)/libzstd.a + $(LD_COMMAND) + +test/%Test$(EXT): PZSTD_LDFLAGS += $(GTEST_LIB) +test/%Test$(EXT): LIBS += -lgtest -lgtest_main +test/%Test$(EXT): test/%Test.o $(PROGDIR)/datagen.o \ + $(PROGDIR)/util.o Options.o Pzstd.o \ + SkippableFrame.o $(ZSTDDIR)/libzstd.a + $(LD_COMMAND) + +utils/test/%Test$(EXT): PZSTD_LDFLAGS += $(GTEST_LIB) +utils/test/%Test$(EXT): LIBS += -lgtest -lgtest_main +utils/test/%Test$(EXT): utils/test/%Test.o + $(LD_COMMAND) + + +GTEST_CMAKEFLAGS = + +# Install googletest +.PHONY: googletest +googletest: PZSTD_CCXXFLAGS += -fPIC +googletest: + @$(RM) -rf googletest + @git clone https://github.com/google/googletest + @mkdir -p googletest/build + @cd googletest/build && cmake $(GTEST_CMAKEFLAGS) -DCMAKE_CXX_FLAGS="$(ALL_CXXFLAGS)" .. && $(MAKE) + +.PHONY: googletest32 +googletest32: PZSTD_CCXXFLAGS += -m32 +googletest32: googletest + +.PHONY: googletest-mingw64 +googletest-mingw64: GTEST_CMAKEFLAGS += -G "MSYS Makefiles" +googletest-mingw64: googletest + +.PHONY: clean +clean: + $(RM) -f *.o pzstd$(EXT) *.Td *.d + $(RM) -f test/*.o test/*Test$(EXT) test/*.Td test/*.d + $(RM) -f utils/test/*.o utils/test/*Test$(EXT) utils/test/*.Td utils/test/*.d + $(RM) -f $(PROGDIR)/*.o $(PROGDIR)/*.Td $(PROGDIR)/*.d + $(MAKE) -C $(ZSTDDIR) clean + @echo Cleaning completed + + +# Cancel implicit rules +%.o: %.c +%.o: %.cpp + +# Object file rules +%.o: %.c + $(CC_COMMAND) + $(POSTCOMPILE) + +$(PROGDIR)/%.o: $(PROGDIR)/%.c + $(CC_COMMAND) + $(POSTCOMPILE) + +%.o: %.cpp + $(CXX_COMMAND) + $(POSTCOMPILE) + +test/%.o: PZSTD_CPPFLAGS += $(GTEST_INC) +test/%.o: test/%.cpp + $(CXX_COMMAND) + $(POSTCOMPILE) + +utils/test/%.o: PZSTD_CPPFLAGS += $(GTEST_INC) +utils/test/%.o: utils/test/%.cpp + $(CXX_COMMAND) + $(POSTCOMPILE) + +# Dependency file stuff +.PRECIOUS: %.d test/%.d utils/test/%.d + +# Include rules that specify header file dependencies +-include $(patsubst %,%.d,$(basename $(ALL_SRCS))) diff --git a/3rdparty/zstd/contrib/pzstd/Options.cpp b/3rdparty/zstd/contrib/pzstd/Options.cpp new file mode 100644 index 00000000000..a77824edb05 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/Options.cpp @@ -0,0 +1,424 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +#include "Options.h" +#include "util.h" +#include "utils/ScopeGuard.h" + +#include <algorithm> +#include <cassert> +#include <cstdio> +#include <cstring> +#include <iterator> +#include <thread> +#include <vector> + + +namespace pzstd { + +namespace { +unsigned defaultNumThreads() { +#ifdef PZSTD_NUM_THREADS + return PZSTD_NUM_THREADS; +#else + return std::thread::hardware_concurrency(); +#endif +} + +unsigned parseUnsigned(const char **arg) { + unsigned result = 0; + while (**arg >= '0' && **arg <= '9') { + result *= 10; + result += **arg - '0'; + ++(*arg); + } + return result; +} + +const char *getArgument(const char *options, const char **argv, int &i, + int argc) { + if (options[1] != 0) { + return options + 1; + } + ++i; + if (i == argc) { + std::fprintf(stderr, "Option -%c requires an argument, but none provided\n", + *options); + return nullptr; + } + return argv[i]; +} + +const std::string kZstdExtension = ".zst"; +constexpr char kStdIn[] = "-"; +constexpr char kStdOut[] = "-"; +constexpr unsigned kDefaultCompressionLevel = 3; +constexpr unsigned kMaxNonUltraCompressionLevel = 19; + +#ifdef _WIN32 +const char nullOutput[] = "nul"; +#else +const char nullOutput[] = "/dev/null"; +#endif + +void notSupported(const char *option) { + std::fprintf(stderr, "Operation not supported: %s\n", option); +} + +void usage() { + std::fprintf(stderr, "Usage:\n"); + std::fprintf(stderr, " pzstd [args] [FILE(s)]\n"); + std::fprintf(stderr, "Parallel ZSTD options:\n"); + std::fprintf(stderr, " -p, --processes # : number of threads to use for (de)compression (default:<numcpus>)\n"); + + std::fprintf(stderr, "ZSTD options:\n"); + std::fprintf(stderr, " -# : # compression level (1-%d, default:%d)\n", kMaxNonUltraCompressionLevel, kDefaultCompressionLevel); + std::fprintf(stderr, " -d, --decompress : decompression\n"); + std::fprintf(stderr, " -o file : result stored into `file` (only if 1 input file)\n"); + std::fprintf(stderr, " -f, --force : overwrite output without prompting, (de)compress links\n"); + std::fprintf(stderr, " --rm : remove source file(s) after successful (de)compression\n"); + std::fprintf(stderr, " -k, --keep : preserve source file(s) (default)\n"); + std::fprintf(stderr, " -h, --help : display help and exit\n"); + std::fprintf(stderr, " -V, --version : display version number and exit\n"); + std::fprintf(stderr, " -v, --verbose : verbose mode; specify multiple times to increase log level (default:2)\n"); + std::fprintf(stderr, " -q, --quiet : suppress warnings; specify twice to suppress errors too\n"); + std::fprintf(stderr, " -c, --stdout : write to standard output (even if it is the console)\n"); +#ifdef UTIL_HAS_CREATEFILELIST + std::fprintf(stderr, " -r : operate recursively on directories\n"); +#endif + std::fprintf(stderr, " --ultra : enable levels beyond %i, up to %i (requires more memory)\n", kMaxNonUltraCompressionLevel, ZSTD_maxCLevel()); + std::fprintf(stderr, " -C, --check : integrity check (default)\n"); + std::fprintf(stderr, " --no-check : no integrity check\n"); + std::fprintf(stderr, " -t, --test : test compressed file integrity\n"); + std::fprintf(stderr, " -- : all arguments after \"--\" are treated as files\n"); +} +} // anonymous namespace + +Options::Options() + : numThreads(defaultNumThreads()), maxWindowLog(23), + compressionLevel(kDefaultCompressionLevel), decompress(false), + overwrite(false), keepSource(true), writeMode(WriteMode::Auto), + checksum(true), verbosity(2) {} + +Options::Status Options::parse(int argc, const char **argv) { + bool test = false; + bool recursive = false; + bool ultra = false; + bool forceStdout = false; + bool followLinks = false; + // Local copy of input files, which are pointers into argv. + std::vector<const char *> localInputFiles; + for (int i = 1; i < argc; ++i) { + const char *arg = argv[i]; + // Protect against empty arguments + if (arg[0] == 0) { + continue; + } + // Everything after "--" is an input file + if (!std::strcmp(arg, "--")) { + ++i; + std::copy(argv + i, argv + argc, std::back_inserter(localInputFiles)); + break; + } + // Long arguments that don't have a short option + { + bool isLongOption = true; + if (!std::strcmp(arg, "--rm")) { + keepSource = false; + } else if (!std::strcmp(arg, "--ultra")) { + ultra = true; + maxWindowLog = 0; + } else if (!std::strcmp(arg, "--no-check")) { + checksum = false; + } else if (!std::strcmp(arg, "--sparse")) { + writeMode = WriteMode::Sparse; + notSupported("Sparse mode"); + return Status::Failure; + } else if (!std::strcmp(arg, "--no-sparse")) { + writeMode = WriteMode::Regular; + notSupported("Sparse mode"); + return Status::Failure; + } else if (!std::strcmp(arg, "--dictID")) { + notSupported(arg); + return Status::Failure; + } else if (!std::strcmp(arg, "--no-dictID")) { + notSupported(arg); + return Status::Failure; + } else { + isLongOption = false; + } + if (isLongOption) { + continue; + } + } + // Arguments with a short option simply set their short option. + const char *options = nullptr; + if (!std::strcmp(arg, "--processes")) { + options = "p"; + } else if (!std::strcmp(arg, "--version")) { + options = "V"; + } else if (!std::strcmp(arg, "--help")) { + options = "h"; + } else if (!std::strcmp(arg, "--decompress")) { + options = "d"; + } else if (!std::strcmp(arg, "--force")) { + options = "f"; + } else if (!std::strcmp(arg, "--stdout")) { + options = "c"; + } else if (!std::strcmp(arg, "--keep")) { + options = "k"; + } else if (!std::strcmp(arg, "--verbose")) { + options = "v"; + } else if (!std::strcmp(arg, "--quiet")) { + options = "q"; + } else if (!std::strcmp(arg, "--check")) { + options = "C"; + } else if (!std::strcmp(arg, "--test")) { + options = "t"; + } else if (arg[0] == '-' && arg[1] != 0) { + options = arg + 1; + } else { + localInputFiles.emplace_back(arg); + continue; + } + assert(options != nullptr); + + bool finished = false; + while (!finished && *options != 0) { + // Parse the compression level + if (*options >= '0' && *options <= '9') { + compressionLevel = parseUnsigned(&options); + continue; + } + + switch (*options) { + case 'h': + case 'H': + usage(); + return Status::Message; + case 'V': + std::fprintf(stderr, "PZSTD version: %s.\n", ZSTD_VERSION_STRING); + return Status::Message; + case 'p': { + finished = true; + const char *optionArgument = getArgument(options, argv, i, argc); + if (optionArgument == nullptr) { + return Status::Failure; + } + if (*optionArgument < '0' || *optionArgument > '9') { + std::fprintf(stderr, "Option -p expects a number, but %s provided\n", + optionArgument); + return Status::Failure; + } + numThreads = parseUnsigned(&optionArgument); + if (*optionArgument != 0) { + std::fprintf(stderr, + "Option -p expects a number, but %u%s provided\n", + numThreads, optionArgument); + return Status::Failure; + } + break; + } + case 'o': { + finished = true; + const char *optionArgument = getArgument(options, argv, i, argc); + if (optionArgument == nullptr) { + return Status::Failure; + } + outputFile = optionArgument; + break; + } + case 'C': + checksum = true; + break; + case 'k': + keepSource = true; + break; + case 'd': + decompress = true; + break; + case 'f': + overwrite = true; + forceStdout = true; + followLinks = true; + break; + case 't': + test = true; + decompress = true; + break; +#ifdef UTIL_HAS_CREATEFILELIST + case 'r': + recursive = true; + break; +#endif + case 'c': + outputFile = kStdOut; + forceStdout = true; + break; + case 'v': + ++verbosity; + break; + case 'q': + --verbosity; + // Ignore them for now + break; + // Unsupported options from Zstd + case 'D': + case 's': + notSupported("Zstd dictionaries."); + return Status::Failure; + case 'b': + case 'e': + case 'i': + case 'B': + notSupported("Zstd benchmarking options."); + return Status::Failure; + default: + std::fprintf(stderr, "Invalid argument: %s\n", arg); + return Status::Failure; + } + if (!finished) { + ++options; + } + } // while (*options != 0); + } // for (int i = 1; i < argc; ++i); + + // Set options for test mode + if (test) { + outputFile = nullOutput; + keepSource = true; + } + + // Input file defaults to standard input if not provided. + if (localInputFiles.empty()) { + localInputFiles.emplace_back(kStdIn); + } + + // Check validity of input files + if (localInputFiles.size() > 1) { + const auto it = std::find(localInputFiles.begin(), localInputFiles.end(), + std::string{kStdIn}); + if (it != localInputFiles.end()) { + std::fprintf( + stderr, + "Cannot specify standard input when handling multiple files\n"); + return Status::Failure; + } + } + if (localInputFiles.size() > 1 || recursive) { + if (!outputFile.empty() && outputFile != nullOutput) { + std::fprintf( + stderr, + "Cannot specify an output file when handling multiple inputs\n"); + return Status::Failure; + } + } + + g_utilDisplayLevel = verbosity; + // Remove local input files that are symbolic links + if (!followLinks) { + std::remove_if(localInputFiles.begin(), localInputFiles.end(), + [&](const char *path) { + bool isLink = UTIL_isLink(path); + if (isLink && verbosity >= 2) { + std::fprintf( + stderr, + "Warning : %s is symbolic link, ignoring\n", + path); + } + return isLink; + }); + } + + // Translate input files/directories into files to (de)compress + if (recursive) { + FileNamesTable* const files = UTIL_createExpandedFNT(localInputFiles.data(), localInputFiles.size(), followLinks); + if (files == nullptr) { + std::fprintf(stderr, "Error traversing directories\n"); + return Status::Failure; + } + auto guard = + makeScopeGuard([&] { UTIL_freeFileNamesTable(files); }); + if (files->tableSize == 0) { + std::fprintf(stderr, "No files found\n"); + return Status::Failure; + } + inputFiles.resize(files->tableSize); + std::copy(files->fileNames, files->fileNames + files->tableSize, inputFiles.begin()); + } else { + inputFiles.resize(localInputFiles.size()); + std::copy(localInputFiles.begin(), localInputFiles.end(), + inputFiles.begin()); + } + localInputFiles.clear(); + assert(!inputFiles.empty()); + + // If reading from standard input, default to standard output + if (inputFiles[0] == kStdIn && outputFile.empty()) { + assert(inputFiles.size() == 1); + outputFile = "-"; + } + + if (inputFiles[0] == kStdIn && IS_CONSOLE(stdin)) { + assert(inputFiles.size() == 1); + std::fprintf(stderr, "Cannot read input from interactive console\n"); + return Status::Failure; + } + if (outputFile == "-" && IS_CONSOLE(stdout) && !(forceStdout && decompress)) { + std::fprintf(stderr, "Will not write to console stdout unless -c or -f is " + "specified and decompressing\n"); + return Status::Failure; + } + + // Check compression level + { + unsigned maxCLevel = + ultra ? ZSTD_maxCLevel() : kMaxNonUltraCompressionLevel; + if (compressionLevel > maxCLevel || compressionLevel == 0) { + std::fprintf(stderr, "Invalid compression level %u.\n", compressionLevel); + return Status::Failure; + } + } + + // Check that numThreads is set + if (numThreads == 0) { + std::fprintf(stderr, "Invalid arguments: # of threads not specified " + "and unable to determine hardware concurrency.\n"); + return Status::Failure; + } + + // Modify verbosity + // If we are piping input and output, turn off interaction + if (inputFiles[0] == kStdIn && outputFile == kStdOut && verbosity == 2) { + verbosity = 1; + } + // If we are in multi-file mode, turn off interaction + if (inputFiles.size() > 1 && verbosity == 2) { + verbosity = 1; + } + + return Status::Success; +} + +std::string Options::getOutputFile(const std::string &inputFile) const { + if (!outputFile.empty()) { + return outputFile; + } + // Attempt to add/remove zstd extension from the input file + if (decompress) { + int stemSize = inputFile.size() - kZstdExtension.size(); + if (stemSize > 0 && inputFile.substr(stemSize) == kZstdExtension) { + return inputFile.substr(0, stemSize); + } else { + return ""; + } + } else { + return inputFile + kZstdExtension; + } +} +} diff --git a/3rdparty/zstd/contrib/pzstd/Options.h b/3rdparty/zstd/contrib/pzstd/Options.h new file mode 100644 index 00000000000..92c18a350cf --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/Options.h @@ -0,0 +1,71 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +#pragma once + +#define ZSTD_STATIC_LINKING_ONLY +#define ZSTD_DISABLE_DEPRECATE_WARNINGS /* No deprecation warnings, pzstd itself is deprecated + * and uses deprecated functions + */ +#include "zstd.h" +#undef ZSTD_STATIC_LINKING_ONLY + +#include <cstdint> +#include <string> +#include <vector> + +namespace pzstd { + +struct Options { + enum class WriteMode { Regular, Auto, Sparse }; + + unsigned numThreads; + unsigned maxWindowLog; + unsigned compressionLevel; + bool decompress; + std::vector<std::string> inputFiles; + std::string outputFile; + bool overwrite; + bool keepSource; + WriteMode writeMode; + bool checksum; + int verbosity; + + enum class Status { + Success, // Successfully parsed options + Failure, // Failure to parse options + Message // Options specified to print a message (e.g. "-h") + }; + + Options(); + Options(unsigned numThreads, unsigned maxWindowLog, unsigned compressionLevel, + bool decompress, std::vector<std::string> inputFiles, + std::string outputFile, bool overwrite, bool keepSource, + WriteMode writeMode, bool checksum, int verbosity) + : numThreads(numThreads), maxWindowLog(maxWindowLog), + compressionLevel(compressionLevel), decompress(decompress), + inputFiles(std::move(inputFiles)), outputFile(std::move(outputFile)), + overwrite(overwrite), keepSource(keepSource), writeMode(writeMode), + checksum(checksum), verbosity(verbosity) {} + + Status parse(int argc, const char **argv); + + ZSTD_parameters determineParameters() const { + ZSTD_parameters params = ZSTD_getParams(compressionLevel, 0, 0); + params.fParams.contentSizeFlag = 0; + params.fParams.checksumFlag = checksum; + if (maxWindowLog != 0 && params.cParams.windowLog > maxWindowLog) { + params.cParams.windowLog = maxWindowLog; + params.cParams = ZSTD_adjustCParams(params.cParams, 0, 0); + } + return params; + } + + std::string getOutputFile(const std::string &inputFile) const; +}; +} diff --git a/3rdparty/zstd/contrib/pzstd/Pzstd.cpp b/3rdparty/zstd/contrib/pzstd/Pzstd.cpp new file mode 100644 index 00000000000..67b94199149 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/Pzstd.cpp @@ -0,0 +1,618 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +#include "platform.h" /* Large Files support, SET_BINARY_MODE */ +#include "Pzstd.h" +#include "SkippableFrame.h" +#include "utils/FileSystem.h" +#include "utils/Portability.h" +#include "utils/Range.h" +#include "utils/ScopeGuard.h" +#include "utils/ThreadPool.h" +#include "utils/WorkQueue.h" + +#include <algorithm> +#include <chrono> +#include <cinttypes> +#include <cstddef> +#include <cstdio> +#include <memory> +#include <string> + + +namespace pzstd { + +namespace { +#ifdef _WIN32 +const std::string nullOutput = "nul"; +#else +const std::string nullOutput = "/dev/null"; +#endif +} + +using std::size_t; + +static std::uintmax_t fileSizeOrZero(const std::string &file) { + if (file == "-") { + return 0; + } + std::error_code ec; + auto size = file_size(file, ec); + if (ec) { + size = 0; + } + return size; +} + +static std::uint64_t handleOneInput(const Options &options, + const std::string &inputFile, + FILE* inputFd, + const std::string &outputFile, + FILE* outputFd, + SharedState& state) { + auto inputSize = fileSizeOrZero(inputFile); + // WorkQueue outlives ThreadPool so in the case of error we are certain + // we don't accidentally try to call push() on it after it is destroyed + WorkQueue<std::shared_ptr<BufferWorkQueue>> outs{options.numThreads + 1}; + std::uint64_t bytesRead; + std::uint64_t bytesWritten; + { + // Initialize the (de)compression thread pool with numThreads + ThreadPool executor(options.numThreads); + // Run the reader thread on an extra thread + ThreadPool readExecutor(1); + if (!options.decompress) { + // Add a job that reads the input and starts all the compression jobs + readExecutor.add( + [&state, &outs, &executor, inputFd, inputSize, &options, &bytesRead] { + bytesRead = asyncCompressChunks( + state, + outs, + executor, + inputFd, + inputSize, + options.numThreads, + options.determineParameters()); + }); + // Start writing + bytesWritten = writeFile(state, outs, outputFd, options.decompress); + } else { + // Add a job that reads the input and starts all the decompression jobs + readExecutor.add([&state, &outs, &executor, inputFd, &bytesRead] { + bytesRead = asyncDecompressFrames(state, outs, executor, inputFd); + }); + // Start writing + bytesWritten = writeFile(state, outs, outputFd, options.decompress); + } + } + if (!state.errorHolder.hasError()) { + std::string inputFileName = inputFile == "-" ? "stdin" : inputFile; + std::string outputFileName = outputFile == "-" ? "stdout" : outputFile; + if (!options.decompress) { + double ratio = static_cast<double>(bytesWritten) / + static_cast<double>(bytesRead + !bytesRead); + state.log(kLogInfo, "%-20s :%6.2f%% (%6" PRIu64 " => %6" PRIu64 + " bytes, %s)\n", + inputFileName.c_str(), ratio * 100, bytesRead, bytesWritten, + outputFileName.c_str()); + } else { + state.log(kLogInfo, "%-20s: %" PRIu64 " bytes \n", + inputFileName.c_str(),bytesWritten); + } + } + return bytesWritten; +} + +static FILE *openInputFile(const std::string &inputFile, + ErrorHolder &errorHolder) { + if (inputFile == "-") { + SET_BINARY_MODE(stdin); + return stdin; + } + // Check if input file is a directory + { + std::error_code ec; + if (is_directory(inputFile, ec)) { + errorHolder.setError("Output file is a directory -- ignored"); + return nullptr; + } + } + auto inputFd = std::fopen(inputFile.c_str(), "rb"); + if (!errorHolder.check(inputFd != nullptr, "Failed to open input file")) { + return nullptr; + } + return inputFd; +} + +static FILE *openOutputFile(const Options &options, + const std::string &outputFile, + SharedState& state) { + if (outputFile == "-") { + SET_BINARY_MODE(stdout); + return stdout; + } + // Check if the output file exists and then open it + if (!options.overwrite && outputFile != nullOutput) { + auto outputFd = std::fopen(outputFile.c_str(), "rb"); + if (outputFd != nullptr) { + std::fclose(outputFd); + if (!state.log.logsAt(kLogInfo)) { + state.errorHolder.setError("Output file exists"); + return nullptr; + } + state.log( + kLogInfo, + "pzstd: %s already exists; do you wish to overwrite (y/n) ? ", + outputFile.c_str()); + int c = getchar(); + if (c != 'y' && c != 'Y') { + state.errorHolder.setError("Not overwritten"); + return nullptr; + } + } + } + auto outputFd = std::fopen(outputFile.c_str(), "wb"); + if (!state.errorHolder.check( + outputFd != nullptr, "Failed to open output file")) { + return nullptr; + } + return outputFd; +} + +int pzstdMain(const Options &options) { + int returnCode = 0; + SharedState state(options); + for (const auto& input : options.inputFiles) { + // Setup the shared state + auto printErrorGuard = makeScopeGuard([&] { + if (state.errorHolder.hasError()) { + returnCode = 1; + state.log(kLogError, "pzstd: %s: %s.\n", input.c_str(), + state.errorHolder.getError().c_str()); + } + }); + // Open the input file + auto inputFd = openInputFile(input, state.errorHolder); + if (inputFd == nullptr) { + continue; + } + auto closeInputGuard = makeScopeGuard([&] { std::fclose(inputFd); }); + // Open the output file + auto outputFile = options.getOutputFile(input); + if (!state.errorHolder.check(outputFile != "", + "Input file does not have extension .zst")) { + continue; + } + auto outputFd = openOutputFile(options, outputFile, state); + if (outputFd == nullptr) { + continue; + } + auto closeOutputGuard = makeScopeGuard([&] { std::fclose(outputFd); }); + // (de)compress the file + handleOneInput(options, input, inputFd, outputFile, outputFd, state); + if (state.errorHolder.hasError()) { + continue; + } + // Delete the input file if necessary + if (!options.keepSource) { + // Be sure that we are done and have written everything before we delete + if (!state.errorHolder.check(std::fclose(inputFd) == 0, + "Failed to close input file")) { + continue; + } + closeInputGuard.dismiss(); + if (!state.errorHolder.check(std::fclose(outputFd) == 0, + "Failed to close output file")) { + continue; + } + closeOutputGuard.dismiss(); + if (std::remove(input.c_str()) != 0) { + state.errorHolder.setError("Failed to remove input file"); + continue; + } + } + } + // Returns 1 if any of the files failed to (de)compress. + return returnCode; +} + +/// Construct a `ZSTD_inBuffer` that points to the data in `buffer`. +static ZSTD_inBuffer makeZstdInBuffer(const Buffer& buffer) { + return ZSTD_inBuffer{buffer.data(), buffer.size(), 0}; +} + +/** + * Advance `buffer` and `inBuffer` by the amount of data read, as indicated by + * `inBuffer.pos`. + */ +void advance(Buffer& buffer, ZSTD_inBuffer& inBuffer) { + auto pos = inBuffer.pos; + inBuffer.src = static_cast<const unsigned char*>(inBuffer.src) + pos; + inBuffer.size -= pos; + inBuffer.pos = 0; + return buffer.advance(pos); +} + +/// Construct a `ZSTD_outBuffer` that points to the data in `buffer`. +static ZSTD_outBuffer makeZstdOutBuffer(Buffer& buffer) { + return ZSTD_outBuffer{buffer.data(), buffer.size(), 0}; +} + +/** + * Split `buffer` and advance `outBuffer` by the amount of data written, as + * indicated by `outBuffer.pos`. + */ +Buffer split(Buffer& buffer, ZSTD_outBuffer& outBuffer) { + auto pos = outBuffer.pos; + outBuffer.dst = static_cast<unsigned char*>(outBuffer.dst) + pos; + outBuffer.size -= pos; + outBuffer.pos = 0; + return buffer.splitAt(pos); +} + +/** + * Stream chunks of input from `in`, compress it, and stream it out to `out`. + * + * @param state The shared state + * @param in Queue that we `pop()` input buffers from + * @param out Queue that we `push()` compressed output buffers to + * @param maxInputSize An upper bound on the size of the input + */ +static void compress( + SharedState& state, + std::shared_ptr<BufferWorkQueue> in, + std::shared_ptr<BufferWorkQueue> out, + size_t maxInputSize) { + auto& errorHolder = state.errorHolder; + auto guard = makeScopeGuard([&] { out->finish(); }); + // Initialize the CCtx + auto ctx = state.cStreamPool->get(); + if (!errorHolder.check(ctx != nullptr, "Failed to allocate ZSTD_CStream")) { + return; + } + { + auto err = ZSTD_CCtx_reset(ctx.get(), ZSTD_reset_session_only); + if (!errorHolder.check(!ZSTD_isError(err), ZSTD_getErrorName(err))) { + return; + } + } + + // Allocate space for the result + auto outBuffer = Buffer(ZSTD_compressBound(maxInputSize)); + auto zstdOutBuffer = makeZstdOutBuffer(outBuffer); + { + Buffer inBuffer; + // Read a buffer in from the input queue + while (in->pop(inBuffer) && !errorHolder.hasError()) { + auto zstdInBuffer = makeZstdInBuffer(inBuffer); + // Compress the whole buffer and send it to the output queue + while (!inBuffer.empty() && !errorHolder.hasError()) { + if (!errorHolder.check( + !outBuffer.empty(), "ZSTD_compressBound() was too small")) { + return; + } + // Compress + auto err = + ZSTD_compressStream(ctx.get(), &zstdOutBuffer, &zstdInBuffer); + if (!errorHolder.check(!ZSTD_isError(err), ZSTD_getErrorName(err))) { + return; + } + // Split the compressed data off outBuffer and pass to the output queue + out->push(split(outBuffer, zstdOutBuffer)); + // Forget about the data we already compressed + advance(inBuffer, zstdInBuffer); + } + } + } + // Write the epilog + size_t bytesLeft; + do { + if (!errorHolder.check( + !outBuffer.empty(), "ZSTD_compressBound() was too small")) { + return; + } + bytesLeft = ZSTD_endStream(ctx.get(), &zstdOutBuffer); + if (!errorHolder.check( + !ZSTD_isError(bytesLeft), ZSTD_getErrorName(bytesLeft))) { + return; + } + out->push(split(outBuffer, zstdOutBuffer)); + } while (bytesLeft != 0 && !errorHolder.hasError()); +} + +/** + * Calculates how large each independently compressed frame should be. + * + * @param size The size of the source if known, 0 otherwise + * @param numThreads The number of threads available to run compression jobs on + * @param params The zstd parameters to be used for compression + */ +static size_t calculateStep( + std::uintmax_t size, + size_t numThreads, + const ZSTD_parameters ¶ms) { + (void)size; + (void)numThreads; + // Not validated to work correctly for window logs > 23. + // It will definitely fail if windowLog + 2 is >= 4GB because + // the skippable frame can only store sizes up to 4GB. + assert(params.cParams.windowLog <= 23); + return size_t{1} << (params.cParams.windowLog + 2); +} + +namespace { +enum class FileStatus { Continue, Done, Error }; +/// Determines the status of the file descriptor `fd`. +FileStatus fileStatus(FILE* fd) { + if (std::feof(fd)) { + return FileStatus::Done; + } else if (std::ferror(fd)) { + return FileStatus::Error; + } + return FileStatus::Continue; +} +} // anonymous namespace + +/** + * Reads `size` data in chunks of `chunkSize` and puts it into `queue`. + * Will read less if an error or EOF occurs. + * Returns the status of the file after all of the reads have occurred. + */ +static FileStatus +readData(BufferWorkQueue& queue, size_t chunkSize, size_t size, FILE* fd, + std::uint64_t *totalBytesRead) { + Buffer buffer(size); + while (!buffer.empty()) { + auto bytesRead = + std::fread(buffer.data(), 1, std::min(chunkSize, buffer.size()), fd); + *totalBytesRead += bytesRead; + queue.push(buffer.splitAt(bytesRead)); + auto status = fileStatus(fd); + if (status != FileStatus::Continue) { + return status; + } + } + return FileStatus::Continue; +} + +std::uint64_t asyncCompressChunks( + SharedState& state, + WorkQueue<std::shared_ptr<BufferWorkQueue>>& chunks, + ThreadPool& executor, + FILE* fd, + std::uintmax_t size, + size_t numThreads, + ZSTD_parameters params) { + auto chunksGuard = makeScopeGuard([&] { chunks.finish(); }); + std::uint64_t bytesRead = 0; + + // Break the input up into chunks of size `step` and compress each chunk + // independently. + size_t step = calculateStep(size, numThreads, params); + state.log(kLogDebug, "Chosen frame size: %zu\n", step); + auto status = FileStatus::Continue; + while (status == FileStatus::Continue && !state.errorHolder.hasError()) { + // Make a new input queue that we will put the chunk's input data into. + auto in = std::make_shared<BufferWorkQueue>(); + auto inGuard = makeScopeGuard([&] { in->finish(); }); + // Make a new output queue that compress will put the compressed data into. + auto out = std::make_shared<BufferWorkQueue>(); + // Start compression in the thread pool + executor.add([&state, in, out, step] { + return compress( + state, std::move(in), std::move(out), step); + }); + // Pass the output queue to the writer thread. + chunks.push(std::move(out)); + state.log(kLogVerbose, "%s\n", "Starting a new frame"); + // Fill the input queue for the compression job we just started + status = readData(*in, ZSTD_CStreamInSize(), step, fd, &bytesRead); + } + state.errorHolder.check(status != FileStatus::Error, "Error reading input"); + return bytesRead; +} + +/** + * Decompress a frame, whose data is streamed into `in`, and stream the output + * to `out`. + * + * @param state The shared state + * @param in Queue that we `pop()` input buffers from. It contains + * exactly one compressed frame. + * @param out Queue that we `push()` decompressed output buffers to + */ +static void decompress( + SharedState& state, + std::shared_ptr<BufferWorkQueue> in, + std::shared_ptr<BufferWorkQueue> out) { + auto& errorHolder = state.errorHolder; + auto guard = makeScopeGuard([&] { out->finish(); }); + // Initialize the DCtx + auto ctx = state.dStreamPool->get(); + if (!errorHolder.check(ctx != nullptr, "Failed to allocate ZSTD_DStream")) { + return; + } + { + auto err = ZSTD_DCtx_reset(ctx.get(), ZSTD_reset_session_only); + if (!errorHolder.check(!ZSTD_isError(err), ZSTD_getErrorName(err))) { + return; + } + } + + const size_t outSize = ZSTD_DStreamOutSize(); + Buffer inBuffer; + size_t returnCode = 0; + // Read a buffer in from the input queue + while (in->pop(inBuffer) && !errorHolder.hasError()) { + auto zstdInBuffer = makeZstdInBuffer(inBuffer); + // Decompress the whole buffer and send it to the output queue + while (!inBuffer.empty() && !errorHolder.hasError()) { + // Allocate a buffer with at least outSize bytes. + Buffer outBuffer(outSize); + auto zstdOutBuffer = makeZstdOutBuffer(outBuffer); + // Decompress + returnCode = + ZSTD_decompressStream(ctx.get(), &zstdOutBuffer, &zstdInBuffer); + if (!errorHolder.check( + !ZSTD_isError(returnCode), ZSTD_getErrorName(returnCode))) { + return; + } + // Pass the buffer with the decompressed data to the output queue + out->push(split(outBuffer, zstdOutBuffer)); + // Advance past the input we already read + advance(inBuffer, zstdInBuffer); + if (returnCode == 0) { + // The frame is over, prepare to (maybe) start a new frame + ZSTD_initDStream(ctx.get()); + } + } + } + if (!errorHolder.check(returnCode <= 1, "Incomplete block")) { + return; + } + // We've given ZSTD_decompressStream all of our data, but there may still + // be data to read. + while (returnCode == 1) { + // Allocate a buffer with at least outSize bytes. + Buffer outBuffer(outSize); + auto zstdOutBuffer = makeZstdOutBuffer(outBuffer); + // Pass in no input. + ZSTD_inBuffer zstdInBuffer{nullptr, 0, 0}; + // Decompress + returnCode = + ZSTD_decompressStream(ctx.get(), &zstdOutBuffer, &zstdInBuffer); + if (!errorHolder.check( + !ZSTD_isError(returnCode), ZSTD_getErrorName(returnCode))) { + return; + } + // Pass the buffer with the decompressed data to the output queue + out->push(split(outBuffer, zstdOutBuffer)); + } +} + +std::uint64_t asyncDecompressFrames( + SharedState& state, + WorkQueue<std::shared_ptr<BufferWorkQueue>>& frames, + ThreadPool& executor, + FILE* fd) { + auto framesGuard = makeScopeGuard([&] { frames.finish(); }); + std::uint64_t totalBytesRead = 0; + + // Split the source up into its component frames. + // If we find our recognized skippable frame we know the next frames size + // which means that we can decompress each standard frame in independently. + // Otherwise, we will decompress using only one decompression task. + const size_t chunkSize = ZSTD_DStreamInSize(); + auto status = FileStatus::Continue; + while (status == FileStatus::Continue && !state.errorHolder.hasError()) { + // Make a new input queue that we will put the frames's bytes into. + auto in = std::make_shared<BufferWorkQueue>(); + auto inGuard = makeScopeGuard([&] { in->finish(); }); + // Make a output queue that decompress will put the decompressed data into + auto out = std::make_shared<BufferWorkQueue>(); + + size_t frameSize; + { + // Calculate the size of the next frame. + // frameSize is 0 if the frame info can't be decoded. + Buffer buffer(SkippableFrame::kSize); + auto bytesRead = std::fread(buffer.data(), 1, buffer.size(), fd); + totalBytesRead += bytesRead; + status = fileStatus(fd); + if (bytesRead == 0 && status != FileStatus::Continue) { + break; + } + buffer.subtract(buffer.size() - bytesRead); + frameSize = SkippableFrame::tryRead(buffer.range()); + in->push(std::move(buffer)); + } + if (frameSize == 0) { + // We hit a non SkippableFrame, so this will be the last job. + // Make sure that we don't use too much memory + in->setMaxSize(64); + out->setMaxSize(64); + } + // Start decompression in the thread pool + executor.add([&state, in, out] { + return decompress(state, std::move(in), std::move(out)); + }); + // Pass the output queue to the writer thread + frames.push(std::move(out)); + if (frameSize == 0) { + // We hit a non SkippableFrame ==> not compressed by pzstd or corrupted + // Pass the rest of the source to this decompression task + state.log(kLogVerbose, "%s\n", + "Input not in pzstd format, falling back to serial decompression"); + while (status == FileStatus::Continue && !state.errorHolder.hasError()) { + status = readData(*in, chunkSize, chunkSize, fd, &totalBytesRead); + } + break; + } + state.log(kLogVerbose, "Decompressing a frame of size %zu", frameSize); + // Fill the input queue for the decompression job we just started + status = readData(*in, chunkSize, frameSize, fd, &totalBytesRead); + } + state.errorHolder.check(status != FileStatus::Error, "Error reading input"); + return totalBytesRead; +} + +/// Write `data` to `fd`, returns true iff success. +static bool writeData(ByteRange data, FILE* fd) { + while (!data.empty()) { + data.advance(std::fwrite(data.begin(), 1, data.size(), fd)); + if (std::ferror(fd)) { + return false; + } + } + return true; +} + +std::uint64_t writeFile( + SharedState& state, + WorkQueue<std::shared_ptr<BufferWorkQueue>>& outs, + FILE* outputFd, + bool decompress) { + auto& errorHolder = state.errorHolder; + auto lineClearGuard = makeScopeGuard([&state] { + state.log.clear(kLogInfo); + }); + std::uint64_t bytesWritten = 0; + std::shared_ptr<BufferWorkQueue> out; + // Grab the output queue for each decompression job (in order). + while (outs.pop(out)) { + if (errorHolder.hasError()) { + continue; + } + if (!decompress) { + // If we are compressing and want to write skippable frames we can't + // start writing before compression is done because we need to know the + // compressed size. + // Wait for the compressed size to be available and write skippable frame + assert(uint64_t(out->size()) < uint64_t(1) << 32); + SkippableFrame frame(uint32_t(out->size())); + if (!writeData(frame.data(), outputFd)) { + errorHolder.setError("Failed to write output"); + return bytesWritten; + } + bytesWritten += frame.kSize; + } + // For each chunk of the frame: Pop it from the queue and write it + Buffer buffer; + while (out->pop(buffer) && !errorHolder.hasError()) { + if (!writeData(buffer.range(), outputFd)) { + errorHolder.setError("Failed to write output"); + return bytesWritten; + } + bytesWritten += buffer.size(); + state.log.update(kLogInfo, "Written: %u MB ", + static_cast<std::uint32_t>(bytesWritten >> 20)); + } + } + return bytesWritten; +} +} diff --git a/3rdparty/zstd/contrib/pzstd/Pzstd.h b/3rdparty/zstd/contrib/pzstd/Pzstd.h new file mode 100644 index 00000000000..3645e594268 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/Pzstd.h @@ -0,0 +1,153 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +#pragma once + +#include "ErrorHolder.h" +#include "Logging.h" +#include "Options.h" +#include "utils/Buffer.h" +#include "utils/Range.h" +#include "utils/ResourcePool.h" +#include "utils/ThreadPool.h" +#include "utils/WorkQueue.h" +#define ZSTD_STATIC_LINKING_ONLY +#define ZSTD_DISABLE_DEPRECATE_WARNINGS /* No deprecation warnings, pzstd itself is deprecated + * and uses deprecated functions + */ +#include "zstd.h" +#undef ZSTD_STATIC_LINKING_ONLY + +#include <cstddef> +#include <cstdint> +#include <memory> + +namespace pzstd { +/** + * Runs pzstd with `options` and returns the number of bytes written. + * An error occurred if `errorHandler.hasError()`. + * + * @param options The pzstd options to use for (de)compression + * @returns 0 upon success and non-zero on failure. + */ +int pzstdMain(const Options& options); + +class SharedState { + public: + SharedState(const Options& options) : log(options.verbosity) { + if (!options.decompress) { + auto parameters = options.determineParameters(); + cStreamPool.reset(new ResourcePool<ZSTD_CStream>{ + [this, parameters]() -> ZSTD_CStream* { + this->log(kLogVerbose, "%s\n", "Creating new ZSTD_CStream"); + auto zcs = ZSTD_createCStream(); + if (zcs) { + auto err = ZSTD_initCStream_advanced( + zcs, nullptr, 0, parameters, 0); + if (ZSTD_isError(err)) { + ZSTD_freeCStream(zcs); + return nullptr; + } + } + return zcs; + }, + [](ZSTD_CStream *zcs) { + ZSTD_freeCStream(zcs); + }}); + } else { + dStreamPool.reset(new ResourcePool<ZSTD_DStream>{ + [this]() -> ZSTD_DStream* { + this->log(kLogVerbose, "%s\n", "Creating new ZSTD_DStream"); + auto zds = ZSTD_createDStream(); + if (zds) { + auto err = ZSTD_initDStream(zds); + if (ZSTD_isError(err)) { + ZSTD_freeDStream(zds); + return nullptr; + } + } + return zds; + }, + [](ZSTD_DStream *zds) { + ZSTD_freeDStream(zds); + }}); + } + } + + ~SharedState() { + // The resource pools have references to this, so destroy them first. + cStreamPool.reset(); + dStreamPool.reset(); + } + + Logger log; + ErrorHolder errorHolder; + std::unique_ptr<ResourcePool<ZSTD_CStream>> cStreamPool; + std::unique_ptr<ResourcePool<ZSTD_DStream>> dStreamPool; +}; + +/** + * Streams input from `fd`, breaks input up into chunks, and compresses each + * chunk independently. Output of each chunk gets streamed to a queue, and + * the output queues get put into `chunks` in order. + * + * @param state The shared state + * @param chunks Each compression jobs output queue gets `pushed()` here + * as soon as it is available + * @param executor The thread pool to run compression jobs in + * @param fd The input file descriptor + * @param size The size of the input file if known, 0 otherwise + * @param numThreads The number of threads in the thread pool + * @param parameters The zstd parameters to use for compression + * @returns The number of bytes read from the file + */ +std::uint64_t asyncCompressChunks( + SharedState& state, + WorkQueue<std::shared_ptr<BufferWorkQueue>>& chunks, + ThreadPool& executor, + FILE* fd, + std::uintmax_t size, + std::size_t numThreads, + ZSTD_parameters parameters); + +/** + * Streams input from `fd`. If pzstd headers are available it breaks the input + * up into independent frames. It sends each frame to an independent + * decompression job. Output of each frame gets streamed to a queue, and + * the output queues get put into `frames` in order. + * + * @param state The shared state + * @param frames Each decompression jobs output queue gets `pushed()` here + * as soon as it is available + * @param executor The thread pool to run compression jobs in + * @param fd The input file descriptor + * @returns The number of bytes read from the file + */ +std::uint64_t asyncDecompressFrames( + SharedState& state, + WorkQueue<std::shared_ptr<BufferWorkQueue>>& frames, + ThreadPool& executor, + FILE* fd); + +/** + * Streams input in from each queue in `outs` in order, and writes the data to + * `outputFd`. + * + * @param state The shared state + * @param outs A queue of output queues, one for each + * (de)compression job. + * @param outputFd The file descriptor to write to + * @param decompress Are we decompressing? + * @returns The number of bytes written + */ +std::uint64_t writeFile( + SharedState& state, + WorkQueue<std::shared_ptr<BufferWorkQueue>>& outs, + FILE* outputFd, + bool decompress); +} diff --git a/3rdparty/zstd/contrib/pzstd/README.md b/3rdparty/zstd/contrib/pzstd/README.md new file mode 100644 index 00000000000..bc8f831f82c --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/README.md @@ -0,0 +1,56 @@ +# Parallel Zstandard (PZstandard) + +Parallel Zstandard is a Pigz-like tool for Zstandard. +It provides Zstandard format compatible compression and decompression that is able to utilize multiple cores. +It breaks the input up into equal sized chunks and compresses each chunk independently into a Zstandard frame. +It then concatenates the frames together to produce the final compressed output. +Pzstandard will write a 12 byte header for each frame that is a skippable frame in the Zstandard format, which tells PZstandard the size of the next compressed frame. +PZstandard supports parallel decompression of files compressed with PZstandard. +When decompressing files compressed with Zstandard, PZstandard does IO in one thread, and decompression in another. + +## Usage + +PZstandard supports the same command line interface as Zstandard, but also provides the `-p` option to specify the number of threads. +Dictionary mode is not currently supported. + +Basic usage + + pzstd input-file -o output-file -p num-threads -# # Compression + pzstd -d input-file -o output-file -p num-threads # Decompression + +PZstandard also supports piping and fifo pipes + + cat input-file | pzstd -p num-threads -# -c > /dev/null + +For more options + + pzstd --help + +PZstandard tries to pick a smart default number of threads if not specified (displayed in `pzstd --help`). +If this number is not suitable, during compilation you can define `PZSTD_NUM_THREADS` to the number of threads you prefer. + +## Benchmarks + +As a reference, PZstandard and Pigz were compared on an Intel Core i7 @ 3.1 GHz, each using 4 threads, with the [Silesia compression corpus](https://sun.aei.polsl.pl//~sdeor/index.php?page=silesia). + +Compression Speed vs Ratio with 4 Threads | Decompression Speed with 4 Threads +------------------------------------------|----------------------------------- + |  + +The test procedure was to run each of the following commands 2 times for each compression level, and take the minimum time. + + time pzstd -# -p 4 -c silesia.tar > silesia.tar.zst + time pzstd -d -p 4 -c silesia.tar.zst > /dev/null + + time pigz -# -p 4 -k -c silesia.tar > silesia.tar.gz + time pigz -d -p 4 -k -c silesia.tar.gz > /dev/null + +PZstandard was tested using compression levels 1-19, and Pigz was tested using compression levels 1-9. +Pigz cannot do parallel decompression, it simply does each of reading, decompression, and writing on separate threads. + +## Tests + +Tests require that you have [gtest](https://github.com/google/googletest) installed. +Set `GTEST_INC` and `GTEST_LIB` in `Makefile` to specify the location of the gtest headers and libraries. +Alternatively, run `make googletest`, which will clone googletest and build it. +Run `make tests && make check` to run tests. diff --git a/3rdparty/zstd/contrib/pzstd/SkippableFrame.cpp b/3rdparty/zstd/contrib/pzstd/SkippableFrame.cpp new file mode 100644 index 00000000000..3bea4eb65bf --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/SkippableFrame.cpp @@ -0,0 +1,30 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +#include "SkippableFrame.h" +#include "mem.h" +#include "utils/Range.h" + +#include <cstdio> + +using namespace pzstd; + +SkippableFrame::SkippableFrame(std::uint32_t size) : frameSize_(size) { + MEM_writeLE32(data_.data(), kSkippableFrameMagicNumber); + MEM_writeLE32(data_.data() + 4, kFrameContentsSize); + MEM_writeLE32(data_.data() + 8, frameSize_); +} + +/* static */ std::size_t SkippableFrame::tryRead(ByteRange bytes) { + if (bytes.size() < SkippableFrame::kSize || + MEM_readLE32(bytes.begin()) != kSkippableFrameMagicNumber || + MEM_readLE32(bytes.begin() + 4) != kFrameContentsSize) { + return 0; + } + return MEM_readLE32(bytes.begin() + 8); +} diff --git a/3rdparty/zstd/contrib/pzstd/SkippableFrame.h b/3rdparty/zstd/contrib/pzstd/SkippableFrame.h new file mode 100644 index 00000000000..817415e9231 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/SkippableFrame.h @@ -0,0 +1,64 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +#pragma once + +#include "utils/Range.h" + +#include <array> +#include <cstddef> +#include <cstdint> +#include <cstdio> + +namespace pzstd { +/** + * We put a skippable frame before each frame. + * It contains a skippable frame magic number, the size of the skippable frame, + * and the size of the next frame. + * Each skippable frame is exactly 12 bytes in little endian format. + * The first 8 bytes are for compatibility with the ZSTD format. + * If we have N threads, the output will look like + * + * [0x184D2A50|4|size1] [frame1 of size size1] + * [0x184D2A50|4|size2] [frame2 of size size2] + * ... + * [0x184D2A50|4|sizeN] [frameN of size sizeN] + * + * Each sizeX is 4 bytes. + * + * These skippable frames should allow us to skip through the compressed file + * and only load at most N pages. + */ +class SkippableFrame { + public: + static constexpr std::size_t kSize = 12; + + private: + std::uint32_t frameSize_; + std::array<std::uint8_t, kSize> data_; + static constexpr std::uint32_t kSkippableFrameMagicNumber = 0x184D2A50; + // Could be improved if the size fits in less bytes + static constexpr std::uint32_t kFrameContentsSize = kSize - 8; + + public: + // Write the skippable frame to data_ in LE format. + explicit SkippableFrame(std::uint32_t size); + + // Read the skippable frame from bytes in LE format. + static std::size_t tryRead(ByteRange bytes); + + ByteRange data() const { + return {data_.data(), data_.size()}; + } + + // Size of the next frame. + std::size_t frameSize() const { + return frameSize_; + } +}; +} diff --git a/3rdparty/zstd/contrib/pzstd/images/Cspeed.png b/3rdparty/zstd/contrib/pzstd/images/Cspeed.png Binary files differnew file mode 100644 index 00000000000..aca4f663ea2 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/images/Cspeed.png diff --git a/3rdparty/zstd/contrib/pzstd/images/Dspeed.png b/3rdparty/zstd/contrib/pzstd/images/Dspeed.png Binary files differnew file mode 100644 index 00000000000..e48881bcd05 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/images/Dspeed.png diff --git a/3rdparty/zstd/contrib/pzstd/main.cpp b/3rdparty/zstd/contrib/pzstd/main.cpp new file mode 100644 index 00000000000..422b4a56a71 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/main.cpp @@ -0,0 +1,27 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +#include "ErrorHolder.h" +#include "Options.h" +#include "Pzstd.h" + +using namespace pzstd; + +int main(int argc, const char** argv) { + Options options; + switch (options.parse(argc, argv)) { + case Options::Status::Failure: + return 1; + case Options::Status::Message: + return 0; + default: + break; + } + + return pzstdMain(options); +} diff --git a/3rdparty/zstd/contrib/pzstd/test/BUCK b/3rdparty/zstd/contrib/pzstd/test/BUCK new file mode 100644 index 00000000000..6d3fdd3c269 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/test/BUCK @@ -0,0 +1,37 @@ +cxx_test( + name='options_test', + srcs=['OptionsTest.cpp'], + deps=['//contrib/pzstd:options'], +) + +cxx_test( + name='pzstd_test', + srcs=['PzstdTest.cpp'], + deps=[ + ':round_trip', + '//contrib/pzstd:libpzstd', + '//contrib/pzstd/utils:scope_guard', + '//programs:datagen', + ], +) + +cxx_binary( + name='round_trip_test', + srcs=['RoundTripTest.cpp'], + deps=[ + ':round_trip', + '//contrib/pzstd/utils:scope_guard', + '//programs:datagen', + ] +) + +cxx_library( + name='round_trip', + header_namespace='test', + exported_headers=['RoundTrip.h'], + deps=[ + '//contrib/pzstd:libpzstd', + '//contrib/pzstd:options', + '//contrib/pzstd/utils:scope_guard', + ] +) diff --git a/3rdparty/zstd/contrib/pzstd/test/OptionsTest.cpp b/3rdparty/zstd/contrib/pzstd/test/OptionsTest.cpp new file mode 100644 index 00000000000..91e39750d0c --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/test/OptionsTest.cpp @@ -0,0 +1,536 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +#include "Options.h" + +#include <array> +#include <gtest/gtest.h> + +using namespace pzstd; + +namespace pzstd { +bool operator==(const Options &lhs, const Options &rhs) { + return lhs.numThreads == rhs.numThreads && + lhs.maxWindowLog == rhs.maxWindowLog && + lhs.compressionLevel == rhs.compressionLevel && + lhs.decompress == rhs.decompress && lhs.inputFiles == rhs.inputFiles && + lhs.outputFile == rhs.outputFile && lhs.overwrite == rhs.overwrite && + lhs.keepSource == rhs.keepSource && lhs.writeMode == rhs.writeMode && + lhs.checksum == rhs.checksum && lhs.verbosity == rhs.verbosity; +} + +std::ostream &operator<<(std::ostream &out, const Options &opt) { + out << "{"; + { + out << "\n\t" + << "numThreads: " << opt.numThreads; + out << ",\n\t" + << "maxWindowLog: " << opt.maxWindowLog; + out << ",\n\t" + << "compressionLevel: " << opt.compressionLevel; + out << ",\n\t" + << "decompress: " << opt.decompress; + out << ",\n\t" + << "inputFiles: {"; + { + bool first = true; + for (const auto &file : opt.inputFiles) { + if (!first) { + out << ","; + } + first = false; + out << "\n\t\t" << file; + } + } + out << "\n\t}"; + out << ",\n\t" + << "outputFile: " << opt.outputFile; + out << ",\n\t" + << "overwrite: " << opt.overwrite; + out << ",\n\t" + << "keepSource: " << opt.keepSource; + out << ",\n\t" + << "writeMode: " << static_cast<int>(opt.writeMode); + out << ",\n\t" + << "checksum: " << opt.checksum; + out << ",\n\t" + << "verbosity: " << opt.verbosity; + } + out << "\n}"; + return out; +} +} + +namespace { +#ifdef _WIN32 +const char nullOutput[] = "nul"; +#else +const char nullOutput[] = "/dev/null"; +#endif + +constexpr auto autoMode = Options::WriteMode::Auto; +} // anonymous namespace + +#define EXPECT_SUCCESS(...) EXPECT_EQ(Options::Status::Success, __VA_ARGS__) +#define EXPECT_FAILURE(...) EXPECT_EQ(Options::Status::Failure, __VA_ARGS__) +#define EXPECT_MESSAGE(...) EXPECT_EQ(Options::Status::Message, __VA_ARGS__) + +template <typename... Args> +std::array<const char *, sizeof...(Args) + 1> makeArray(Args... args) { + return {{nullptr, args...}}; +} + +TEST(Options, ValidInputs) { + { + Options options; + auto args = makeArray("--processes", "5", "-o", "x", "y", "-f"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + Options expected = {5, 23, 3, false, {"y"}, "x", + true, true, autoMode, true, 2}; + EXPECT_EQ(expected, options); + } + { + Options options; + auto args = makeArray("-p", "1", "input", "-19"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + Options expected = {1, 23, 19, false, {"input"}, "", + false, true, autoMode, true, 2}; + EXPECT_EQ(expected, options); + } + { + Options options; + auto args = + makeArray("--ultra", "-22", "-p", "1", "-o", "x", "-d", "x.zst", "-f"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + Options expected = {1, 0, 22, true, {"x.zst"}, "x", + true, true, autoMode, true, 2}; + EXPECT_EQ(expected, options); + } + { + Options options; + auto args = makeArray("--processes", "100", "hello.zst", "--decompress", + "--force"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + Options expected = {100, 23, 3, true, {"hello.zst"}, "", true, + true, autoMode, true, 2}; + EXPECT_EQ(expected, options); + } + { + Options options; + auto args = makeArray("x", "-dp", "1", "-c"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + Options expected = {1, 23, 3, true, {"x"}, "-", + false, true, autoMode, true, 2}; + EXPECT_EQ(expected, options); + } + { + Options options; + auto args = makeArray("x", "-dp", "1", "--stdout"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + Options expected = {1, 23, 3, true, {"x"}, "-", + false, true, autoMode, true, 2}; + EXPECT_EQ(expected, options); + } + { + Options options; + auto args = makeArray("-p", "1", "x", "-5", "-fo", "-", "--ultra", "-d"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + Options expected = {1, 0, 5, true, {"x"}, "-", + true, true, autoMode, true, 2}; + EXPECT_EQ(expected, options); + } + { + Options options; + auto args = makeArray("silesia.tar", "-o", "silesia.tar.pzstd", "-p", "2"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + Options expected = {2, + 23, + 3, + false, + {"silesia.tar"}, + "silesia.tar.pzstd", + false, + true, + autoMode, + true, + 2}; + EXPECT_EQ(expected, options); + } + { + Options options; + auto args = makeArray("x", "-p", "1"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + } + { + Options options; + auto args = makeArray("x", "-p", "1"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + } +} + +TEST(Options, GetOutputFile) { + { + Options options; + auto args = makeArray("x"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + EXPECT_EQ("x.zst", options.getOutputFile(options.inputFiles[0])); + } + { + Options options; + auto args = makeArray("x", "y", "-o", nullOutput); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + EXPECT_EQ(nullOutput, options.getOutputFile(options.inputFiles[0])); + } + { + Options options; + auto args = makeArray("x.zst", "-do", nullOutput); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + EXPECT_EQ(nullOutput, options.getOutputFile(options.inputFiles[0])); + } + { + Options options; + auto args = makeArray("x.zst", "-d"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + EXPECT_EQ("x", options.getOutputFile(options.inputFiles[0])); + } + { + Options options; + auto args = makeArray("xzst", "-d"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + EXPECT_EQ("", options.getOutputFile(options.inputFiles[0])); + } + { + Options options; + auto args = makeArray("xzst", "-doxx"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + EXPECT_EQ("xx", options.getOutputFile(options.inputFiles[0])); + } +} + +TEST(Options, MultipleFiles) { + { + Options options; + auto args = makeArray("x", "y", "z"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + Options expected; + expected.inputFiles = {"x", "y", "z"}; + expected.verbosity = 1; + EXPECT_EQ(expected, options); + } + { + Options options; + auto args = makeArray("x", "y", "z", "-o", nullOutput); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + Options expected; + expected.inputFiles = {"x", "y", "z"}; + expected.outputFile = nullOutput; + expected.verbosity = 1; + EXPECT_EQ(expected, options); + } + { + Options options; + auto args = makeArray("x", "y", "-o-"); + EXPECT_FAILURE(options.parse(args.size(), args.data())); + } + { + Options options; + auto args = makeArray("x", "y", "-o", "file"); + EXPECT_FAILURE(options.parse(args.size(), args.data())); + } + { + Options options; + auto args = makeArray("-qqvd12qp4", "-f", "x", "--", "--rm", "-c"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + Options expected = {4, 23, 12, true, {"x", "--rm", "-c"}, + "", true, true, autoMode, true, + 0}; + EXPECT_EQ(expected, options); + } +} + +TEST(Options, NumThreads) { + { + Options options; + auto args = makeArray("x", "-dfo", "-"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + } + { + Options options; + auto args = makeArray("x", "-p", "0", "-fo", "-"); + EXPECT_FAILURE(options.parse(args.size(), args.data())); + } + { + Options options; + auto args = makeArray("-f", "-p", "-o", "-"); + EXPECT_FAILURE(options.parse(args.size(), args.data())); + } +} + +TEST(Options, BadCompressionLevel) { + { + Options options; + auto args = makeArray("x", "-20"); + EXPECT_FAILURE(options.parse(args.size(), args.data())); + } + { + Options options; + auto args = makeArray("x", "--ultra", "-23"); + EXPECT_FAILURE(options.parse(args.size(), args.data())); + } + { + Options options; + auto args = makeArray("x", "--1"); // negative 1? + EXPECT_FAILURE(options.parse(args.size(), args.data())); + } +} + +TEST(Options, InvalidOption) { + { + Options options; + auto args = makeArray("x", "-x"); + EXPECT_FAILURE(options.parse(args.size(), args.data())); + } +} + +TEST(Options, BadOutputFile) { + { + Options options; + auto args = makeArray("notzst", "-d", "-p", "1"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + EXPECT_EQ("", options.getOutputFile(options.inputFiles.front())); + } +} + +TEST(Options, BadOptionsWithArguments) { + { + Options options; + auto args = makeArray("x", "-pf"); + EXPECT_FAILURE(options.parse(args.size(), args.data())); + } + { + Options options; + auto args = makeArray("x", "-p", "10f"); + EXPECT_FAILURE(options.parse(args.size(), args.data())); + } + { + Options options; + auto args = makeArray("x", "-p"); + EXPECT_FAILURE(options.parse(args.size(), args.data())); + } + { + Options options; + auto args = makeArray("x", "-o"); + EXPECT_FAILURE(options.parse(args.size(), args.data())); + } + { + Options options; + auto args = makeArray("x", "-o"); + EXPECT_FAILURE(options.parse(args.size(), args.data())); + } +} + +TEST(Options, KeepSource) { + { + Options options; + auto args = makeArray("x", "--rm", "-k"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + EXPECT_EQ(true, options.keepSource); + } + { + Options options; + auto args = makeArray("x", "--rm", "--keep"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + EXPECT_EQ(true, options.keepSource); + } + { + Options options; + auto args = makeArray("x"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + EXPECT_EQ(true, options.keepSource); + } + { + Options options; + auto args = makeArray("x", "--rm"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + EXPECT_EQ(false, options.keepSource); + } +} + +TEST(Options, Verbosity) { + { + Options options; + auto args = makeArray("x"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + EXPECT_EQ(2, options.verbosity); + } + { + Options options; + auto args = makeArray("--quiet", "-qq", "x"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + EXPECT_EQ(-1, options.verbosity); + } + { + Options options; + auto args = makeArray("x", "y"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + EXPECT_EQ(1, options.verbosity); + } + { + Options options; + auto args = makeArray("--", "x", "y"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + EXPECT_EQ(1, options.verbosity); + } + { + Options options; + auto args = makeArray("-qv", "x", "y"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + EXPECT_EQ(1, options.verbosity); + } + { + Options options; + auto args = makeArray("-v", "x", "y"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + EXPECT_EQ(3, options.verbosity); + } + { + Options options; + auto args = makeArray("-v", "x"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + EXPECT_EQ(3, options.verbosity); + } +} + +TEST(Options, TestMode) { + { + Options options; + auto args = makeArray("x", "-t"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + EXPECT_EQ(true, options.keepSource); + EXPECT_EQ(true, options.decompress); + EXPECT_EQ(nullOutput, options.outputFile); + } + { + Options options; + auto args = makeArray("x", "--test", "--rm", "-ohello"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + EXPECT_EQ(true, options.keepSource); + EXPECT_EQ(true, options.decompress); + EXPECT_EQ(nullOutput, options.outputFile); + } +} + +TEST(Options, Checksum) { + { + Options options; + auto args = makeArray("x.zst", "--no-check", "-Cd"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + EXPECT_EQ(true, options.checksum); + } + { + Options options; + auto args = makeArray("x"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + EXPECT_EQ(true, options.checksum); + } + { + Options options; + auto args = makeArray("x", "--no-check", "--check"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + EXPECT_EQ(true, options.checksum); + } + { + Options options; + auto args = makeArray("x", "--no-check"); + EXPECT_SUCCESS(options.parse(args.size(), args.data())); + EXPECT_EQ(false, options.checksum); + } +} + +TEST(Options, InputFiles) { + { + Options options; + auto args = makeArray("-cd"); + options.parse(args.size(), args.data()); + EXPECT_EQ(1, options.inputFiles.size()); + EXPECT_EQ("-", options.inputFiles[0]); + EXPECT_EQ("-", options.outputFile); + } + { + Options options; + auto args = makeArray(); + options.parse(args.size(), args.data()); + EXPECT_EQ(1, options.inputFiles.size()); + EXPECT_EQ("-", options.inputFiles[0]); + EXPECT_EQ("-", options.outputFile); + } + { + Options options; + auto args = makeArray("-d"); + options.parse(args.size(), args.data()); + EXPECT_EQ(1, options.inputFiles.size()); + EXPECT_EQ("-", options.inputFiles[0]); + EXPECT_EQ("-", options.outputFile); + } + { + Options options; + auto args = makeArray("x", "-"); + EXPECT_FAILURE(options.parse(args.size(), args.data())); + } +} + +TEST(Options, InvalidOptions) { + { + Options options; + auto args = makeArray("-ibasdf"); + EXPECT_FAILURE(options.parse(args.size(), args.data())); + } + { + Options options; + auto args = makeArray("- "); + EXPECT_FAILURE(options.parse(args.size(), args.data())); + } + { + Options options; + auto args = makeArray("-n15"); + EXPECT_FAILURE(options.parse(args.size(), args.data())); + } + { + Options options; + auto args = makeArray("-0", "x"); + EXPECT_FAILURE(options.parse(args.size(), args.data())); + } +} + +TEST(Options, Extras) { + { + Options options; + auto args = makeArray("-h"); + EXPECT_MESSAGE(options.parse(args.size(), args.data())); + } + { + Options options; + auto args = makeArray("-H"); + EXPECT_MESSAGE(options.parse(args.size(), args.data())); + } + { + Options options; + auto args = makeArray("-V"); + EXPECT_MESSAGE(options.parse(args.size(), args.data())); + } + { + Options options; + auto args = makeArray("--help"); + EXPECT_MESSAGE(options.parse(args.size(), args.data())); + } + { + Options options; + auto args = makeArray("--version"); + EXPECT_MESSAGE(options.parse(args.size(), args.data())); + } +} diff --git a/3rdparty/zstd/contrib/pzstd/test/PzstdTest.cpp b/3rdparty/zstd/contrib/pzstd/test/PzstdTest.cpp new file mode 100644 index 00000000000..75453f5f6f3 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/test/PzstdTest.cpp @@ -0,0 +1,149 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +#include "Pzstd.h" +extern "C" { +#include "datagen.h" +} +#include "test/RoundTrip.h" +#include "utils/ScopeGuard.h" + +#include <cstddef> +#include <cstdio> +#include <gtest/gtest.h> +#include <memory> +#include <random> + +using namespace std; +using namespace pzstd; + +TEST(Pzstd, SmallSizes) { + unsigned seed = std::random_device{}(); + std::fprintf(stderr, "Pzstd.SmallSizes seed: %u\n", seed); + std::mt19937 gen(seed); + + for (unsigned len = 1; len < 256; ++len) { + if (len % 16 == 0) { + std::fprintf(stderr, "%u / 16\n", len / 16); + } + std::string inputFile = std::tmpnam(nullptr); + auto guard = makeScopeGuard([&] { std::remove(inputFile.c_str()); }); + { + static uint8_t buf[256]; + RDG_genBuffer(buf, len, 0.5, 0.0, gen()); + auto fd = std::fopen(inputFile.c_str(), "wb"); + auto written = std::fwrite(buf, 1, len, fd); + std::fclose(fd); + ASSERT_EQ(written, len); + } + for (unsigned numThreads = 1; numThreads <= 2; ++numThreads) { + for (unsigned level = 1; level <= 4; level *= 4) { + auto errorGuard = makeScopeGuard([&] { + std::fprintf(stderr, "# threads: %u\n", numThreads); + std::fprintf(stderr, "compression level: %u\n", level); + }); + Options options; + options.overwrite = true; + options.inputFiles = {inputFile}; + options.numThreads = numThreads; + options.compressionLevel = level; + options.verbosity = 1; + ASSERT_TRUE(roundTrip(options)); + errorGuard.dismiss(); + } + } + } +} + +TEST(Pzstd, LargeSizes) { + unsigned seed = std::random_device{}(); + std::fprintf(stderr, "Pzstd.LargeSizes seed: %u\n", seed); + std::mt19937 gen(seed); + + for (unsigned len = 1 << 20; len <= (1 << 24); len *= 2) { + std::string inputFile = std::tmpnam(nullptr); + auto guard = makeScopeGuard([&] { std::remove(inputFile.c_str()); }); + { + std::unique_ptr<uint8_t[]> buf(new uint8_t[len]); + RDG_genBuffer(buf.get(), len, 0.5, 0.0, gen()); + auto fd = std::fopen(inputFile.c_str(), "wb"); + auto written = std::fwrite(buf.get(), 1, len, fd); + std::fclose(fd); + ASSERT_EQ(written, len); + } + for (unsigned numThreads = 1; numThreads <= 16; numThreads *= 4) { + for (unsigned level = 1; level <= 4; level *= 4) { + auto errorGuard = makeScopeGuard([&] { + std::fprintf(stderr, "# threads: %u\n", numThreads); + std::fprintf(stderr, "compression level: %u\n", level); + }); + Options options; + options.overwrite = true; + options.inputFiles = {inputFile}; + options.numThreads = std::min(numThreads, options.numThreads); + options.compressionLevel = level; + options.verbosity = 1; + ASSERT_TRUE(roundTrip(options)); + errorGuard.dismiss(); + } + } + } +} + +TEST(Pzstd, DISABLED_ExtremelyLargeSize) { + unsigned seed = std::random_device{}(); + std::fprintf(stderr, "Pzstd.ExtremelyLargeSize seed: %u\n", seed); + std::mt19937 gen(seed); + + std::string inputFile = std::tmpnam(nullptr); + auto guard = makeScopeGuard([&] { std::remove(inputFile.c_str()); }); + + { + // Write 4GB + 64 MB + constexpr size_t kLength = 1 << 26; + std::unique_ptr<uint8_t[]> buf(new uint8_t[kLength]); + auto fd = std::fopen(inputFile.c_str(), "wb"); + auto closeGuard = makeScopeGuard([&] { std::fclose(fd); }); + for (size_t i = 0; i < (1 << 6) + 1; ++i) { + RDG_genBuffer(buf.get(), kLength, 0.5, 0.0, gen()); + auto written = std::fwrite(buf.get(), 1, kLength, fd); + if (written != kLength) { + std::fprintf(stderr, "Failed to write file, skipping test\n"); + return; + } + } + } + + Options options; + options.overwrite = true; + options.inputFiles = {inputFile}; + options.compressionLevel = 1; + if (options.numThreads == 0) { + options.numThreads = 1; + } + ASSERT_TRUE(roundTrip(options)); +} + +TEST(Pzstd, ExtremelyCompressible) { + std::string inputFile = std::tmpnam(nullptr); + auto guard = makeScopeGuard([&] { std::remove(inputFile.c_str()); }); + { + std::unique_ptr<uint8_t[]> buf(new uint8_t[10000]); + std::memset(buf.get(), 'a', 10000); + auto fd = std::fopen(inputFile.c_str(), "wb"); + auto written = std::fwrite(buf.get(), 1, 10000, fd); + std::fclose(fd); + ASSERT_EQ(written, 10000); + } + Options options; + options.overwrite = true; + options.inputFiles = {inputFile}; + options.numThreads = 1; + options.compressionLevel = 1; + ASSERT_TRUE(roundTrip(options)); +} diff --git a/3rdparty/zstd/contrib/pzstd/test/RoundTrip.h b/3rdparty/zstd/contrib/pzstd/test/RoundTrip.h new file mode 100644 index 00000000000..f777622a393 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/test/RoundTrip.h @@ -0,0 +1,86 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +#pragma once + +#include "Options.h" +#include "Pzstd.h" +#include "utils/ScopeGuard.h" + +#include <cstdio> +#include <string> +#include <cstdint> +#include <memory> + +namespace pzstd { + +inline bool check(std::string source, std::string decompressed) { + std::unique_ptr<std::uint8_t[]> sBuf(new std::uint8_t[1024]); + std::unique_ptr<std::uint8_t[]> dBuf(new std::uint8_t[1024]); + + auto sFd = std::fopen(source.c_str(), "rb"); + auto dFd = std::fopen(decompressed.c_str(), "rb"); + auto guard = makeScopeGuard([&] { + std::fclose(sFd); + std::fclose(dFd); + }); + + size_t sRead, dRead; + + do { + sRead = std::fread(sBuf.get(), 1, 1024, sFd); + dRead = std::fread(dBuf.get(), 1, 1024, dFd); + if (std::ferror(sFd) || std::ferror(dFd)) { + return false; + } + if (sRead != dRead) { + return false; + } + + for (size_t i = 0; i < sRead; ++i) { + if (sBuf.get()[i] != dBuf.get()[i]) { + return false; + } + } + } while (sRead == 1024); + if (!std::feof(sFd) || !std::feof(dFd)) { + return false; + } + return true; +} + +inline bool roundTrip(Options& options) { + if (options.inputFiles.size() != 1) { + return false; + } + std::string source = options.inputFiles.front(); + std::string compressedFile = std::tmpnam(nullptr); + std::string decompressedFile = std::tmpnam(nullptr); + auto guard = makeScopeGuard([&] { + std::remove(compressedFile.c_str()); + std::remove(decompressedFile.c_str()); + }); + + { + options.outputFile = compressedFile; + options.decompress = false; + if (pzstdMain(options) != 0) { + return false; + } + } + { + options.decompress = true; + options.inputFiles.front() = compressedFile; + options.outputFile = decompressedFile; + if (pzstdMain(options) != 0) { + return false; + } + } + return check(source, decompressedFile); +} +} diff --git a/3rdparty/zstd/contrib/pzstd/test/RoundTripTest.cpp b/3rdparty/zstd/contrib/pzstd/test/RoundTripTest.cpp new file mode 100644 index 00000000000..c37646d1c44 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/test/RoundTripTest.cpp @@ -0,0 +1,86 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +extern "C" { +#include "datagen.h" +} +#include "Options.h" +#include "test/RoundTrip.h" +#include "utils/ScopeGuard.h" + +#include <cstddef> +#include <cstdio> +#include <cstdlib> +#include <memory> +#include <random> + +using namespace std; +using namespace pzstd; + +namespace { +string +writeData(size_t size, double matchProba, double litProba, unsigned seed) { + std::unique_ptr<uint8_t[]> buf(new uint8_t[size]); + RDG_genBuffer(buf.get(), size, matchProba, litProba, seed); + string file = tmpnam(nullptr); + auto fd = std::fopen(file.c_str(), "wb"); + auto guard = makeScopeGuard([&] { std::fclose(fd); }); + auto bytesWritten = std::fwrite(buf.get(), 1, size, fd); + if (bytesWritten != size) { + std::abort(); + } + return file; +} + +template <typename Generator> +string generateInputFile(Generator& gen) { + // Use inputs ranging from 1 Byte to 2^16 Bytes + std::uniform_int_distribution<size_t> size{1, 1 << 16}; + std::uniform_real_distribution<> prob{0, 1}; + return writeData(size(gen), prob(gen), prob(gen), gen()); +} + +template <typename Generator> +Options generateOptions(Generator& gen, const string& inputFile) { + Options options; + options.inputFiles = {inputFile}; + options.overwrite = true; + + std::uniform_int_distribution<unsigned> numThreads{1, 32}; + std::uniform_int_distribution<unsigned> compressionLevel{1, 10}; + + options.numThreads = numThreads(gen); + options.compressionLevel = compressionLevel(gen); + + return options; +} +} + +int main() { + std::mt19937 gen(std::random_device{}()); + + auto newlineGuard = makeScopeGuard([] { std::fprintf(stderr, "\n"); }); + for (unsigned i = 0; i < 10000; ++i) { + if (i % 100 == 0) { + std::fprintf(stderr, "Progress: %u%%\r", i / 100); + } + auto inputFile = generateInputFile(gen); + auto inputGuard = makeScopeGuard([&] { std::remove(inputFile.c_str()); }); + for (unsigned i = 0; i < 10; ++i) { + auto options = generateOptions(gen, inputFile); + if (!roundTrip(options)) { + std::fprintf(stderr, "numThreads: %u\n", options.numThreads); + std::fprintf(stderr, "level: %u\n", options.compressionLevel); + std::fprintf(stderr, "decompress? %u\n", (unsigned)options.decompress); + std::fprintf(stderr, "file: %s\n", inputFile.c_str()); + return 1; + } + } + } + return 0; +} diff --git a/3rdparty/zstd/contrib/pzstd/utils/BUCK b/3rdparty/zstd/contrib/pzstd/utils/BUCK new file mode 100644 index 00000000000..e757f412070 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/utils/BUCK @@ -0,0 +1,75 @@ +cxx_library( + name='buffer', + visibility=['PUBLIC'], + header_namespace='utils', + exported_headers=['Buffer.h'], + deps=[':range'], +) + +cxx_library( + name='file_system', + visibility=['PUBLIC'], + header_namespace='utils', + exported_headers=['FileSystem.h'], + deps=[':range'], +) + +cxx_library( + name='likely', + visibility=['PUBLIC'], + header_namespace='utils', + exported_headers=['Likely.h'], +) + +cxx_library( + name='range', + visibility=['PUBLIC'], + header_namespace='utils', + exported_headers=['Range.h'], + deps=[':likely'], +) + +cxx_library( + name='resource_pool', + visibility=['PUBLIC'], + header_namespace='utils', + exported_headers=['ResourcePool.h'], +) + +cxx_library( + name='scope_guard', + visibility=['PUBLIC'], + header_namespace='utils', + exported_headers=['ScopeGuard.h'], +) + +cxx_library( + name='thread_pool', + visibility=['PUBLIC'], + header_namespace='utils', + exported_headers=['ThreadPool.h'], + deps=[':work_queue'], +) + +cxx_library( + name='work_queue', + visibility=['PUBLIC'], + header_namespace='utils', + exported_headers=['WorkQueue.h'], + deps=[':buffer'], +) + +cxx_library( + name='utils', + visibility=['PUBLIC'], + deps=[ + ':buffer', + ':file_system', + ':likely', + ':range', + ':resource_pool', + ':scope_guard', + ':thread_pool', + ':work_queue', + ], +) diff --git a/3rdparty/zstd/contrib/pzstd/utils/Buffer.h b/3rdparty/zstd/contrib/pzstd/utils/Buffer.h new file mode 100644 index 00000000000..a85f770ba18 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/utils/Buffer.h @@ -0,0 +1,99 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +#pragma once + +#include "utils/Range.h" + +#include <array> +#include <cstddef> +#include <memory> + +namespace pzstd { + +/** + * A `Buffer` has a pointer to a shared buffer, and a range of the buffer that + * it owns. + * The idea is that you can allocate one buffer, and write chunks into it + * and break off those chunks. + * The underlying buffer is reference counted, and will be destroyed when all + * `Buffer`s that reference it are destroyed. + */ +class Buffer { + std::shared_ptr<unsigned char> buffer_; + MutableByteRange range_; + + static void delete_buffer(unsigned char* buffer) { + delete[] buffer; + } + + public: + /// Construct an empty buffer that owns no data. + explicit Buffer() {} + + /// Construct a `Buffer` that owns a new underlying buffer of size `size`. + explicit Buffer(std::size_t size) + : buffer_(new unsigned char[size], delete_buffer), + range_(buffer_.get(), buffer_.get() + size) {} + + explicit Buffer(std::shared_ptr<unsigned char> buffer, MutableByteRange data) + : buffer_(buffer), range_(data) {} + + Buffer(Buffer&&) = default; + Buffer& operator=(Buffer&&) = default; + + /** + * Splits the data into two pieces: [begin, begin + n), [begin + n, end). + * Their data both points into the same underlying buffer. + * Modifies the original `Buffer` to point to only [begin + n, end). + * + * @param n The offset to split at. + * @returns A buffer that owns the data [begin, begin + n). + */ + Buffer splitAt(std::size_t n) { + auto firstPiece = range_.subpiece(0, n); + range_.advance(n); + return Buffer(buffer_, firstPiece); + } + + /// Modifies the buffer to point to the range [begin + n, end). + void advance(std::size_t n) { + range_.advance(n); + } + + /// Modifies the buffer to point to the range [begin, end - n). + void subtract(std::size_t n) { + range_.subtract(n); + } + + /// Returns a read only `Range` pointing to the `Buffer`s data. + ByteRange range() const { + return range_; + } + /// Returns a mutable `Range` pointing to the `Buffer`s data. + MutableByteRange range() { + return range_; + } + + const unsigned char* data() const { + return range_.data(); + } + + unsigned char* data() { + return range_.data(); + } + + std::size_t size() const { + return range_.size(); + } + + bool empty() const { + return range_.empty(); + } +}; +} diff --git a/3rdparty/zstd/contrib/pzstd/utils/FileSystem.h b/3rdparty/zstd/contrib/pzstd/utils/FileSystem.h new file mode 100644 index 00000000000..8d57d05f0fa --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/utils/FileSystem.h @@ -0,0 +1,96 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +#pragma once + +#include "utils/Portability.h" +#include "utils/Range.h" + +#include <sys/stat.h> +#include <cerrno> +#include <cstdint> +#include <limits> +#include <system_error> + +// A small subset of `std::filesystem`. +// `std::filesystem` should be a drop in replacement. +// See https://en.cppreference.com/w/cpp/filesystem for documentation. + +namespace pzstd { + +// using file_status = ... causes gcc to emit a false positive warning +#if defined(_MSC_VER) +typedef struct ::_stat64 file_status; +#else +typedef struct ::stat file_status; +#endif + +/// https://en.cppreference.com/w/cpp/filesystem/status +inline file_status status(StringPiece path, std::error_code& ec) noexcept { + file_status status; +#if defined(_MSC_VER) + const auto error = ::_stat64(path.data(), &status); +#else + const auto error = ::stat(path.data(), &status); +#endif + if (error) { + ec.assign(errno, std::generic_category()); + } else { + ec.clear(); + } + return status; +} + +/// https://en.cppreference.com/w/cpp/filesystem/is_regular_file +inline bool is_regular_file(file_status status) noexcept { +#if defined(S_ISREG) + return S_ISREG(status.st_mode); +#elif !defined(S_ISREG) && defined(S_IFMT) && defined(S_IFREG) + return (status.st_mode & S_IFMT) == S_IFREG; +#else + static_assert(false, "No POSIX stat() support."); +#endif +} + +/// https://en.cppreference.com/w/cpp/filesystem/is_regular_file +inline bool is_regular_file(StringPiece path, std::error_code& ec) noexcept { + return is_regular_file(status(path, ec)); +} + +/// https://en.cppreference.com/w/cpp/filesystem/is_directory +inline bool is_directory(file_status status) noexcept { +#if defined(S_ISDIR) + return S_ISDIR(status.st_mode); +#elif !defined(S_ISDIR) && defined(S_IFMT) && defined(S_IFDIR) + return (status.st_mode & S_IFMT) == S_IFDIR; +#else + static_assert(false, "NO POSIX stat() support."); +#endif +} + +/// https://en.cppreference.com/w/cpp/filesystem/is_directory +inline bool is_directory(StringPiece path, std::error_code& ec) noexcept { + return is_directory(status(path, ec)); +} + +/// https://en.cppreference.com/w/cpp/filesystem/file_size +inline std::uintmax_t file_size( + StringPiece path, + std::error_code& ec) noexcept { + auto stat = status(path, ec); + if (ec) { + return std::numeric_limits<uintmax_t>::max(); + } + if (!is_regular_file(stat)) { + ec.assign(ENOTSUP, std::generic_category()); + return std::numeric_limits<uintmax_t>::max(); + } + ec.clear(); + return stat.st_size; +} +} diff --git a/3rdparty/zstd/contrib/pzstd/utils/Likely.h b/3rdparty/zstd/contrib/pzstd/utils/Likely.h new file mode 100644 index 00000000000..52243a64eab --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/utils/Likely.h @@ -0,0 +1,28 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ + +/** + * Compiler hints to indicate the fast path of an "if" branch: whether + * the if condition is likely to be true or false. + * + * @author Tudor Bosman (tudorb@fb.com) + */ + +#pragma once + +#undef LIKELY +#undef UNLIKELY + +#if defined(__GNUC__) && __GNUC__ >= 4 +#define LIKELY(x) (__builtin_expect((x), 1)) +#define UNLIKELY(x) (__builtin_expect((x), 0)) +#else +#define LIKELY(x) (x) +#define UNLIKELY(x) (x) +#endif diff --git a/3rdparty/zstd/contrib/pzstd/utils/Portability.h b/3rdparty/zstd/contrib/pzstd/utils/Portability.h new file mode 100644 index 00000000000..ef1f86e51f6 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/utils/Portability.h @@ -0,0 +1,16 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ + +#pragma once + +#include <algorithm> + +// Required for windows, which defines min/max, but we want the std:: version. +#undef min +#undef max diff --git a/3rdparty/zstd/contrib/pzstd/utils/Range.h b/3rdparty/zstd/contrib/pzstd/utils/Range.h new file mode 100644 index 00000000000..0fd8f9f8655 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/utils/Range.h @@ -0,0 +1,133 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ + +/** + * A subset of `folly/Range.h`. + * All code copied verbatim modulo formatting + */ +#pragma once + +#include "utils/Likely.h" +#include "utils/Portability.h" + +#include <algorithm> +#include <cstddef> +#include <cstring> +#include <stdexcept> +#include <string> +#include <type_traits> + +namespace pzstd { + +namespace detail { +/* + *Use IsCharPointer<T>::type to enable const char* or char*. + *Use IsCharPointer<T>::const_type to enable only const char*. +*/ +template <class T> +struct IsCharPointer {}; + +template <> +struct IsCharPointer<char*> { + typedef int type; +}; + +template <> +struct IsCharPointer<const char*> { + typedef int const_type; + typedef int type; +}; + +} // namespace detail + +template <typename Iter> +class Range { + Iter b_; + Iter e_; + + public: + using size_type = std::size_t; + using iterator = Iter; + using const_iterator = Iter; + using value_type = typename std::remove_reference< + typename std::iterator_traits<Iter>::reference>::type; + using reference = typename std::iterator_traits<Iter>::reference; + + constexpr Range() : b_(), e_() {} + constexpr Range(Iter begin, Iter end) : b_(begin), e_(end) {} + + constexpr Range(Iter begin, size_type size) : b_(begin), e_(begin + size) {} + + template <class T = Iter, typename detail::IsCharPointer<T>::type = 0> + /* implicit */ Range(Iter str) : b_(str), e_(str + std::strlen(str)) {} + + template <class T = Iter, typename detail::IsCharPointer<T>::const_type = 0> + /* implicit */ Range(const std::string& str) + : b_(str.data()), e_(b_ + str.size()) {} + + // Allow implicit conversion from Range<From> to Range<To> if From is + // implicitly convertible to To. + template < + class OtherIter, + typename std::enable_if< + (!std::is_same<Iter, OtherIter>::value && + std::is_convertible<OtherIter, Iter>::value), + int>::type = 0> + constexpr /* implicit */ Range(const Range<OtherIter>& other) + : b_(other.begin()), e_(other.end()) {} + + Range(const Range&) = default; + Range(Range&&) = default; + + Range& operator=(const Range&) = default; + Range& operator=(Range&&) = default; + + constexpr size_type size() const { + return e_ - b_; + } + bool empty() const { + return b_ == e_; + } + Iter data() const { + return b_; + } + Iter begin() const { + return b_; + } + Iter end() const { + return e_; + } + + void advance(size_type n) { + if (UNLIKELY(n > size())) { + throw std::out_of_range("index out of range"); + } + b_ += n; + } + + void subtract(size_type n) { + if (UNLIKELY(n > size())) { + throw std::out_of_range("index out of range"); + } + e_ -= n; + } + + Range subpiece(size_type first, size_type length = std::string::npos) const { + if (UNLIKELY(first > size())) { + throw std::out_of_range("index out of range"); + } + + return Range(b_ + first, std::min(length, size() - first)); + } +}; + +using ByteRange = Range<const unsigned char*>; +using MutableByteRange = Range<unsigned char*>; +using StringPiece = Range<const char*>; +} diff --git a/3rdparty/zstd/contrib/pzstd/utils/ResourcePool.h b/3rdparty/zstd/contrib/pzstd/utils/ResourcePool.h new file mode 100644 index 00000000000..7c4bb623580 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/utils/ResourcePool.h @@ -0,0 +1,96 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +#pragma once + +#include <cassert> +#include <functional> +#include <memory> +#include <mutex> +#include <vector> + +namespace pzstd { + +/** + * An unbounded pool of resources. + * A `ResourcePool<T>` requires a factory function that takes allocates `T*` and + * a free function that frees a `T*`. + * Calling `ResourcePool::get()` will give you a new `ResourcePool::UniquePtr` + * to a `T`, and when it goes out of scope the resource will be returned to the + * pool. + * The `ResourcePool<T>` *must* survive longer than any resources it hands out. + * Remember that `ResourcePool<T>` hands out mutable `T`s, so make sure to clean + * up the resource before or after every use. + */ +template <typename T> +class ResourcePool { + public: + class Deleter; + using Factory = std::function<T*()>; + using Free = std::function<void(T*)>; + using UniquePtr = std::unique_ptr<T, Deleter>; + + private: + std::mutex mutex_; + Factory factory_; + Free free_; + std::vector<T*> resources_; + unsigned inUse_; + + public: + /** + * Creates a `ResourcePool`. + * + * @param factory The function to use to create new resources. + * @param free The function to use to free resources created by `factory`. + */ + ResourcePool(Factory factory, Free free) + : factory_(std::move(factory)), free_(std::move(free)), inUse_(0) {} + + /** + * @returns A unique pointer to a resource. The resource is null iff + * there are no available resources and `factory()` returns null. + */ + UniquePtr get() { + std::lock_guard<std::mutex> lock(mutex_); + if (!resources_.empty()) { + UniquePtr resource{resources_.back(), Deleter{*this}}; + resources_.pop_back(); + ++inUse_; + return resource; + } + UniquePtr resource{factory_(), Deleter{*this}}; + ++inUse_; + return resource; + } + + ~ResourcePool() noexcept { + assert(inUse_ == 0); + for (const auto resource : resources_) { + free_(resource); + } + } + + class Deleter { + ResourcePool *pool_; + public: + explicit Deleter(ResourcePool &pool) : pool_(&pool) {} + + void operator() (T *resource) { + std::lock_guard<std::mutex> lock(pool_->mutex_); + // Make sure we don't put null resources into the pool + if (resource) { + pool_->resources_.push_back(resource); + } + assert(pool_->inUse_ > 0); + --pool_->inUse_; + } + }; +}; + +} diff --git a/3rdparty/zstd/contrib/pzstd/utils/ScopeGuard.h b/3rdparty/zstd/contrib/pzstd/utils/ScopeGuard.h new file mode 100644 index 00000000000..911fd984214 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/utils/ScopeGuard.h @@ -0,0 +1,50 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +#pragma once + +#include <utility> + +namespace pzstd { + +/** + * Dismissable scope guard. + * `Function` must be callable and take no parameters. + * Unless `dismiss()` is called, the callable is executed upon destruction of + * `ScopeGuard`. + * + * Example: + * + * auto guard = makeScopeGuard([&] { cleanup(); }); + */ +template <typename Function> +class ScopeGuard { + Function function; + bool dismissed; + + public: + explicit ScopeGuard(Function&& function) + : function(std::move(function)), dismissed(false) {} + + void dismiss() { + dismissed = true; + } + + ~ScopeGuard() noexcept { + if (!dismissed) { + function(); + } + } +}; + +/// Creates a scope guard from `function`. +template <typename Function> +ScopeGuard<Function> makeScopeGuard(Function&& function) { + return ScopeGuard<Function>(std::forward<Function>(function)); +} +} diff --git a/3rdparty/zstd/contrib/pzstd/utils/ThreadPool.h b/3rdparty/zstd/contrib/pzstd/utils/ThreadPool.h new file mode 100644 index 00000000000..a087d7c1cff --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/utils/ThreadPool.h @@ -0,0 +1,58 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +#pragma once + +#include "utils/WorkQueue.h" + +#include <cstddef> +#include <functional> +#include <thread> +#include <vector> + +namespace pzstd { +/// A simple thread pool that pulls tasks off its queue in FIFO order. +class ThreadPool { + std::vector<std::thread> threads_; + + WorkQueue<std::function<void()>> tasks_; + + public: + /// Constructs a thread pool with `numThreads` threads. + explicit ThreadPool(std::size_t numThreads) { + threads_.reserve(numThreads); + for (std::size_t i = 0; i < numThreads; ++i) { + threads_.emplace_back([this] { + std::function<void()> task; + while (tasks_.pop(task)) { + task(); + } + }); + } + } + + /// Finishes all tasks currently in the queue. + ~ThreadPool() { + tasks_.finish(); + for (auto& thread : threads_) { + thread.join(); + } + } + + /** + * Adds `task` to the queue of tasks to execute. Since `task` is a + * `std::function<>`, it cannot be a move only type. So any lambda passed must + * not capture move only types (like `std::unique_ptr`). + * + * @param task The task to execute. + */ + void add(std::function<void()> task) { + tasks_.push(std::move(task)); + } +}; +} diff --git a/3rdparty/zstd/contrib/pzstd/utils/WorkQueue.h b/3rdparty/zstd/contrib/pzstd/utils/WorkQueue.h new file mode 100644 index 00000000000..d7947b814dd --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/utils/WorkQueue.h @@ -0,0 +1,181 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +#pragma once + +#include "utils/Buffer.h" + +#include <atomic> +#include <cassert> +#include <cstddef> +#include <condition_variable> +#include <cstddef> +#include <functional> +#include <mutex> +#include <queue> + +namespace pzstd { + +/// Unbounded thread-safe work queue. +template <typename T> +class WorkQueue { + // Protects all member variable access + std::mutex mutex_; + std::condition_variable readerCv_; + std::condition_variable writerCv_; + std::condition_variable finishCv_; + + std::queue<T> queue_; + bool done_; + std::size_t maxSize_; + + // Must have lock to call this function + bool full() const { + if (maxSize_ == 0) { + return false; + } + return queue_.size() >= maxSize_; + } + + public: + /** + * Constructs an empty work queue with an optional max size. + * If `maxSize == 0` the queue size is unbounded. + * + * @param maxSize The maximum allowed size of the work queue. + */ + WorkQueue(std::size_t maxSize = 0) : done_(false), maxSize_(maxSize) {} + + /** + * Push an item onto the work queue. Notify a single thread that work is + * available. If `finish()` has been called, do nothing and return false. + * If `push()` returns false, then `item` has not been moved from. + * + * @param item Item to push onto the queue. + * @returns True upon success, false if `finish()` has been called. An + * item was pushed iff `push()` returns true. + */ + bool push(T&& item) { + { + std::unique_lock<std::mutex> lock(mutex_); + while (full() && !done_) { + writerCv_.wait(lock); + } + if (done_) { + return false; + } + queue_.push(std::move(item)); + } + readerCv_.notify_one(); + return true; + } + + /** + * Attempts to pop an item off the work queue. It will block until data is + * available or `finish()` has been called. + * + * @param[out] item If `pop` returns `true`, it contains the popped item. + * If `pop` returns `false`, it is unmodified. + * @returns True upon success. False if the queue is empty and + * `finish()` has been called. + */ + bool pop(T& item) { + { + std::unique_lock<std::mutex> lock(mutex_); + while (queue_.empty() && !done_) { + readerCv_.wait(lock); + } + if (queue_.empty()) { + assert(done_); + return false; + } + item = std::move(queue_.front()); + queue_.pop(); + } + writerCv_.notify_one(); + return true; + } + + /** + * Sets the maximum queue size. If `maxSize == 0` then it is unbounded. + * + * @param maxSize The new maximum queue size. + */ + void setMaxSize(std::size_t maxSize) { + { + std::lock_guard<std::mutex> lock(mutex_); + maxSize_ = maxSize; + } + writerCv_.notify_all(); + } + + /** + * Promise that `push()` won't be called again, so once the queue is empty + * there will never any more work. + */ + void finish() { + { + std::lock_guard<std::mutex> lock(mutex_); + assert(!done_); + done_ = true; + } + readerCv_.notify_all(); + writerCv_.notify_all(); + finishCv_.notify_all(); + } + + /// Blocks until `finish()` has been called (but the queue may not be empty). + void waitUntilFinished() { + std::unique_lock<std::mutex> lock(mutex_); + while (!done_) { + finishCv_.wait(lock); + } + } +}; + +/// Work queue for `Buffer`s that knows the total number of bytes in the queue. +class BufferWorkQueue { + WorkQueue<Buffer> queue_; + std::atomic<std::size_t> size_; + + public: + BufferWorkQueue(std::size_t maxSize = 0) : queue_(maxSize), size_(0) {} + + void push(Buffer buffer) { + size_.fetch_add(buffer.size()); + queue_.push(std::move(buffer)); + } + + bool pop(Buffer& buffer) { + bool result = queue_.pop(buffer); + if (result) { + size_.fetch_sub(buffer.size()); + } + return result; + } + + void setMaxSize(std::size_t maxSize) { + queue_.setMaxSize(maxSize); + } + + void finish() { + queue_.finish(); + } + + /** + * Blocks until `finish()` has been called. + * + * @returns The total number of bytes of all the `Buffer`s currently in the + * queue. + */ + std::size_t size() { + queue_.waitUntilFinished(); + return size_.load(); + } +}; +} diff --git a/3rdparty/zstd/contrib/pzstd/utils/test/BUCK b/3rdparty/zstd/contrib/pzstd/utils/test/BUCK new file mode 100644 index 00000000000..a5113cab6b0 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/utils/test/BUCK @@ -0,0 +1,35 @@ +cxx_test( + name='buffer_test', + srcs=['BufferTest.cpp'], + deps=['//contrib/pzstd/utils:buffer'], +) + +cxx_test( + name='range_test', + srcs=['RangeTest.cpp'], + deps=['//contrib/pzstd/utils:range'], +) + +cxx_test( + name='resource_pool_test', + srcs=['ResourcePoolTest.cpp'], + deps=['//contrib/pzstd/utils:resource_pool'], +) + +cxx_test( + name='scope_guard_test', + srcs=['ScopeGuardTest.cpp'], + deps=['//contrib/pzstd/utils:scope_guard'], +) + +cxx_test( + name='thread_pool_test', + srcs=['ThreadPoolTest.cpp'], + deps=['//contrib/pzstd/utils:thread_pool'], +) + +cxx_test( + name='work_queue_test', + srcs=['RangeTest.cpp'], + deps=['//contrib/pzstd/utils:work_queue'], +) diff --git a/3rdparty/zstd/contrib/pzstd/utils/test/BufferTest.cpp b/3rdparty/zstd/contrib/pzstd/utils/test/BufferTest.cpp new file mode 100644 index 00000000000..58bf08dcd69 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/utils/test/BufferTest.cpp @@ -0,0 +1,89 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +#include "utils/Buffer.h" +#include "utils/Range.h" + +#include <gtest/gtest.h> +#include <memory> + +using namespace pzstd; + +namespace { +void deleter(const unsigned char* buf) { + delete[] buf; +} +} + +TEST(Buffer, Constructors) { + Buffer empty; + EXPECT_TRUE(empty.empty()); + EXPECT_EQ(0, empty.size()); + + Buffer sized(5); + EXPECT_FALSE(sized.empty()); + EXPECT_EQ(5, sized.size()); + + Buffer moved(std::move(sized)); + EXPECT_FALSE(sized.empty()); + EXPECT_EQ(5, sized.size()); + + Buffer assigned; + assigned = std::move(moved); + EXPECT_FALSE(sized.empty()); + EXPECT_EQ(5, sized.size()); +} + +TEST(Buffer, BufferManagement) { + std::shared_ptr<unsigned char> buf(new unsigned char[10], deleter); + { + Buffer acquired(buf, MutableByteRange(buf.get(), buf.get() + 10)); + EXPECT_EQ(2, buf.use_count()); + Buffer moved(std::move(acquired)); + EXPECT_EQ(2, buf.use_count()); + Buffer assigned; + assigned = std::move(moved); + EXPECT_EQ(2, buf.use_count()); + + Buffer split = assigned.splitAt(5); + EXPECT_EQ(3, buf.use_count()); + + split.advance(1); + assigned.subtract(1); + EXPECT_EQ(3, buf.use_count()); + } + EXPECT_EQ(1, buf.use_count()); +} + +TEST(Buffer, Modifiers) { + Buffer buf(10); + { + unsigned char i = 0; + for (auto& byte : buf.range()) { + byte = i++; + } + } + + auto prefix = buf.splitAt(2); + + ASSERT_EQ(2, prefix.size()); + EXPECT_EQ(0, *prefix.data()); + + ASSERT_EQ(8, buf.size()); + EXPECT_EQ(2, *buf.data()); + + buf.advance(2); + EXPECT_EQ(4, *buf.data()); + + EXPECT_EQ(9, *(buf.range().end() - 1)); + + buf.subtract(2); + EXPECT_EQ(7, *(buf.range().end() - 1)); + + EXPECT_EQ(4, buf.size()); +} diff --git a/3rdparty/zstd/contrib/pzstd/utils/test/RangeTest.cpp b/3rdparty/zstd/contrib/pzstd/utils/test/RangeTest.cpp new file mode 100644 index 00000000000..8b7dee27193 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/utils/test/RangeTest.cpp @@ -0,0 +1,82 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +#include "utils/Range.h" + +#include <gtest/gtest.h> +#include <string> + +using namespace pzstd; + +// Range is directly copied from folly. +// Just some sanity tests to make sure everything seems to work. + +TEST(Range, Constructors) { + StringPiece empty; + EXPECT_TRUE(empty.empty()); + EXPECT_EQ(0, empty.size()); + + std::string str = "hello"; + { + Range<std::string::const_iterator> piece(str.begin(), str.end()); + EXPECT_EQ(5, piece.size()); + EXPECT_EQ('h', *piece.data()); + EXPECT_EQ('o', *(piece.end() - 1)); + } + + { + StringPiece piece(str.data(), str.size()); + EXPECT_EQ(5, piece.size()); + EXPECT_EQ('h', *piece.data()); + EXPECT_EQ('o', *(piece.end() - 1)); + } + + { + StringPiece piece(str); + EXPECT_EQ(5, piece.size()); + EXPECT_EQ('h', *piece.data()); + EXPECT_EQ('o', *(piece.end() - 1)); + } + + { + StringPiece piece(str.c_str()); + EXPECT_EQ(5, piece.size()); + EXPECT_EQ('h', *piece.data()); + EXPECT_EQ('o', *(piece.end() - 1)); + } +} + +TEST(Range, Modifiers) { + StringPiece range("hello world"); + ASSERT_EQ(11, range.size()); + + { + auto hello = range.subpiece(0, 5); + EXPECT_EQ(5, hello.size()); + EXPECT_EQ('h', *hello.data()); + EXPECT_EQ('o', *(hello.end() - 1)); + } + { + auto hello = range; + hello.subtract(6); + EXPECT_EQ(5, hello.size()); + EXPECT_EQ('h', *hello.data()); + EXPECT_EQ('o', *(hello.end() - 1)); + } + { + auto world = range; + world.advance(6); + EXPECT_EQ(5, world.size()); + EXPECT_EQ('w', *world.data()); + EXPECT_EQ('d', *(world.end() - 1)); + } + + std::string expected = "hello world"; + EXPECT_EQ(expected, std::string(range.begin(), range.end())); + EXPECT_EQ(expected, std::string(range.data(), range.size())); +} diff --git a/3rdparty/zstd/contrib/pzstd/utils/test/ResourcePoolTest.cpp b/3rdparty/zstd/contrib/pzstd/utils/test/ResourcePoolTest.cpp new file mode 100644 index 00000000000..750ee084b07 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/utils/test/ResourcePoolTest.cpp @@ -0,0 +1,72 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +#include "utils/ResourcePool.h" + +#include <gtest/gtest.h> +#include <atomic> +#include <thread> + +using namespace pzstd; + +TEST(ResourcePool, FullTest) { + unsigned numCreated = 0; + unsigned numDeleted = 0; + { + ResourcePool<int> pool( + [&numCreated] { ++numCreated; return new int{5}; }, + [&numDeleted](int *x) { ++numDeleted; delete x; }); + + { + auto i = pool.get(); + EXPECT_EQ(5, *i); + *i = 6; + } + { + auto i = pool.get(); + EXPECT_EQ(6, *i); + auto j = pool.get(); + EXPECT_EQ(5, *j); + *j = 7; + } + { + auto i = pool.get(); + EXPECT_EQ(6, *i); + auto j = pool.get(); + EXPECT_EQ(7, *j); + } + } + EXPECT_EQ(2, numCreated); + EXPECT_EQ(numCreated, numDeleted); +} + +TEST(ResourcePool, ThreadSafe) { + std::atomic<unsigned> numCreated{0}; + std::atomic<unsigned> numDeleted{0}; + { + ResourcePool<int> pool( + [&numCreated] { ++numCreated; return new int{0}; }, + [&numDeleted](int *x) { ++numDeleted; delete x; }); + auto push = [&pool] { + for (int i = 0; i < 100; ++i) { + auto x = pool.get(); + ++*x; + } + }; + std::thread t1{push}; + std::thread t2{push}; + t1.join(); + t2.join(); + + auto x = pool.get(); + auto y = pool.get(); + EXPECT_EQ(200, *x + *y); + } + EXPECT_GE(2, numCreated); + EXPECT_EQ(numCreated, numDeleted); +} diff --git a/3rdparty/zstd/contrib/pzstd/utils/test/ScopeGuardTest.cpp b/3rdparty/zstd/contrib/pzstd/utils/test/ScopeGuardTest.cpp new file mode 100644 index 00000000000..0f77cdf38b7 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/utils/test/ScopeGuardTest.cpp @@ -0,0 +1,28 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +#include "utils/ScopeGuard.h" + +#include <gtest/gtest.h> + +using namespace pzstd; + +TEST(ScopeGuard, Dismiss) { + { + auto guard = makeScopeGuard([&] { EXPECT_TRUE(false); }); + guard.dismiss(); + } +} + +TEST(ScopeGuard, Executes) { + bool executed = false; + { + auto guard = makeScopeGuard([&] { executed = true; }); + } + EXPECT_TRUE(executed); +} diff --git a/3rdparty/zstd/contrib/pzstd/utils/test/ThreadPoolTest.cpp b/3rdparty/zstd/contrib/pzstd/utils/test/ThreadPoolTest.cpp new file mode 100644 index 00000000000..a01052e605f --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/utils/test/ThreadPoolTest.cpp @@ -0,0 +1,71 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +#include "utils/ThreadPool.h" + +#include <gtest/gtest.h> +#include <atomic> +#include <iostream> +#include <thread> +#include <vector> + +using namespace pzstd; + +TEST(ThreadPool, Ordering) { + std::vector<int> results; + + { + ThreadPool executor(1); + for (int i = 0; i < 10; ++i) { + executor.add([ &results, i ] { results.push_back(i); }); + } + } + + for (int i = 0; i < 10; ++i) { + EXPECT_EQ(i, results[i]); + } +} + +TEST(ThreadPool, AllJobsFinished) { + std::atomic<unsigned> numFinished{0}; + std::atomic<bool> start{false}; + { + std::cerr << "Creating executor" << std::endl; + ThreadPool executor(5); + for (int i = 0; i < 10; ++i) { + executor.add([ &numFinished, &start ] { + while (!start.load()) { + std::this_thread::yield(); + } + ++numFinished; + }); + } + std::cerr << "Starting" << std::endl; + start.store(true); + std::cerr << "Finishing" << std::endl; + } + EXPECT_EQ(10, numFinished.load()); +} + +TEST(ThreadPool, AddJobWhileJoining) { + std::atomic<bool> done{false}; + { + ThreadPool executor(1); + executor.add([&executor, &done] { + while (!done.load()) { + std::this_thread::yield(); + } + // Sleep for a second to be sure that we are joining + std::this_thread::sleep_for(std::chrono::seconds(1)); + executor.add([] { + EXPECT_TRUE(false); + }); + }); + done.store(true); + } +} diff --git a/3rdparty/zstd/contrib/pzstd/utils/test/WorkQueueTest.cpp b/3rdparty/zstd/contrib/pzstd/utils/test/WorkQueueTest.cpp new file mode 100644 index 00000000000..16600bb6031 --- /dev/null +++ b/3rdparty/zstd/contrib/pzstd/utils/test/WorkQueueTest.cpp @@ -0,0 +1,282 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ +#include "utils/Buffer.h" +#include "utils/WorkQueue.h" + +#include <gtest/gtest.h> +#include <iostream> +#include <memory> +#include <mutex> +#include <thread> +#include <vector> + +using namespace pzstd; + +namespace { +struct Popper { + WorkQueue<int>* queue; + int* results; + std::mutex* mutex; + + void operator()() { + int result; + while (queue->pop(result)) { + std::lock_guard<std::mutex> lock(*mutex); + results[result] = result; + } + } +}; +} + +TEST(WorkQueue, SingleThreaded) { + WorkQueue<int> queue; + int result; + + queue.push(5); + EXPECT_TRUE(queue.pop(result)); + EXPECT_EQ(5, result); + + queue.push(1); + queue.push(2); + EXPECT_TRUE(queue.pop(result)); + EXPECT_EQ(1, result); + EXPECT_TRUE(queue.pop(result)); + EXPECT_EQ(2, result); + + queue.push(1); + queue.push(2); + queue.finish(); + EXPECT_TRUE(queue.pop(result)); + EXPECT_EQ(1, result); + EXPECT_TRUE(queue.pop(result)); + EXPECT_EQ(2, result); + EXPECT_FALSE(queue.pop(result)); + + queue.waitUntilFinished(); +} + +TEST(WorkQueue, SPSC) { + WorkQueue<int> queue; + const int max = 100; + + for (int i = 0; i < 10; ++i) { + queue.push(int{i}); + } + + std::thread thread([ &queue, max ] { + int result; + for (int i = 0;; ++i) { + if (!queue.pop(result)) { + EXPECT_EQ(i, max); + break; + } + EXPECT_EQ(i, result); + } + }); + + std::this_thread::yield(); + for (int i = 10; i < max; ++i) { + queue.push(int{i}); + } + queue.finish(); + + thread.join(); +} + +TEST(WorkQueue, SPMC) { + WorkQueue<int> queue; + std::vector<int> results(50, -1); + std::mutex mutex; + std::vector<std::thread> threads; + for (int i = 0; i < 5; ++i) { + threads.emplace_back(Popper{&queue, results.data(), &mutex}); + } + + for (int i = 0; i < 50; ++i) { + queue.push(int{i}); + } + queue.finish(); + + for (auto& thread : threads) { + thread.join(); + } + + for (int i = 0; i < 50; ++i) { + EXPECT_EQ(i, results[i]); + } +} + +TEST(WorkQueue, MPMC) { + WorkQueue<int> queue; + std::vector<int> results(100, -1); + std::mutex mutex; + std::vector<std::thread> popperThreads; + for (int i = 0; i < 4; ++i) { + popperThreads.emplace_back(Popper{&queue, results.data(), &mutex}); + } + + std::vector<std::thread> pusherThreads; + for (int i = 0; i < 2; ++i) { + auto min = i * 50; + auto max = (i + 1) * 50; + pusherThreads.emplace_back( + [ &queue, min, max ] { + for (int i = min; i < max; ++i) { + queue.push(int{i}); + } + }); + } + + for (auto& thread : pusherThreads) { + thread.join(); + } + queue.finish(); + + for (auto& thread : popperThreads) { + thread.join(); + } + + for (int i = 0; i < 100; ++i) { + EXPECT_EQ(i, results[i]); + } +} + +TEST(WorkQueue, BoundedSizeWorks) { + WorkQueue<int> queue(1); + int result; + queue.push(5); + queue.pop(result); + queue.push(5); + queue.pop(result); + queue.push(5); + queue.finish(); + queue.pop(result); + EXPECT_EQ(5, result); +} + +TEST(WorkQueue, BoundedSizePushAfterFinish) { + WorkQueue<int> queue(1); + int result; + queue.push(5); + std::thread pusher([&queue] { + queue.push(6); + }); + // Dirtily try and make sure that pusher has run. + std::this_thread::sleep_for(std::chrono::seconds(1)); + queue.finish(); + EXPECT_TRUE(queue.pop(result)); + EXPECT_EQ(5, result); + EXPECT_FALSE(queue.pop(result)); + + pusher.join(); +} + +TEST(WorkQueue, SetMaxSize) { + WorkQueue<int> queue(2); + int result; + queue.push(5); + queue.push(6); + queue.setMaxSize(1); + std::thread pusher([&queue] { + queue.push(7); + }); + // Dirtily try and make sure that pusher has run. + std::this_thread::sleep_for(std::chrono::seconds(1)); + queue.finish(); + EXPECT_TRUE(queue.pop(result)); + EXPECT_EQ(5, result); + EXPECT_TRUE(queue.pop(result)); + EXPECT_EQ(6, result); + EXPECT_FALSE(queue.pop(result)); + + pusher.join(); +} + +TEST(WorkQueue, BoundedSizeMPMC) { + WorkQueue<int> queue(10); + std::vector<int> results(200, -1); + std::mutex mutex; + std::cerr << "Creating popperThreads" << std::endl; + std::vector<std::thread> popperThreads; + for (int i = 0; i < 4; ++i) { + popperThreads.emplace_back(Popper{&queue, results.data(), &mutex}); + } + + std::cerr << "Creating pusherThreads" << std::endl; + std::vector<std::thread> pusherThreads; + for (int i = 0; i < 2; ++i) { + auto min = i * 100; + auto max = (i + 1) * 100; + pusherThreads.emplace_back( + [ &queue, min, max ] { + for (int i = min; i < max; ++i) { + queue.push(int{i}); + } + }); + } + + std::cerr << "Joining pusherThreads" << std::endl; + for (auto& thread : pusherThreads) { + thread.join(); + } + std::cerr << "Finishing queue" << std::endl; + queue.finish(); + + std::cerr << "Joining popperThreads" << std::endl; + for (auto& thread : popperThreads) { + thread.join(); + } + + std::cerr << "Inspecting results" << std::endl; + for (int i = 0; i < 200; ++i) { + EXPECT_EQ(i, results[i]); + } +} + +TEST(WorkQueue, FailedPush) { + WorkQueue<std::unique_ptr<int>> queue; + std::unique_ptr<int> x(new int{5}); + EXPECT_TRUE(queue.push(std::move(x))); + EXPECT_EQ(nullptr, x); + queue.finish(); + x.reset(new int{6}); + EXPECT_FALSE(queue.push(std::move(x))); + EXPECT_NE(nullptr, x); + EXPECT_EQ(6, *x); +} + +TEST(BufferWorkQueue, SizeCalculatedCorrectly) { + { + BufferWorkQueue queue; + queue.finish(); + EXPECT_EQ(0, queue.size()); + } + { + BufferWorkQueue queue; + queue.push(Buffer(10)); + queue.finish(); + EXPECT_EQ(10, queue.size()); + } + { + BufferWorkQueue queue; + queue.push(Buffer(10)); + queue.push(Buffer(5)); + queue.finish(); + EXPECT_EQ(15, queue.size()); + } + { + BufferWorkQueue queue; + queue.push(Buffer(10)); + queue.push(Buffer(5)); + queue.finish(); + Buffer buffer; + queue.pop(buffer); + EXPECT_EQ(5, queue.size()); + } +} diff --git a/3rdparty/zstd/contrib/recovery/Makefile b/3rdparty/zstd/contrib/recovery/Makefile new file mode 100644 index 00000000000..be6ea4b0e9c --- /dev/null +++ b/3rdparty/zstd/contrib/recovery/Makefile @@ -0,0 +1,35 @@ +# ################################################################ +# Copyright (c) Meta Platforms, Inc. and affiliates. +# All rights reserved. +# +# This source code is licensed under both the BSD-style license (found in the +# LICENSE file in the root directory of this source tree) and the GPLv2 (found +# in the COPYING file in the root directory of this source tree). +# ################################################################ + +.PHONY: all +all: recover_directory + +ZSTDLIBDIR ?= ../../lib +PROGRAMDIR ?= ../../programs + +CFLAGS ?= -O3 +CFLAGS += -I$(ZSTDLIBDIR) -I$(PROGRAMDIR) +CFLAGS += -Wall -Wextra -Wcast-qual -Wcast-align -Wshadow \ + -Wstrict-aliasing=1 -Wswitch-enum \ + -Wstrict-prototypes -Wundef \ + -Wvla -Wformat=2 -Winit-self -Wfloat-equal -Wwrite-strings \ + -Wredundant-decls -Wmissing-prototypes +CFLAGS += $(DEBUGFLAGS) $(MOREFLAGS) +FLAGS = $(CPPFLAGS) $(CFLAGS) $(LDFLAGS) + +.PHONY: $(ZSTDLIBDIR)/libzstd.a +$(ZSTDLIBDIR)/libzstd.a: + $(MAKE) -C $(ZSTDLIBDIR) libzstd.a + +recover_directory: recover_directory.c $(ZSTDLIBDIR)/libzstd.a $(PROGRAMDIR)/util.c + $(CC) $(FLAGS) $^ -o $@$(EXT) + +.PHONY: clean +clean: + rm -f recover_directory diff --git a/3rdparty/zstd/contrib/recovery/recover_directory.c b/3rdparty/zstd/contrib/recovery/recover_directory.c new file mode 100644 index 00000000000..b9bd7ab4996 --- /dev/null +++ b/3rdparty/zstd/contrib/recovery/recover_directory.c @@ -0,0 +1,152 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ + +#include <string.h> +#include <stdio.h> +#include <stdlib.h> + +#define ZSTD_STATIC_LINKING_ONLY +#include "util.h" +#include "zstd.h" + +#define CHECK(cond, ...) \ + do { \ + if (!(cond)) { \ + fprintf(stderr, "%s:%d CHECK(%s) failed: ", __FILE__, __LINE__, #cond); \ + fprintf(stderr, "" __VA_ARGS__); \ + fprintf(stderr, "\n"); \ + exit(1); \ + } \ + } while (0) + +static void usage(char const *program) { + fprintf(stderr, "USAGE: %s FILE.zst PREFIX\n", program); + fprintf(stderr, "FILE.zst: A zstd compressed file with multiple frames\n"); + fprintf(stderr, "PREFIX: The output prefix. Uncompressed files will be " + "created named ${PREFIX}0 ${PREFIX}1...\n\n"); + fprintf(stderr, "This program takes concatenated zstd frames and " + "decompresses them into individual files.\n"); + fprintf(stderr, "E.g. files created with a command like: zstd -r directory " + "-o file.zst\n"); +} + +typedef struct { + char *data; + size_t size; + size_t frames; + size_t maxFrameSize; +} ZstdFrames; + +static ZstdFrames readFile(char const *fileName) { + U64 const fileSize = UTIL_getFileSize(fileName); + CHECK(fileSize != UTIL_FILESIZE_UNKNOWN, "Unknown file size!"); + + char *const data = (char *)malloc(fileSize); + CHECK(data != NULL, "Allocation failed"); + + FILE *file = fopen(fileName, "rb"); + CHECK(file != NULL, "fopen failed"); + + size_t const readSize = fread(data, 1, fileSize, file); + CHECK(readSize == fileSize, "fread failed"); + + fclose(file); + ZstdFrames frames; + frames.data = (char *)data; + frames.size = fileSize; + frames.frames = 0; + + size_t index; + size_t maxFrameSize = 0; + for (index = 0; index < fileSize;) { + size_t const frameSize = + ZSTD_findFrameCompressedSize(data + index, fileSize - index); + CHECK(!ZSTD_isError(frameSize), "Bad zstd frame: %s", + ZSTD_getErrorName(frameSize)); + if (frameSize > maxFrameSize) + maxFrameSize = frameSize; + frames.frames += 1; + index += frameSize; + } + CHECK(index == fileSize, "Zstd file corrupt!"); + frames.maxFrameSize = maxFrameSize; + + return frames; +} + +static int computePadding(size_t numFrames) { + return snprintf(NULL, 0, "%u", (unsigned)numFrames); +} + +int main(int argc, char **argv) { + if (argc != 3) { + usage(argv[0]); + exit(1); + } + char const *const zstdFile = argv[1]; + char const *const prefix = argv[2]; + + ZstdFrames frames = readFile(zstdFile); + + if (frames.frames <= 1) { + fprintf( + stderr, + "%s only has %u zstd frame. Simply use `zstd -d` to decompress it.\n", + zstdFile, (unsigned)frames.frames); + exit(1); + } + + int const padding = computePadding(frames.frames - 1); + + size_t const outFileNameSize = strlen(prefix) + padding + 1; + char* outFileName = malloc(outFileNameSize); + CHECK(outFileName != NULL, "Allocation failure"); + + size_t const bufferSize = 128 * 1024; + void *buffer = malloc(bufferSize); + CHECK(buffer != NULL, "Allocation failure"); + + ZSTD_DCtx* dctx = ZSTD_createDCtx(); + CHECK(dctx != NULL, "Allocation failure"); + + fprintf(stderr, "Recovering %u files...\n", (unsigned)frames.frames); + + size_t index; + size_t frame = 0; + for (index = 0; index < frames.size; ++frame) { + size_t const frameSize = + ZSTD_findFrameCompressedSize(frames.data + index, frames.size - index); + + int const ret = snprintf(outFileName, outFileNameSize, "%s%0*u", prefix, padding, (unsigned)frame); + CHECK(ret >= 0 && (size_t)ret <= outFileNameSize, "snprintf failed!"); + + FILE* outFile = fopen(outFileName, "wb"); + CHECK(outFile != NULL, "fopen failed"); + + ZSTD_DCtx_reset(dctx, ZSTD_reset_session_only); + ZSTD_inBuffer in = {frames.data + index, frameSize, 0}; + while (in.pos < in.size) { + ZSTD_outBuffer out = {buffer, bufferSize, 0}; + CHECK(!ZSTD_isError(ZSTD_decompressStream(dctx, &out, &in)), "decompression failed"); + size_t const writeSize = fwrite(out.dst, 1, out.pos, outFile); + CHECK(writeSize == out.pos, "fwrite failed"); + } + fclose(outFile); + fprintf(stderr, "Recovered %s\n", outFileName); + index += frameSize; + } + fprintf(stderr, "Complete\n"); + + free(outFileName); + ZSTD_freeDCtx(dctx); + free(buffer); + free(frames.data); + return 0; +} diff --git a/3rdparty/zstd/contrib/seekable_format/README.md b/3rdparty/zstd/contrib/seekable_format/README.md new file mode 100644 index 00000000000..fedf96bab4d --- /dev/null +++ b/3rdparty/zstd/contrib/seekable_format/README.md @@ -0,0 +1,42 @@ +# Zstandard Seekable Format + +The seekable format splits compressed data into a series of independent "frames", +each compressed individually, +so that decompression of a section in the middle of an archive +only requires zstd to decompress at most a frame's worth of extra data, +instead of the entire archive. + +The frames are appended, so that the decompression of the entire payload +still regenerates the original content, using any compliant zstd decoder. + +On top of that, the seekable format generates a jump table, +which makes it possible to jump directly to the position of the relevant frame +when requesting only a segment of the data. +The jump table is simply ignored by zstd decoders unaware of the seekable format. + +The format is delivered with an API to create seekable archives +and to retrieve arbitrary segments inside the archive. + +### Maximum Frame Size parameter + +When creating a seekable archive, the main parameter is the maximum frame size. + +At compression time, user can manually select the boundaries between segments, +but they don't have to: long segments will be automatically split +when larger than selected maximum frame size. + +Small frame sizes reduce decompression cost when requesting small segments, +because the decoder will nonetheless have to decompress an entire frame +to recover just a single byte from it. + +A good rule of thumb is to select a maximum frame size roughly equivalent +to the access pattern when it's known. +For example, if the application tends to request 4KB blocks, +then it's a good idea to set a maximum frame size in the vicinity of 4 KB. + +But small frame sizes also reduce compression ratio, +and increase the cost for the jump table, +so there is a balance to find. + +In general, try to avoid really tiny frame sizes (<1 KB), +which would have a large negative impact on compression ratio. diff --git a/3rdparty/zstd/contrib/seekable_format/examples/.gitignore b/3rdparty/zstd/contrib/seekable_format/examples/.gitignore new file mode 100644 index 00000000000..0b83f5e11c6 --- /dev/null +++ b/3rdparty/zstd/contrib/seekable_format/examples/.gitignore @@ -0,0 +1,5 @@ +seekable_compression +seekable_decompression +seekable_decompression_mem +parallel_processing +parallel_compression diff --git a/3rdparty/zstd/contrib/seekable_format/examples/Makefile b/3rdparty/zstd/contrib/seekable_format/examples/Makefile new file mode 100644 index 00000000000..fcc04587ce8 --- /dev/null +++ b/3rdparty/zstd/contrib/seekable_format/examples/Makefile @@ -0,0 +1,53 @@ +# ################################################################ +# Copyright (c) Meta Platforms, Inc. and affiliates. +# All rights reserved. +# +# This source code is licensed under both the BSD-style license (found in the +# LICENSE file in the root directory of this source tree) and the GPLv2 (found +# in the COPYING file in the root directory of this source tree). +# ################################################################ + +# This Makefile presumes libzstd is built, using `make` in / or /lib/ + +ZSTDLIB_PATH = ../../../lib +ZSTDLIB_NAME = libzstd.a +ZSTDLIB = $(ZSTDLIB_PATH)/$(ZSTDLIB_NAME) + +CPPFLAGS += -DXXH_NAMESPACE=ZSTD_ -I../ -I../../../lib -I../../../lib/common + +CFLAGS ?= -O3 +CFLAGS += -g + +SEEKABLE_OBJS = ../zstdseek_compress.c ../zstdseek_decompress.c $(ZSTDLIB) + +.PHONY: default all clean test + +default: all + +all: seekable_compression seekable_decompression seekable_decompression_mem \ + parallel_processing + +$(ZSTDLIB): + make -C $(ZSTDLIB_PATH) $(ZSTDLIB_NAME) + +seekable_compression : seekable_compression.c $(SEEKABLE_OBJS) + $(CC) $(CPPFLAGS) $(CFLAGS) $^ $(LDFLAGS) -o $@ + +seekable_decompression : seekable_decompression.c $(SEEKABLE_OBJS) + $(CC) $(CPPFLAGS) $(CFLAGS) $^ $(LDFLAGS) -o $@ + +seekable_decompression_mem : seekable_decompression_mem.c $(SEEKABLE_OBJS) + $(CC) $(CPPFLAGS) $(CFLAGS) $^ $(LDFLAGS) -o $@ + +parallel_processing : parallel_processing.c $(SEEKABLE_OBJS) + $(CC) $(CPPFLAGS) $(CFLAGS) $^ $(LDFLAGS) -o $@ -pthread + +parallel_compression : parallel_compression.c $(SEEKABLE_OBJS) + $(CC) $(CPPFLAGS) $(CFLAGS) $^ $(LDFLAGS) -o $@ -pthread + +clean: + @rm -f core *.o tmp* result* *.zst \ + seekable_compression seekable_decompression \ + seekable_decompression_mem \ + parallel_processing parallel_compression + @echo Cleaning completed diff --git a/3rdparty/zstd/contrib/seekable_format/examples/parallel_compression.c b/3rdparty/zstd/contrib/seekable_format/examples/parallel_compression.c new file mode 100644 index 00000000000..4e06fae3241 --- /dev/null +++ b/3rdparty/zstd/contrib/seekable_format/examples/parallel_compression.c @@ -0,0 +1,214 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ + +#include <stdlib.h> // malloc, free, exit, atoi +#include <stdio.h> // fprintf, perror, feof, fopen, etc. +#include <string.h> // strlen, memset, strcat +#define ZSTD_STATIC_LINKING_ONLY +#include <zstd.h> // presumes zstd library is installed +#include <zstd_errors.h> +#if defined(WIN32) || defined(_WIN32) +# include <windows.h> +# define SLEEP(x) Sleep(x) +#else +# include <unistd.h> +# define SLEEP(x) usleep(x * 1000) +#endif + +#include "xxhash.h" + +#include "pool.h" // use zstd thread pool for demo + +#include "../zstd_seekable.h" + +static void* malloc_orDie(size_t size) +{ + void* const buff = malloc(size); + if (buff) return buff; + /* error */ + perror("malloc:"); + exit(1); +} + +static FILE* fopen_orDie(const char *filename, const char *instruction) +{ + FILE* const inFile = fopen(filename, instruction); + if (inFile) return inFile; + /* error */ + perror(filename); + exit(3); +} + +static size_t fread_orDie(void* buffer, size_t sizeToRead, FILE* file) +{ + size_t const readSize = fread(buffer, 1, sizeToRead, file); + if (readSize == sizeToRead) return readSize; /* good */ + if (feof(file)) return readSize; /* good, reached end of file */ + /* error */ + perror("fread"); + exit(4); +} + +static size_t fwrite_orDie(const void* buffer, size_t sizeToWrite, FILE* file) +{ + size_t const writtenSize = fwrite(buffer, 1, sizeToWrite, file); + if (writtenSize == sizeToWrite) return sizeToWrite; /* good */ + /* error */ + perror("fwrite"); + exit(5); +} + +static size_t fclose_orDie(FILE* file) +{ + if (!fclose(file)) return 0; + /* error */ + perror("fclose"); + exit(6); +} + +static void fseek_orDie(FILE* file, long int offset, int origin) +{ + if (!fseek(file, offset, origin)) { + if (!fflush(file)) return; + } + /* error */ + perror("fseek"); + exit(7); +} + +static long int ftell_orDie(FILE* file) +{ + long int off = ftell(file); + if (off != -1) return off; + /* error */ + perror("ftell"); + exit(8); +} + +struct job { + const void* src; + size_t srcSize; + void* dst; + size_t dstSize; + + unsigned checksum; + + int compressionLevel; + int done; +}; + +static void compressFrame(void* opaque) +{ + struct job* job = opaque; + + job->checksum = XXH64(job->src, job->srcSize, 0); + + size_t ret = ZSTD_compress(job->dst, job->dstSize, job->src, job->srcSize, job->compressionLevel); + if (ZSTD_isError(ret)) { + fprintf(stderr, "ZSTD_compress() error : %s \n", ZSTD_getErrorName(ret)); + exit(20); + } + + job->dstSize = ret; + job->done = 1; +} + +static void compressFile_orDie(const char* fname, const char* outName, int cLevel, unsigned frameSize, int nbThreads) +{ + POOL_ctx* pool = POOL_create(nbThreads, nbThreads); + if (pool == NULL) { fprintf(stderr, "POOL_create() error \n"); exit(9); } + + FILE* const fin = fopen_orDie(fname, "rb"); + FILE* const fout = fopen_orDie(outName, "wb"); + + if (ZSTD_compressBound(frameSize) > 0xFFFFFFFFU) { fprintf(stderr, "Frame size too large \n"); exit(10); } + unsigned dstSize = ZSTD_compressBound(frameSize); + + + fseek_orDie(fin, 0, SEEK_END); + long int length = ftell_orDie(fin); + fseek_orDie(fin, 0, SEEK_SET); + + size_t numFrames = (length + frameSize - 1) / frameSize; + + struct job* jobs = malloc_orDie(sizeof(struct job) * numFrames); + + size_t i; + for(i = 0; i < numFrames; i++) { + void* in = malloc_orDie(frameSize); + void* out = malloc_orDie(dstSize); + + size_t inSize = fread_orDie(in, frameSize, fin); + + jobs[i].src = in; + jobs[i].srcSize = inSize; + jobs[i].dst = out; + jobs[i].dstSize = dstSize; + jobs[i].compressionLevel = cLevel; + jobs[i].done = 0; + POOL_add(pool, compressFrame, &jobs[i]); + } + + ZSTD_frameLog* fl = ZSTD_seekable_createFrameLog(1); + if (fl == NULL) { fprintf(stderr, "ZSTD_seekable_createFrameLog() failed \n"); exit(11); } + for (i = 0; i < numFrames; i++) { + while (!jobs[i].done) SLEEP(5); /* wake up every 5 milliseconds to check */ + fwrite_orDie(jobs[i].dst, jobs[i].dstSize, fout); + free((void*)jobs[i].src); + free(jobs[i].dst); + + size_t ret = ZSTD_seekable_logFrame(fl, jobs[i].dstSize, jobs[i].srcSize, jobs[i].checksum); + if (ZSTD_isError(ret)) { fprintf(stderr, "ZSTD_seekable_logFrame() error : %s \n", ZSTD_getErrorName(ret)); } + } + + { unsigned char seekTableBuff[1024]; + ZSTD_outBuffer out = {seekTableBuff, 1024, 0}; + while (ZSTD_seekable_writeSeekTable(fl, &out) != 0) { + fwrite_orDie(seekTableBuff, out.pos, fout); + out.pos = 0; + } + fwrite_orDie(seekTableBuff, out.pos, fout); + } + + ZSTD_seekable_freeFrameLog(fl); + free(jobs); + fclose_orDie(fout); + fclose_orDie(fin); +} + +static const char* createOutFilename_orDie(const char* filename) +{ + size_t const inL = strlen(filename); + size_t const outL = inL + 5; + void* outSpace = malloc_orDie(outL); + memset(outSpace, 0, outL); + strcat(outSpace, filename); + strcat(outSpace, ".zst"); + return (const char*)outSpace; +} + +int main(int argc, const char** argv) { + const char* const exeName = argv[0]; + if (argc!=4) { + printf("wrong arguments\n"); + printf("usage:\n"); + printf("%s FILE FRAME_SIZE NB_THREADS\n", exeName); + return 1; + } + + { const char* const inFileName = argv[1]; + unsigned const frameSize = (unsigned)atoi(argv[2]); + int const nbThreads = atoi(argv[3]); + + const char* const outFileName = createOutFilename_orDie(inFileName); + compressFile_orDie(inFileName, outFileName, 5, frameSize, nbThreads); + } + + return 0; +} diff --git a/3rdparty/zstd/contrib/seekable_format/examples/parallel_processing.c b/3rdparty/zstd/contrib/seekable_format/examples/parallel_processing.c new file mode 100644 index 00000000000..356561e5a60 --- /dev/null +++ b/3rdparty/zstd/contrib/seekable_format/examples/parallel_processing.c @@ -0,0 +1,194 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ + +/* + * A simple demo that sums up all the bytes in the file in parallel using + * seekable decompression and the zstd thread pool + */ + +#include <stdlib.h> // malloc, exit +#include <stdio.h> // fprintf, perror, feof +#include <string.h> // strerror +#include <errno.h> // errno +#define ZSTD_STATIC_LINKING_ONLY +#include <zstd.h> // presumes zstd library is installed +#include <zstd_errors.h> +#if defined(WIN32) || defined(_WIN32) +# include <windows.h> +# define SLEEP(x) Sleep(x) +#else +# include <unistd.h> +# define SLEEP(x) usleep(x * 1000) +#endif + +#include "pool.h" // use zstd thread pool for demo + +#include "../zstd_seekable.h" + +#define MIN(a, b) ((a) < (b) ? (a) : (b)) + +static void* malloc_orDie(size_t size) +{ + void* const buff = malloc(size); + if (buff) return buff; + /* error */ + perror("malloc"); + exit(1); +} + +static void* realloc_orDie(void* ptr, size_t size) +{ + ptr = realloc(ptr, size); + if (ptr) return ptr; + /* error */ + perror("realloc"); + exit(1); +} + +static FILE* fopen_orDie(const char *filename, const char *instruction) +{ + FILE* const inFile = fopen(filename, instruction); + if (inFile) return inFile; + /* error */ + perror(filename); + exit(3); +} + +static size_t fread_orDie(void* buffer, size_t sizeToRead, FILE* file) +{ + size_t const readSize = fread(buffer, 1, sizeToRead, file); + if (readSize == sizeToRead) return readSize; /* good */ + if (feof(file)) return readSize; /* good, reached end of file */ + /* error */ + perror("fread"); + exit(4); +} + +static size_t fwrite_orDie(const void* buffer, size_t sizeToWrite, FILE* file) +{ + size_t const writtenSize = fwrite(buffer, 1, sizeToWrite, file); + if (writtenSize == sizeToWrite) return sizeToWrite; /* good */ + /* error */ + perror("fwrite"); + exit(5); +} + +static size_t fclose_orDie(FILE* file) +{ + if (!fclose(file)) return 0; + /* error */ + perror("fclose"); + exit(6); +} + +static void fseek_orDie(FILE* file, long int offset, int origin) { + if (!fseek(file, offset, origin)) { + if (!fflush(file)) return; + } + /* error */ + perror("fseek"); + exit(7); +} + +struct sum_job { + const char* fname; + unsigned long long sum; + unsigned frameNb; + int done; +}; + +static void sumFrame(void* opaque) +{ + struct sum_job* job = (struct sum_job*)opaque; + job->done = 0; + + FILE* const fin = fopen_orDie(job->fname, "rb"); + + ZSTD_seekable* const seekable = ZSTD_seekable_create(); + if (seekable==NULL) { fprintf(stderr, "ZSTD_seekable_create() error \n"); exit(10); } + + size_t const initResult = ZSTD_seekable_initFile(seekable, fin); + if (ZSTD_isError(initResult)) { fprintf(stderr, "ZSTD_seekable_init() error : %s \n", ZSTD_getErrorName(initResult)); exit(11); } + + size_t const frameSize = ZSTD_seekable_getFrameDecompressedSize(seekable, job->frameNb); + unsigned char* data = malloc_orDie(frameSize); + + size_t result = ZSTD_seekable_decompressFrame(seekable, data, frameSize, job->frameNb); + if (ZSTD_isError(result)) { fprintf(stderr, "ZSTD_seekable_decompressFrame() error : %s \n", ZSTD_getErrorName(result)); exit(12); } + + unsigned long long sum = 0; + size_t i; + for (i = 0; i < frameSize; i++) { + sum += data[i]; + } + job->sum = sum; + job->done = 1; + + fclose(fin); + ZSTD_seekable_free(seekable); + free(data); +} + +static void sumFile_orDie(const char* fname, int nbThreads) +{ + POOL_ctx* pool = POOL_create(nbThreads, nbThreads); + if (pool == NULL) { fprintf(stderr, "POOL_create() error \n"); exit(9); } + + FILE* const fin = fopen_orDie(fname, "rb"); + + ZSTD_seekable* const seekable = ZSTD_seekable_create(); + if (seekable==NULL) { fprintf(stderr, "ZSTD_seekable_create() error \n"); exit(10); } + + size_t const initResult = ZSTD_seekable_initFile(seekable, fin); + if (ZSTD_isError(initResult)) { fprintf(stderr, "ZSTD_seekable_init() error : %s \n", ZSTD_getErrorName(initResult)); exit(11); } + + unsigned const numFrames = ZSTD_seekable_getNumFrames(seekable); + struct sum_job* jobs = (struct sum_job*)malloc(numFrames * sizeof(struct sum_job)); + + unsigned fnb; + for (fnb = 0; fnb < numFrames; fnb++) { + jobs[fnb] = (struct sum_job){ fname, 0, fnb, 0 }; + POOL_add(pool, sumFrame, &jobs[fnb]); + } + + unsigned long long total = 0; + + for (fnb = 0; fnb < numFrames; fnb++) { + while (!jobs[fnb].done) SLEEP(5); /* wake up every 5 milliseconds to check */ + total += jobs[fnb].sum; + } + + printf("Sum: %llu\n", total); + + POOL_free(pool); + ZSTD_seekable_free(seekable); + fclose(fin); + free(jobs); +} + + +int main(int argc, const char** argv) +{ + const char* const exeName = argv[0]; + + if (argc!=3) { + fprintf(stderr, "wrong arguments\n"); + fprintf(stderr, "usage:\n"); + fprintf(stderr, "%s FILE NB_THREADS\n", exeName); + return 1; + } + + { + const char* const inFilename = argv[1]; + int const nbThreads = atoi(argv[2]); + sumFile_orDie(inFilename, nbThreads); + } + + return 0; +} diff --git a/3rdparty/zstd/contrib/seekable_format/examples/seekable_compression.c b/3rdparty/zstd/contrib/seekable_format/examples/seekable_compression.c new file mode 100644 index 00000000000..c3d227dd05a --- /dev/null +++ b/3rdparty/zstd/contrib/seekable_format/examples/seekable_compression.c @@ -0,0 +1,136 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ + +#include <stdlib.h> // malloc, free, exit, atoi +#include <stdio.h> // fprintf, perror, feof, fopen, etc. +#include <string.h> // strlen, memset, strcat +#define ZSTD_STATIC_LINKING_ONLY +#include <zstd.h> // presumes zstd library is installed + +#include "../zstd_seekable.h" + +static void* malloc_orDie(size_t size) +{ + void* const buff = malloc(size); + if (buff) return buff; + /* error */ + perror("malloc:"); + exit(1); +} + +static FILE* fopen_orDie(const char *filename, const char *instruction) +{ + FILE* const inFile = fopen(filename, instruction); + if (inFile) return inFile; + /* error */ + perror(filename); + exit(3); +} + +static size_t fread_orDie(void* buffer, size_t sizeToRead, FILE* file) +{ + size_t const readSize = fread(buffer, 1, sizeToRead, file); + if (readSize == sizeToRead) return readSize; /* good */ + if (feof(file)) return readSize; /* good, reached end of file */ + /* error */ + perror("fread"); + exit(4); +} + +static size_t fwrite_orDie(const void* buffer, size_t sizeToWrite, FILE* file) +{ + size_t const writtenSize = fwrite(buffer, 1, sizeToWrite, file); + if (writtenSize == sizeToWrite) return sizeToWrite; /* good */ + /* error */ + perror("fwrite"); + exit(5); +} + +static size_t fclose_orDie(FILE* file) +{ + if (!fclose(file)) return 0; + /* error */ + perror("fclose"); + exit(6); +} + +static void compressFile_orDie(const char* fname, const char* outName, int cLevel, unsigned frameSize) +{ + FILE* const fin = fopen_orDie(fname, "rb"); + FILE* const fout = fopen_orDie(outName, "wb"); + size_t const buffInSize = ZSTD_CStreamInSize(); /* can always read one full block */ + void* const buffIn = malloc_orDie(buffInSize); + size_t const buffOutSize = ZSTD_CStreamOutSize(); /* can always flush a full block */ + void* const buffOut = malloc_orDie(buffOutSize); + + ZSTD_seekable_CStream* const cstream = ZSTD_seekable_createCStream(); + if (cstream==NULL) { fprintf(stderr, "ZSTD_seekable_createCStream() error \n"); exit(10); } + size_t const initResult = ZSTD_seekable_initCStream(cstream, cLevel, 1, frameSize); + if (ZSTD_isError(initResult)) { fprintf(stderr, "ZSTD_seekable_initCStream() error : %s \n", ZSTD_getErrorName(initResult)); exit(11); } + + size_t read, toRead = buffInSize; + while( (read = fread_orDie(buffIn, toRead, fin)) ) { + ZSTD_inBuffer input = { buffIn, read, 0 }; + while (input.pos < input.size) { + ZSTD_outBuffer output = { buffOut, buffOutSize, 0 }; + toRead = ZSTD_seekable_compressStream(cstream, &output , &input); /* toRead is guaranteed to be <= ZSTD_CStreamInSize() */ + if (ZSTD_isError(toRead)) { fprintf(stderr, "ZSTD_seekable_compressStream() error : %s \n", ZSTD_getErrorName(toRead)); exit(12); } + if (toRead > buffInSize) toRead = buffInSize; /* Safely handle case when `buffInSize` is manually changed to a value < ZSTD_CStreamInSize()*/ + fwrite_orDie(buffOut, output.pos, fout); + } + } + + while (1) { + ZSTD_outBuffer output = { buffOut, buffOutSize, 0 }; + size_t const remainingToFlush = ZSTD_seekable_endStream(cstream, &output); /* close stream */ + if (ZSTD_isError(remainingToFlush)) { fprintf(stderr, "ZSTD_seekable_endStream() error : %s \n", ZSTD_getErrorName(remainingToFlush)); exit(13); } + fwrite_orDie(buffOut, output.pos, fout); + if (!remainingToFlush) break; + } + + ZSTD_seekable_freeCStream(cstream); + fclose_orDie(fout); + fclose_orDie(fin); + free(buffIn); + free(buffOut); +} + +static char* createOutFilename_orDie(const char* filename) +{ + size_t const inL = strlen(filename); + size_t const outL = inL + 5; + void* outSpace = malloc_orDie(outL); + memset(outSpace, 0, outL); + strcat(outSpace, filename); + strcat(outSpace, ".zst"); + return (char*)outSpace; +} + +#define CLEVEL_DEFAULT 5 +int main(int argc, const char** argv) +{ + const char* const exeName = argv[0]; + if (argc<3 || argc>4) { + printf("wrong arguments \n"); + printf("usage: \n"); + printf("%s FILE FRAME_SIZE [LEVEL] \n", exeName); + return 1; + } + + { const char* const inFileName = argv[1]; + unsigned const frameSize = (unsigned)atoi(argv[2]); + int const cLevel = (argc==4) ? atoi(argv[3]) : CLEVEL_DEFAULT; + + char* const outFileName = createOutFilename_orDie(inFileName); + compressFile_orDie(inFileName, outFileName, cLevel, frameSize); + free(outFileName); + } + + return 0; +} diff --git a/3rdparty/zstd/contrib/seekable_format/examples/seekable_decompression.c b/3rdparty/zstd/contrib/seekable_format/examples/seekable_decompression.c new file mode 100644 index 00000000000..7edbca87d0e --- /dev/null +++ b/3rdparty/zstd/contrib/seekable_format/examples/seekable_decompression.c @@ -0,0 +1,141 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ + + +#include <stdlib.h> // malloc, exit +#include <stdio.h> // fprintf, perror, feof +#include <string.h> // strerror +#include <errno.h> // errno +#define ZSTD_STATIC_LINKING_ONLY +#include <zstd.h> // presumes zstd library is installed +#include <zstd_errors.h> + +#include "../zstd_seekable.h" + +#define MIN(a, b) ((a) < (b) ? (a) : (b)) + +static void* malloc_orDie(size_t size) +{ + void* const buff = malloc(size); + if (buff) return buff; + /* error */ + perror("malloc"); + exit(1); +} + +static void* realloc_orDie(void* ptr, size_t size) +{ + ptr = realloc(ptr, size); + if (ptr) return ptr; + /* error */ + perror("realloc"); + exit(1); +} + +static FILE* fopen_orDie(const char *filename, const char *instruction) +{ + FILE* const inFile = fopen(filename, instruction); + if (inFile) return inFile; + /* error */ + perror(filename); + exit(3); +} + +static size_t fread_orDie(void* buffer, size_t sizeToRead, FILE* file) +{ + size_t const readSize = fread(buffer, 1, sizeToRead, file); + if (readSize == sizeToRead) return readSize; /* good */ + if (feof(file)) return readSize; /* good, reached end of file */ + /* error */ + perror("fread"); + exit(4); +} + +static size_t fwrite_orDie(const void* buffer, size_t sizeToWrite, FILE* file) +{ + size_t const writtenSize = fwrite(buffer, 1, sizeToWrite, file); + if (writtenSize == sizeToWrite) return sizeToWrite; /* good */ + /* error */ + perror("fwrite"); + exit(5); +} + +static size_t fclose_orDie(FILE* file) +{ + if (!fclose(file)) return 0; + /* error */ + perror("fclose"); + exit(6); +} + +static void fseek_orDie(FILE* file, long int offset, int origin) { + if (!fseek(file, offset, origin)) { + if (!fflush(file)) return; + } + /* error */ + perror("fseek"); + exit(7); +} + + +static void decompressFile_orDie(const char* fname, off_t startOffset, off_t endOffset) +{ + FILE* const fin = fopen_orDie(fname, "rb"); + FILE* const fout = stdout; + size_t const buffOutSize = ZSTD_DStreamOutSize(); /* Guarantee to successfully flush at least one complete compressed block in all circumstances. */ + void* const buffOut = malloc_orDie(buffOutSize); + + ZSTD_seekable* const seekable = ZSTD_seekable_create(); + if (seekable==NULL) { fprintf(stderr, "ZSTD_seekable_create() error \n"); exit(10); } + + size_t const initResult = ZSTD_seekable_initFile(seekable, fin); + if (ZSTD_isError(initResult)) { fprintf(stderr, "ZSTD_seekable_init() error : %s \n", ZSTD_getErrorName(initResult)); exit(11); } + + while (startOffset < endOffset) { + size_t const result = ZSTD_seekable_decompress(seekable, buffOut, MIN(endOffset - startOffset, buffOutSize), startOffset); + if (!result) { + break; + } + + if (ZSTD_isError(result)) { + fprintf(stderr, "ZSTD_seekable_decompress() error : %s \n", + ZSTD_getErrorName(result)); + exit(12); + } + fwrite_orDie(buffOut, result, fout); + startOffset += result; + } + + ZSTD_seekable_free(seekable); + fclose_orDie(fin); + fclose_orDie(fout); + free(buffOut); +} + + +int main(int argc, const char** argv) +{ + const char* const exeName = argv[0]; + + if (argc!=4) { + fprintf(stderr, "wrong arguments\n"); + fprintf(stderr, "usage:\n"); + fprintf(stderr, "%s FILE START END\n", exeName); + return 1; + } + + { + const char* const inFilename = argv[1]; + off_t const startOffset = atoll(argv[2]); + off_t const endOffset = atoll(argv[3]); + decompressFile_orDie(inFilename, startOffset, endOffset); + } + + return 0; +} diff --git a/3rdparty/zstd/contrib/seekable_format/examples/seekable_decompression_mem.c b/3rdparty/zstd/contrib/seekable_format/examples/seekable_decompression_mem.c new file mode 100644 index 00000000000..44a06fbbfb0 --- /dev/null +++ b/3rdparty/zstd/contrib/seekable_format/examples/seekable_decompression_mem.c @@ -0,0 +1,147 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ + + +#include <stdlib.h> // malloc, exit +#include <stdio.h> // fprintf, perror, feof +#include <string.h> // strerror +#include <errno.h> // errno +#define ZSTD_STATIC_LINKING_ONLY +#include <zstd.h> // presumes zstd library is installed +#include <zstd_errors.h> + +#include "zstd_seekable.h" + +#define MIN(a, b) ((a) < (b) ? (a) : (b)) + +#define MAX_FILE_SIZE (8 * 1024 * 1024) + +static void* malloc_orDie(size_t size) +{ + void* const buff = malloc(size); + if (buff) return buff; + /* error */ + perror("malloc"); + exit(1); +} + +static void* realloc_orDie(void* ptr, size_t size) +{ + ptr = realloc(ptr, size); + if (ptr) return ptr; + /* error */ + perror("realloc"); + exit(1); +} + +static FILE* fopen_orDie(const char *filename, const char *instruction) +{ + FILE* const inFile = fopen(filename, instruction); + if (inFile) return inFile; + /* error */ + perror(filename); + exit(3); +} + +static size_t fread_orDie(void* buffer, size_t sizeToRead, FILE* file) +{ + size_t const readSize = fread(buffer, 1, sizeToRead, file); + if (readSize == sizeToRead) return readSize; /* good */ + if (feof(file)) return readSize; /* good, reached end of file */ + /* error */ + perror("fread"); + exit(4); +} + +static size_t fwrite_orDie(const void* buffer, size_t sizeToWrite, FILE* file) +{ + size_t const writtenSize = fwrite(buffer, 1, sizeToWrite, file); + if (writtenSize == sizeToWrite) return sizeToWrite; /* good */ + /* error */ + perror("fwrite"); + exit(5); +} + +static size_t fclose_orDie(FILE* file) +{ + if (!fclose(file)) return 0; + /* error */ + perror("fclose"); + exit(6); +} + +static void fseek_orDie(FILE* file, long int offset, int origin) { + if (!fseek(file, offset, origin)) { + if (!fflush(file)) return; + } + /* error */ + perror("fseek"); + exit(7); +} + + +static void decompressFile_orDie(const char* fname, off_t startOffset, off_t endOffset) +{ + FILE* const fin = fopen_orDie(fname, "rb"); + FILE* const fout = stdout; + // Just for demo purposes, assume file is <= MAX_FILE_SIZE + void* const buffIn = malloc_orDie(MAX_FILE_SIZE); + size_t const inSize = fread_orDie(buffIn, MAX_FILE_SIZE, fin); + size_t const buffOutSize = ZSTD_DStreamOutSize(); /* Guarantee to successfully flush at least one complete compressed block in all circumstances. */ + void* const buffOut = malloc_orDie(buffOutSize); + + ZSTD_seekable* const seekable = ZSTD_seekable_create(); + if (seekable==NULL) { fprintf(stderr, "ZSTD_seekable_create() error \n"); exit(10); } + + size_t const initResult = ZSTD_seekable_initBuff(seekable, buffIn, inSize); + if (ZSTD_isError(initResult)) { fprintf(stderr, "ZSTD_seekable_init() error : %s \n", ZSTD_getErrorName(initResult)); exit(11); } + + while (startOffset < endOffset) { + size_t const result = ZSTD_seekable_decompress(seekable, buffOut, MIN(endOffset - startOffset, buffOutSize), startOffset); + if (!result) { + break; + } + + if (ZSTD_isError(result)) { + fprintf(stderr, "ZSTD_seekable_decompress() error : %s \n", + ZSTD_getErrorName(result)); + exit(12); + } + fwrite_orDie(buffOut, result, fout); + startOffset += result; + } + + ZSTD_seekable_free(seekable); + fclose_orDie(fin); + fclose_orDie(fout); + free(buffIn); + free(buffOut); +} + + +int main(int argc, const char** argv) +{ + const char* const exeName = argv[0]; + + if (argc!=4) { + fprintf(stderr, "wrong arguments\n"); + fprintf(stderr, "usage:\n"); + fprintf(stderr, "%s FILE START END\n", exeName); + return 1; + } + + { + const char* const inFilename = argv[1]; + off_t const startOffset = atoll(argv[2]); + off_t const endOffset = atoll(argv[3]); + decompressFile_orDie(inFilename, startOffset, endOffset); + } + + return 0; +} diff --git a/3rdparty/zstd/contrib/seekable_format/tests/.gitignore b/3rdparty/zstd/contrib/seekable_format/tests/.gitignore new file mode 100644 index 00000000000..f831eaf37c4 --- /dev/null +++ b/3rdparty/zstd/contrib/seekable_format/tests/.gitignore @@ -0,0 +1 @@ +seekable_tests diff --git a/3rdparty/zstd/contrib/seekable_format/tests/Makefile b/3rdparty/zstd/contrib/seekable_format/tests/Makefile new file mode 100644 index 00000000000..a81f2229fa5 --- /dev/null +++ b/3rdparty/zstd/contrib/seekable_format/tests/Makefile @@ -0,0 +1,38 @@ +# ################################################################ +# Copyright (c) Meta Platforms, Inc. and affiliates. +# All rights reserved. +# +# This source code is licensed under both the BSD-style license (found in the +# LICENSE file in the root directory of this source tree) and the GPLv2 (found +# in the COPYING file in the root directory of this source tree). +# ################################################################ + +# This Makefile presumes libzstd is built, using `make` in / or /lib/ + +ZSTDLIB_PATH = ../../../lib +ZSTDLIB_NAME = libzstd.a +ZSTDLIB = $(ZSTDLIB_PATH)/$(ZSTDLIB_NAME) + +CPPFLAGS += -DXXH_NAMESPACE=ZSTD_ -I../ -I$(ZSTDLIB_PATH) -I$(ZSTDLIB_PATH)/common + +CFLAGS ?= -O3 +CFLAGS += -g -Wall -Wextra -Wcast-qual -Wcast-align -Wconversion \ + -Wformat=2 -Wstrict-aliasing=1 + +SEEKABLE_OBJS = ../zstdseek_compress.c ../zstdseek_decompress.c $(ZSTDLIB) + +.PHONY: default clean test +default: test + +test: seekable_tests + ./seekable_tests + +$(ZSTDLIB): + $(MAKE) -C $(ZSTDLIB_PATH) $(ZSTDLIB_NAME) + +seekable_tests : $(SEEKABLE_OBJS) + +clean: + @$(RM) core *.o tmp* result* *.zst \ + seekable_tests + @echo Cleaning completed diff --git a/3rdparty/zstd/contrib/seekable_format/tests/seekable_tests.c b/3rdparty/zstd/contrib/seekable_format/tests/seekable_tests.c new file mode 100644 index 00000000000..f89bdc930cd --- /dev/null +++ b/3rdparty/zstd/contrib/seekable_format/tests/seekable_tests.c @@ -0,0 +1,363 @@ +#include <stddef.h> +#include <stdint.h> +#include <stdlib.h> // malloc +#include <stdio.h> +#include <assert.h> +#include <string.h> + +#include "../zstd_seekable.h" + + +/* ZSTD_seekable_customFile implementation that reads/seeks a buffer while keeping track of total bytes read */ +typedef struct { + const void *ptr; + size_t size; + size_t pos; + size_t totalRead; +} buffWrapperWithTotal_t; + +static int readBuffWithTotal(void* opaque, void* buffer, size_t n) +{ + buffWrapperWithTotal_t* const buff = (buffWrapperWithTotal_t*)opaque; + assert(buff != NULL); + if (buff->pos + n > buff->size) return -1; + memcpy(buffer, (const char*)buff->ptr + buff->pos, n); + buff->pos += n; + buff->totalRead += n; + return 0; +} + +static int seekBuffWithTotal(void* opaque, long long offset, int origin) +{ + buffWrapperWithTotal_t* const buff = (buffWrapperWithTotal_t*) opaque; + unsigned long long newOffset; + assert(buff != NULL); + switch (origin) { + case SEEK_SET: + assert(offset >= 0); + newOffset = (unsigned long long)offset; + break; + case SEEK_CUR: + newOffset = (unsigned long long)((long long)buff->pos + offset); + break; + case SEEK_END: + newOffset = (unsigned long long)((long long)buff->size + offset); + break; + default: + assert(0); /* not possible */ + } + if (newOffset > buff->size) { + return -1; + } + buff->pos = newOffset; + return 0; +} + +/* Basic unit tests for zstd seekable format */ +int main(int argc, const char** argv) +{ + unsigned testNb = 1; + (void)argc; (void)argv; + printf("Beginning zstd seekable format tests...\n"); + + printf("Test %u - simple round trip: ", testNb++); + { size_t const inSize = 4000; + void* const inBuffer = malloc(inSize); + assert(inBuffer != NULL); + + size_t const seekCapacity = 5000; + void* const seekBuffer = malloc(seekCapacity); + assert(seekBuffer != NULL); + size_t seekSize; + + size_t const outCapacity = inSize; + void* const outBuffer = malloc(outCapacity); + assert(outBuffer != NULL); + + ZSTD_seekable_CStream* const zscs = ZSTD_seekable_createCStream(); + assert(zscs != NULL); + + { size_t const initStatus = ZSTD_seekable_initCStream(zscs, 9, 0 /* checksumFlag */, (unsigned)inSize /* maxFrameSize */); + assert(!ZSTD_isError(initStatus)); + } + + { ZSTD_outBuffer outb = { .dst=seekBuffer, .pos=0, .size=seekCapacity }; + ZSTD_inBuffer inb = { .src=inBuffer, .pos=0, .size=inSize }; + + size_t const cStatus = ZSTD_seekable_compressStream(zscs, &outb, &inb); + assert(!ZSTD_isError(cStatus)); + assert(inb.pos == inb.size); + + size_t const endStatus = ZSTD_seekable_endStream(zscs, &outb); + assert(!ZSTD_isError(endStatus)); + seekSize = outb.pos; + } + + ZSTD_seekable* const stream = ZSTD_seekable_create(); + assert(stream != NULL); + { size_t const initStatus = ZSTD_seekable_initBuff(stream, seekBuffer, seekSize); + assert(!ZSTD_isError(initStatus)); } + + { size_t const decStatus = ZSTD_seekable_decompress(stream, outBuffer, outCapacity, 0); + assert(decStatus == inSize); } + + /* unit test ZSTD_seekTable functions */ + ZSTD_seekTable* const zst = ZSTD_seekTable_create_fromSeekable(stream); + assert(zst != NULL); + + unsigned const nbFrames = ZSTD_seekTable_getNumFrames(zst); + assert(nbFrames > 0); + + unsigned long long const frame0Offset = ZSTD_seekTable_getFrameCompressedOffset(zst, 0); + assert(frame0Offset == 0); + + unsigned long long const content0Offset = ZSTD_seekTable_getFrameDecompressedOffset(zst, 0); + assert(content0Offset == 0); + + size_t const cSize = ZSTD_seekTable_getFrameCompressedSize(zst, 0); + assert(!ZSTD_isError(cSize)); + assert(cSize <= seekCapacity); + + size_t const origSize = ZSTD_seekTable_getFrameDecompressedSize(zst, 0); + assert(origSize == inSize); + + unsigned const fo1idx = ZSTD_seekTable_offsetToFrameIndex(zst, 1); + assert(fo1idx == 0); + + free(inBuffer); + free(seekBuffer); + free(outBuffer); + ZSTD_seekable_freeCStream(zscs); + ZSTD_seekTable_free(zst); + ZSTD_seekable_free(stream); + } + printf("Success!\n"); + + + printf("Test %u - check that seekable decompress does not hang: ", testNb++); + { /* Github issue #2335 */ + const size_t compressed_size = 17; + const uint8_t compressed_data[17] = { + '^', + '*', + 'M', + '\x18', + '\t', + '\x00', + '\x00', + '\x00', + '\x00', + '\x00', + '\x00', + '\x00', + (uint8_t)('\x03'), + (uint8_t)('\xb1'), + (uint8_t)('\xea'), + (uint8_t)('\x92'), + (uint8_t)('\x8f'), + }; + const size_t uncompressed_size = 32; + uint8_t uncompressed_data[32]; + + ZSTD_seekable* const stream = ZSTD_seekable_create(); + assert(stream != NULL); + { size_t const status = ZSTD_seekable_initBuff(stream, compressed_data, compressed_size); + if (ZSTD_isError(status)) { + ZSTD_seekable_free(stream); + goto _test_error; + } } + + /* Should return an error, but not hang */ + { const size_t offset = 2; + size_t const status = ZSTD_seekable_decompress(stream, uncompressed_data, uncompressed_size, offset); + if (!ZSTD_isError(status)) { + ZSTD_seekable_free(stream); + goto _test_error; + } } + + ZSTD_seekable_free(stream); + } + printf("Success!\n"); + + printf("Test %u - check #2 that seekable decompress does not hang: ", testNb++); + { /* Github issue #FIXME */ + const size_t compressed_size = 27; + const uint8_t compressed_data[27] = { + (uint8_t)'\x28', + (uint8_t)'\xb5', + (uint8_t)'\x2f', + (uint8_t)'\xfd', + (uint8_t)'\x00', + (uint8_t)'\x32', + (uint8_t)'\x91', + (uint8_t)'\x00', + (uint8_t)'\x00', + (uint8_t)'\x00', + (uint8_t)'\x5e', + (uint8_t)'\x2a', + (uint8_t)'\x4d', + (uint8_t)'\x18', + (uint8_t)'\x09', + (uint8_t)'\x00', + (uint8_t)'\x00', + (uint8_t)'\x00', + (uint8_t)'\x00', + (uint8_t)'\x00', + (uint8_t)'\x00', + (uint8_t)'\x00', + (uint8_t)'\x00', + (uint8_t)'\xb1', + (uint8_t)'\xea', + (uint8_t)'\x92', + (uint8_t)'\x8f', + }; + const size_t uncompressed_size = 400; + uint8_t uncompressed_data[400]; + + ZSTD_seekable* stream = ZSTD_seekable_create(); + size_t status = ZSTD_seekable_initBuff(stream, compressed_data, compressed_size); + if (ZSTD_isError(status)) { + ZSTD_seekable_free(stream); + goto _test_error; + } + + const size_t offset = 2; + /* Should return an error, but not hang */ + status = ZSTD_seekable_decompress(stream, uncompressed_data, uncompressed_size, offset); + if (!ZSTD_isError(status)) { + ZSTD_seekable_free(stream); + goto _test_error; + } + + ZSTD_seekable_free(stream); + } + printf("Success!\n"); + + + printf("Test %u - check ZSTD magic in compressing empty string: ", testNb++); + { // compressing empty string should return a zstd header + size_t const capacity = 255; + char* inBuffer = malloc(capacity); + assert(inBuffer != NULL); + inBuffer[0] = '\0'; + void* const outBuffer = malloc(capacity); + assert(outBuffer != NULL); + + ZSTD_seekable_CStream *s = ZSTD_seekable_createCStream(); + ZSTD_seekable_initCStream(s, 1, 1, 255); + + ZSTD_inBuffer input = { .src=inBuffer, .pos=0, .size=0 }; + ZSTD_outBuffer output = { .dst=outBuffer, .pos=0, .size=capacity }; + + ZSTD_seekable_compressStream(s, &output, &input); + ZSTD_seekable_endStream(s, &output); + + if((((char*)output.dst)[0] != '\x28') | (((char*)output.dst)[1] != '\xb5') | (((char*)output.dst)[2] != '\x2f') | (((char*)output.dst)[3] != '\xfd')) { + printf("%#02x %#02x %#02x %#02x\n", ((char*)output.dst)[0], ((char*)output.dst)[1] , ((char*)output.dst)[2] , ((char*)output.dst)[3] ); + + free(inBuffer); + free(outBuffer); + ZSTD_seekable_freeCStream(s); + goto _test_error; + } + + free(inBuffer); + free(outBuffer); + ZSTD_seekable_freeCStream(s); + } + printf("Success!\n"); + + + printf("Test %u - multiple decompress calls: ", testNb++); + { char const inBuffer[] = "Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt"; + size_t const inSize = sizeof(inBuffer); + + size_t const seekCapacity = 5000; + void* const seekBuffer = malloc(seekCapacity); + assert(seekBuffer != NULL); + size_t seekSize; + + size_t const outCapacity = inSize; + char* const outBuffer = malloc(outCapacity); + assert(outBuffer != NULL); + + ZSTD_seekable_CStream* const zscs = ZSTD_seekable_createCStream(); + assert(zscs != NULL); + + /* compress test data with a small frame size to ensure multiple frames in the output */ + unsigned const maxFrameSize = 40; + { size_t const initStatus = ZSTD_seekable_initCStream(zscs, 9, 0 /* checksumFlag */, maxFrameSize); + assert(!ZSTD_isError(initStatus)); + } + + { ZSTD_outBuffer outb = { .dst=seekBuffer, .pos=0, .size=seekCapacity }; + ZSTD_inBuffer inb = { .src=inBuffer, .pos=0, .size=inSize }; + + while (inb.pos < inb.size) { + size_t const cStatus = ZSTD_seekable_compressStream(zscs, &outb, &inb); + assert(!ZSTD_isError(cStatus)); + } + + size_t const endStatus = ZSTD_seekable_endStream(zscs, &outb); + assert(!ZSTD_isError(endStatus)); + seekSize = outb.pos; + } + + ZSTD_seekable* const stream = ZSTD_seekable_create(); + assert(stream != NULL); + buffWrapperWithTotal_t buffWrapper = {seekBuffer, seekSize, 0, 0}; + { ZSTD_seekable_customFile srcFile = {&buffWrapper, &readBuffWithTotal, &seekBuffWithTotal}; + size_t const initStatus = ZSTD_seekable_initAdvanced(stream, srcFile); + assert(!ZSTD_isError(initStatus)); } + + /* Perform a series of small reads and seeks (repeatedly read 1 byte and skip 1 byte) + and check that we didn't reread input data unnecessarily */ + size_t pos; + for (pos = 0; pos < inSize; pos += 2) { + size_t const decStatus = ZSTD_seekable_decompress(stream, outBuffer, 1, pos); + if (decStatus != 1 || outBuffer[0] != inBuffer[pos]) { + goto _test_error; + } + } + if (buffWrapper.totalRead > seekSize) { + /* We read more than the compressed size, meaning there were some rereads. + This is unneeded because we only seeked forward. */ + printf("Too much data read: %zu read, with compressed size %zu\n", buffWrapper.totalRead, seekSize); + goto _test_error; + } + + /* Perform some reads and seeks to ensure correctness */ + struct { + size_t offset; + size_t size; + } const tests[] = { /* Assume the frame size is 40 */ + {20, 40}, /* read partial data from two frames */ + {60, 10}, /* continue reading from the same offset */ + {50, 20}, /* seek backward within the same frame */ + {10, 10}, /* seek backward to a different frame */ + {25, 10}, /* seek forward within the same frame */ + {60, 10}, /* seek forward to a different frame */ + }; + size_t idx; + for (idx = 0; idx < sizeof(tests) / sizeof(tests[0]); idx++) { + size_t const decStatus = ZSTD_seekable_decompress(stream, outBuffer, tests[idx].size, tests[idx].offset); + if (decStatus != tests[idx].size || memcmp(outBuffer, inBuffer + tests[idx].offset, tests[idx].size) != 0) { + goto _test_error; + } + } + + free(seekBuffer); + free(outBuffer); + ZSTD_seekable_freeCStream(zscs); + ZSTD_seekable_free(stream); + } + printf("Success!\n"); + + /* TODO: Add more tests */ + printf("Finished tests\n"); + return 0; + +_test_error: + printf("test failed! Exiting..\n"); + return 1; +} diff --git a/3rdparty/zstd/contrib/seekable_format/zstd_seekable.h b/3rdparty/zstd/contrib/seekable_format/zstd_seekable.h new file mode 100644 index 00000000000..a0e5e3573d0 --- /dev/null +++ b/3rdparty/zstd/contrib/seekable_format/zstd_seekable.h @@ -0,0 +1,226 @@ +#ifndef SEEKABLE_H +#define SEEKABLE_H + +#if defined (__cplusplus) +extern "C" { +#endif + +#include <stdio.h> +#include "zstd.h" /* ZSTDLIB_API */ + + +#define ZSTD_seekTableFooterSize 9 + +#define ZSTD_SEEKABLE_MAGICNUMBER 0x8F92EAB1 + +#define ZSTD_SEEKABLE_MAXFRAMES 0x8000000U + +/* Limit maximum size to avoid potential issues storing the compressed size */ +#define ZSTD_SEEKABLE_MAX_FRAME_DECOMPRESSED_SIZE 0x40000000U + +/*-**************************************************************************** +* Seekable Format +* +* The seekable format splits the compressed data into a series of "frames", +* each compressed individually so that decompression of a section in the +* middle of an archive only requires zstd to decompress at most a frame's +* worth of extra data, instead of the entire archive. +******************************************************************************/ + +typedef struct ZSTD_seekable_CStream_s ZSTD_seekable_CStream; +typedef struct ZSTD_seekable_s ZSTD_seekable; +typedef struct ZSTD_seekTable_s ZSTD_seekTable; + +/*-**************************************************************************** +* Seekable compression - HowTo +* A ZSTD_seekable_CStream object is required to tracking streaming operation. +* Use ZSTD_seekable_createCStream() and ZSTD_seekable_freeCStream() to create/ +* release resources. +* +* Streaming objects are reusable to avoid allocation and deallocation, +* to start a new compression operation call ZSTD_seekable_initCStream() on the +* compressor. +* +* Data streamed to the seekable compressor will automatically be split into +* frames of size `maxFrameSize` (provided in ZSTD_seekable_initCStream()), +* or if none is provided, will be cut off whenever ZSTD_seekable_endFrame() is +* called or when the default maximum frame size (2GB) is reached. +* +* Use ZSTD_seekable_initCStream() to initialize a ZSTD_seekable_CStream object +* for a new compression operation. +* - `maxFrameSize` indicates the size at which to automatically start a new +* seekable frame. +* `maxFrameSize == 0` implies the default maximum size. +* Smaller frame sizes allow faster decompression of small segments, +* since retrieving a single byte requires decompression of +* the full frame where the byte belongs. +* In general, size the frames to roughly correspond to +* the access granularity (when it's known). +* But small sizes also reduce compression ratio. +* Avoid really tiny frame sizes (< 1 KB), +* that would hurt compression ratio considerably. +* - `checksumFlag` indicates whether or not the seek table should include frame +* checksums on the uncompressed data for verification. +* @return : a size hint for input to provide for compression, or an error code +* checkable with ZSTD_isError() +* +* Use ZSTD_seekable_compressStream() repetitively to consume input stream. +* The function will automatically update both `pos` fields. +* Note that it may not consume the entire input, in which case `pos < size`, +* and it's up to the caller to present again remaining data. +* @return : a size hint, preferred nb of bytes to use as input for next +* function call or an error code, which can be tested using +* ZSTD_isError(). +* Note 1 : it's just a hint, to help latency a little, any other +* value will work fine. +* +* At any time, call ZSTD_seekable_endFrame() to end the current frame and +* start a new one. +* +* ZSTD_seekable_endStream() will end the current frame, and then write the seek +* table so that decompressors can efficiently find compressed frames. +* ZSTD_seekable_endStream() may return a number > 0 if it was unable to flush +* all the necessary data to `output`. In this case, it should be called again +* until all remaining data is flushed out and 0 is returned. +******************************************************************************/ + +/*===== Seekable compressor management =====*/ +ZSTDLIB_API ZSTD_seekable_CStream* ZSTD_seekable_createCStream(void); +ZSTDLIB_API size_t ZSTD_seekable_freeCStream(ZSTD_seekable_CStream* zcs); + +/*===== Seekable compression functions =====*/ +ZSTDLIB_API size_t ZSTD_seekable_initCStream(ZSTD_seekable_CStream* zcs, int compressionLevel, int checksumFlag, unsigned maxFrameSize); +ZSTDLIB_API size_t ZSTD_seekable_compressStream(ZSTD_seekable_CStream* zcs, ZSTD_outBuffer* output, ZSTD_inBuffer* input); +ZSTDLIB_API size_t ZSTD_seekable_endFrame(ZSTD_seekable_CStream* zcs, ZSTD_outBuffer* output); +ZSTDLIB_API size_t ZSTD_seekable_endStream(ZSTD_seekable_CStream* zcs, ZSTD_outBuffer* output); + +/*= Raw seek table API + * These functions allow for the seek table to be constructed directly. + * This table can then be appended to a file of concatenated frames. + * This allows the frames to be compressed independently, even in parallel, + * and compiled together afterward into a seekable archive. + * + * Use ZSTD_seekable_createFrameLog() to allocate and initialize a tracking + * structure. + * + * Call ZSTD_seekable_logFrame() once for each frame in the archive. + * checksum is optional, and will not be used if checksumFlag was 0 when the + * frame log was created. If present, it should be the least significant 32 + * bits of the XXH64 hash of the uncompressed data. + * + * Call ZSTD_seekable_writeSeekTable to serialize the data into a seek table. + * If the entire table was written, the return value will be 0. Otherwise, + * it will be equal to the number of bytes left to write. */ +typedef struct ZSTD_frameLog_s ZSTD_frameLog; +ZSTDLIB_API ZSTD_frameLog* ZSTD_seekable_createFrameLog(int checksumFlag); +ZSTDLIB_API size_t ZSTD_seekable_freeFrameLog(ZSTD_frameLog* fl); +ZSTDLIB_API size_t ZSTD_seekable_logFrame(ZSTD_frameLog* fl, unsigned compressedSize, unsigned decompressedSize, unsigned checksum); +ZSTDLIB_API size_t ZSTD_seekable_writeSeekTable(ZSTD_frameLog* fl, ZSTD_outBuffer* output); + + +/*-**************************************************************************** +* Seekable decompression - HowTo +* A ZSTD_seekable object is required to tracking the seekTable. +* +* Call ZSTD_seekable_init* to initialize a ZSTD_seekable object with the +* the seek table provided in the input. +* There are three modes for ZSTD_seekable_init: +* - ZSTD_seekable_initBuff() : An in-memory API. The data contained in +* `src` should be the entire seekable file, including the seek table. +* `src` should be kept alive and unmodified until the ZSTD_seekable object +* is freed or reset. +* - ZSTD_seekable_initFile() : A simplified file API using stdio. fread and +* fseek will be used to access the required data for building the seek +* table and doing decompression operations. `src` should not be closed +* or modified until the ZSTD_seekable object is freed or reset. +* - ZSTD_seekable_initAdvanced() : A general API allowing the client to +* provide its own read and seek callbacks. +* + ZSTD_seekable_read() : read exactly `n` bytes into `buffer`. +* Premature EOF should be treated as an error. +* + ZSTD_seekable_seek() : seek the read head to `offset` from `origin`, +* where origin is either SEEK_SET (beginning of +* file), or SEEK_END (end of file). +* Both functions should return a non-negative value in case of success, and a +* negative value in case of failure. If implementing using this API and +* stdio, be careful with files larger than 4GB and fseek. All of these +* functions return an error code checkable with ZSTD_isError(). +* +* Call ZSTD_seekable_decompress to decompress `dstSize` bytes at decompressed +* offset `offset`. ZSTD_seekable_decompress may have to decompress the entire +* prefix of the frame before the desired data if it has not already processed +* this section. If ZSTD_seekable_decompress is called multiple times for a +* consecutive range of data, it will efficiently retain the decompressor object +* and avoid redecompressing frame prefixes. The return value is the number of +* bytes decompressed, or an error code checkable with ZSTD_isError(). +* +* The seek table access functions can be used to obtain the data contained +* in the seek table. If frameIndex is larger than the value returned by +* ZSTD_seekable_getNumFrames(), they will return error codes checkable with +* ZSTD_isError(). Note that since the offset access functions return +* unsigned long long instead of size_t, in this case they will instead return +* the value ZSTD_SEEKABLE_FRAMEINDEX_TOOLARGE. +******************************************************************************/ + +/*===== Seekable decompressor management =====*/ +ZSTDLIB_API ZSTD_seekable* ZSTD_seekable_create(void); +ZSTDLIB_API size_t ZSTD_seekable_free(ZSTD_seekable* zs); + +/*===== Seekable decompression functions =====*/ +ZSTDLIB_API size_t ZSTD_seekable_initBuff(ZSTD_seekable* zs, const void* src, size_t srcSize); +ZSTDLIB_API size_t ZSTD_seekable_initFile(ZSTD_seekable* zs, FILE* src); +ZSTDLIB_API size_t ZSTD_seekable_decompress(ZSTD_seekable* zs, void* dst, size_t dstSize, unsigned long long offset); +ZSTDLIB_API size_t ZSTD_seekable_decompressFrame(ZSTD_seekable* zs, void* dst, size_t dstSize, unsigned frameIndex); + +#define ZSTD_SEEKABLE_FRAMEINDEX_TOOLARGE (0ULL-2) +/*===== Seekable seek table access functions =====*/ +ZSTDLIB_API unsigned ZSTD_seekable_getNumFrames(const ZSTD_seekable* zs); +ZSTDLIB_API unsigned long long ZSTD_seekable_getFrameCompressedOffset(const ZSTD_seekable* zs, unsigned frameIndex); +ZSTDLIB_API unsigned long long ZSTD_seekable_getFrameDecompressedOffset(const ZSTD_seekable* zs, unsigned frameIndex); +ZSTDLIB_API size_t ZSTD_seekable_getFrameCompressedSize(const ZSTD_seekable* zs, unsigned frameIndex); +ZSTDLIB_API size_t ZSTD_seekable_getFrameDecompressedSize(const ZSTD_seekable* zs, unsigned frameIndex); +ZSTDLIB_API unsigned ZSTD_seekable_offsetToFrameIndex(const ZSTD_seekable* zs, unsigned long long offset); + + +/*-**************************************************************************** +* Direct exploitation of the seekTable +* +* Memory constrained use cases that manage multiple archives +* benefit from retaining multiple archive seek tables +* without retaining a ZSTD_seekable instance for each. +* +* Below API allow the above-mentioned use cases +* to initialize a ZSTD_seekable, extract its (smaller) ZSTD_seekTable, +* then throw the ZSTD_seekable away to save memory. +* +* Standard ZSTD operations can then be used +* to decompress frames based on seek table offsets. +******************************************************************************/ + +/*===== Independent seek table management =====*/ +ZSTDLIB_API ZSTD_seekTable* ZSTD_seekTable_create_fromSeekable(const ZSTD_seekable* zs); +ZSTDLIB_API size_t ZSTD_seekTable_free(ZSTD_seekTable* st); + +/*===== Direct seek table access functions =====*/ +ZSTDLIB_API unsigned ZSTD_seekTable_getNumFrames(const ZSTD_seekTable* st); +ZSTDLIB_API unsigned long long ZSTD_seekTable_getFrameCompressedOffset(const ZSTD_seekTable* st, unsigned frameIndex); +ZSTDLIB_API unsigned long long ZSTD_seekTable_getFrameDecompressedOffset(const ZSTD_seekTable* st, unsigned frameIndex); +ZSTDLIB_API size_t ZSTD_seekTable_getFrameCompressedSize(const ZSTD_seekTable* st, unsigned frameIndex); +ZSTDLIB_API size_t ZSTD_seekTable_getFrameDecompressedSize(const ZSTD_seekTable* st, unsigned frameIndex); +ZSTDLIB_API unsigned ZSTD_seekTable_offsetToFrameIndex(const ZSTD_seekTable* st, unsigned long long offset); + + +/*===== Seekable advanced I/O API =====*/ +typedef int(ZSTD_seekable_read)(void* opaque, void* buffer, size_t n); +typedef int(ZSTD_seekable_seek)(void* opaque, long long offset, int origin); +typedef struct { + void* opaque; + ZSTD_seekable_read* read; + ZSTD_seekable_seek* seek; +} ZSTD_seekable_customFile; +ZSTDLIB_API size_t ZSTD_seekable_initAdvanced(ZSTD_seekable* zs, ZSTD_seekable_customFile src); + +#if defined (__cplusplus) +} +#endif + +#endif diff --git a/3rdparty/zstd/contrib/seekable_format/zstd_seekable_compression_format.md b/3rdparty/zstd/contrib/seekable_format/zstd_seekable_compression_format.md new file mode 100644 index 00000000000..7bd0790e81b --- /dev/null +++ b/3rdparty/zstd/contrib/seekable_format/zstd_seekable_compression_format.md @@ -0,0 +1,116 @@ +# Zstandard Seekable Format + +### Notices + +Copyright (c) Meta Platforms, Inc. and affiliates. + +Permission is granted to copy and distribute this document +for any purpose and without charge, +including translations into other languages +and incorporation into compilations, +provided that the copyright notice and this notice are preserved, +and that any substantive changes or deletions from the original +are clearly marked. +Distribution of this document is unlimited. + +### Version +0.1.0 (11/04/17) + +## Introduction +This document defines a format for compressed data to be stored so that subranges of the data can be efficiently decompressed without requiring the entire document to be decompressed. +This is done by splitting up the input data into frames, +each of which are compressed independently, +and so can be decompressed independently. +Decompression then takes advantage of a provided 'seek table', which allows the decompressor to immediately jump to the desired data. This is done in a way that is compatible with the original Zstandard format by placing the seek table in a Zstandard skippable frame. + +### Overall conventions +In this document: +- square brackets i.e. `[` and `]` are used to indicate optional fields or parameters. +- the naming convention for identifiers is `Mixed_Case_With_Underscores` +- All numeric fields are little-endian unless specified otherwise + +## Format + +The format consists of a number of frames (Zstandard compressed frames and skippable frames), followed by a final skippable frame at the end containing the seek table. + +### Seek Table Format +The structure of the seek table frame is as follows: + +|`Skippable_Magic_Number`|`Frame_Size`|`[Seek_Table_Entries]`|`Seek_Table_Footer`| +|------------------------|------------|----------------------|-------------------| +| 4 bytes | 4 bytes | 8-12 bytes each | 9 bytes | + +__`Skippable_Magic_Number`__ + +Value : 0x184D2A5E. +This is for compatibility with [Zstandard skippable frames]. +Since it is legal for other Zstandard skippable frames to use the same +magic number, it is not recommended for a decoder to recognize frames +solely on this. + +__`Frame_Size`__ + +The total size of the skippable frame, not including the `Skippable_Magic_Number` or `Frame_Size`. +This is for compatibility with [Zstandard skippable frames]. + +[Zstandard skippable frames]: https://github.com/facebook/zstd/blob/release/doc/zstd_compression_format.md#skippable-frames + +#### `Seek_Table_Footer` +The seek table footer format is as follows: + +|`Number_Of_Frames`|`Seek_Table_Descriptor`|`Seekable_Magic_Number`| +|------------------|-----------------------|-----------------------| +| 4 bytes | 1 byte | 4 bytes | + +__`Seekable_Magic_Number`__ + +Value : 0x8F92EAB1. +This value must be the last bytes present in the compressed file so that decoders +can efficiently find it and determine if there is an actual seek table present. + +__`Number_Of_Frames`__ + +The number of stored frames in the data. + +__`Seek_Table_Descriptor`__ + +A bitfield describing the format of the seek table. + +| Bit number | Field name | +| ---------- | ---------- | +| 7 | `Checksum_Flag` | +| 6-2 | `Reserved_Bits` | +| 1-0 | `Unused_Bits` | + +While only `Checksum_Flag` currently exists, there are 7 other bits in this field that can be used for future changes to the format, +for example the addition of inline dictionaries. + +__`Checksum_Flag`__ + +If the checksum flag is set, each of the seek table entries contains a 4 byte checksum of the uncompressed data contained in its frame. + +`Reserved_Bits` are not currently used but may be used in the future for breaking changes, so a compliant decoder should ensure they are set to 0. `Unused_Bits` may be used in the future for non-breaking changes, so a compliant decoder should not interpret these bits. + +#### __`Seek_Table_Entries`__ + +`Seek_Table_Entries` consists of `Number_Of_Frames` (one for each frame in the data, not including the seek table frame) entries of the following form, in sequence: + +|`Compressed_Size`|`Decompressed_Size`|`[Checksum]`| +|-----------------|-------------------|------------| +| 4 bytes | 4 bytes | 4 bytes | + +__`Compressed_Size`__ + +The compressed size of the frame. +The cumulative sum of the `Compressed_Size` fields of frames `0` to `i` gives the offset in the compressed file of frame `i+1`. + +__`Decompressed_Size`__ + +The size of the decompressed data contained in the frame. For skippable or otherwise empty frames, this value is 0. + +__`Checksum`__ + +Only present if `Checksum_Flag` is set in the `Seek_Table_Descriptor`. Value : the least significant 32 bits of the XXH64 digest of the uncompressed data, stored in little-endian format. + +## Version Changes +- 0.1.0: initial version diff --git a/3rdparty/zstd/contrib/seekable_format/zstdseek_compress.c b/3rdparty/zstd/contrib/seekable_format/zstdseek_compress.c new file mode 100644 index 00000000000..4997807d251 --- /dev/null +++ b/3rdparty/zstd/contrib/seekable_format/zstdseek_compress.c @@ -0,0 +1,365 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + */ + +#include <stdlib.h> /* malloc, free */ +#include <limits.h> /* UINT_MAX */ +#include <assert.h> + +#define XXH_STATIC_LINKING_ONLY +#include "xxhash.h" + +#define ZSTD_STATIC_LINKING_ONLY +#include "zstd.h" +#include "zstd_errors.h" +#include "mem.h" + +#include "zstd_seekable.h" + +#define CHECK_Z(f) { size_t const ret = (f); if (ret != 0) return ret; } + +#undef ERROR +#define ERROR(name) ((size_t)-ZSTD_error_##name) + +#undef MIN +#undef MAX +#define MIN(a, b) ((a) < (b) ? (a) : (b)) +#define MAX(a, b) ((a) > (b) ? (a) : (b)) + +typedef struct { + U32 cSize; + U32 dSize; + U32 checksum; +} framelogEntry_t; + +struct ZSTD_frameLog_s { + framelogEntry_t* entries; + U32 size; + U32 capacity; + + int checksumFlag; + + /* for use when streaming out the seek table */ + U32 seekTablePos; + U32 seekTableIndex; +} framelog_t; + +struct ZSTD_seekable_CStream_s { + ZSTD_CStream* cstream; + ZSTD_frameLog framelog; + + U32 frameCSize; + U32 frameDSize; + + XXH64_state_t xxhState; + + U32 maxFrameSize; + + int writingSeekTable; +}; + +static size_t ZSTD_seekable_frameLog_allocVec(ZSTD_frameLog* fl) +{ + /* allocate some initial space */ + size_t const FRAMELOG_STARTING_CAPACITY = 16; + fl->entries = (framelogEntry_t*)malloc( + sizeof(framelogEntry_t) * FRAMELOG_STARTING_CAPACITY); + if (fl->entries == NULL) return ERROR(memory_allocation); + fl->capacity = (U32)FRAMELOG_STARTING_CAPACITY; + return 0; +} + +static size_t ZSTD_seekable_frameLog_freeVec(ZSTD_frameLog* fl) +{ + if (fl != NULL) free(fl->entries); + return 0; +} + +ZSTD_frameLog* ZSTD_seekable_createFrameLog(int checksumFlag) +{ + ZSTD_frameLog* const fl = (ZSTD_frameLog*)malloc(sizeof(ZSTD_frameLog)); + if (fl == NULL) return NULL; + + if (ZSTD_isError(ZSTD_seekable_frameLog_allocVec(fl))) { + free(fl); + return NULL; + } + + fl->checksumFlag = checksumFlag; + fl->seekTablePos = 0; + fl->seekTableIndex = 0; + fl->size = 0; + + return fl; +} + +size_t ZSTD_seekable_freeFrameLog(ZSTD_frameLog* fl) +{ + ZSTD_seekable_frameLog_freeVec(fl); + free(fl); + return 0; +} + +ZSTD_seekable_CStream* ZSTD_seekable_createCStream(void) +{ + ZSTD_seekable_CStream* const zcs = (ZSTD_seekable_CStream*)malloc(sizeof(ZSTD_seekable_CStream)); + if (zcs == NULL) return NULL; + + memset(zcs, 0, sizeof(*zcs)); + + zcs->cstream = ZSTD_createCStream(); + if (zcs->cstream == NULL) goto failed1; + + if (ZSTD_isError(ZSTD_seekable_frameLog_allocVec(&zcs->framelog))) goto failed2; + + return zcs; + +failed2: + ZSTD_freeCStream(zcs->cstream); +failed1: + free(zcs); + return NULL; +} + +size_t ZSTD_seekable_freeCStream(ZSTD_seekable_CStream* zcs) +{ + if (zcs == NULL) return 0; /* support free on null */ + ZSTD_freeCStream(zcs->cstream); + ZSTD_seekable_frameLog_freeVec(&zcs->framelog); + free(zcs); + return 0; +} + +size_t ZSTD_seekable_initCStream(ZSTD_seekable_CStream* zcs, + int compressionLevel, + int checksumFlag, + unsigned maxFrameSize) +{ + zcs->framelog.size = 0; + zcs->frameCSize = 0; + zcs->frameDSize = 0; + + /* make sure maxFrameSize has a reasonable value */ + if (maxFrameSize > ZSTD_SEEKABLE_MAX_FRAME_DECOMPRESSED_SIZE) { + return ERROR(frameParameter_unsupported); + } + + zcs->maxFrameSize = maxFrameSize ? + maxFrameSize : ZSTD_SEEKABLE_MAX_FRAME_DECOMPRESSED_SIZE; + + zcs->framelog.checksumFlag = checksumFlag; + if (zcs->framelog.checksumFlag) { + XXH64_reset(&zcs->xxhState, 0); + } + + zcs->framelog.seekTablePos = 0; + zcs->framelog.seekTableIndex = 0; + zcs->writingSeekTable = 0; + + return ZSTD_initCStream(zcs->cstream, compressionLevel); +} + +size_t ZSTD_seekable_logFrame(ZSTD_frameLog* fl, + unsigned compressedSize, + unsigned decompressedSize, + unsigned checksum) +{ + if (fl->size == ZSTD_SEEKABLE_MAXFRAMES) + return ERROR(frameIndex_tooLarge); + + /* grow the buffer if required */ + if (fl->size == fl->capacity) { + /* exponential size increase for constant amortized runtime */ + size_t const newCapacity = fl->capacity * 2; + framelogEntry_t* const newEntries = (framelogEntry_t*)realloc(fl->entries, + sizeof(framelogEntry_t) * newCapacity); + + if (newEntries == NULL) return ERROR(memory_allocation); + + fl->entries = newEntries; + assert(newCapacity <= UINT_MAX); + fl->capacity = (U32)newCapacity; + } + + fl->entries[fl->size] = (framelogEntry_t){ + compressedSize, decompressedSize, checksum + }; + fl->size++; + + return 0; +} + +size_t ZSTD_seekable_endFrame(ZSTD_seekable_CStream* zcs, ZSTD_outBuffer* output) +{ + size_t const prevOutPos = output->pos; + /* end the frame */ + size_t ret = ZSTD_endStream(zcs->cstream, output); + + zcs->frameCSize += (U32)(output->pos - prevOutPos); + + /* need to flush before doing the rest */ + if (ret) return ret; + + /* frame done */ + + /* store the frame data for later */ + ret = ZSTD_seekable_logFrame( + &zcs->framelog, zcs->frameCSize, zcs->frameDSize, + zcs->framelog.checksumFlag + ? XXH64_digest(&zcs->xxhState) & 0xFFFFFFFFU + : 0); + if (ret) return ret; + + /* reset for the next frame */ + zcs->frameCSize = 0; + zcs->frameDSize = 0; + + ZSTD_CCtx_reset(zcs->cstream, ZSTD_reset_session_only); + if (zcs->framelog.checksumFlag) XXH64_reset(&zcs->xxhState, 0); + + return 0; +} + +size_t ZSTD_seekable_compressStream(ZSTD_seekable_CStream* zcs, ZSTD_outBuffer* output, ZSTD_inBuffer* input) +{ + const BYTE* const inBase = (const BYTE*) input->src + input->pos; + size_t inLen = input->size - input->pos; + + assert(zcs->maxFrameSize < INT_MAX); + ZSTD_CCtx_setParameter(zcs->cstream, ZSTD_c_srcSizeHint, (int)zcs->maxFrameSize); + inLen = MIN(inLen, (size_t)(zcs->maxFrameSize - zcs->frameDSize)); + + /* if we haven't finished flushing the last frame, don't start writing a new one */ + if (inLen > 0) { + ZSTD_inBuffer inTmp = { inBase, inLen, 0 }; + size_t const prevOutPos = output->pos; + + size_t const ret = ZSTD_compressStream(zcs->cstream, output, &inTmp); + + if (zcs->framelog.checksumFlag) { + XXH64_update(&zcs->xxhState, inBase, inTmp.pos); + } + + zcs->frameCSize += (U32)(output->pos - prevOutPos); + zcs->frameDSize += (U32)inTmp.pos; + + input->pos += inTmp.pos; + + if (ZSTD_isError(ret)) return ret; + } + + if (zcs->maxFrameSize == zcs->frameDSize) { + /* log the frame and start over */ + size_t const ret = ZSTD_seekable_endFrame(zcs, output); + if (ZSTD_isError(ret)) return ret; + + /* get the client ready for the next frame */ + return (size_t)zcs->maxFrameSize; + } + + return (size_t)(zcs->maxFrameSize - zcs->frameDSize); +} + +static inline size_t ZSTD_seekable_seekTableSize(const ZSTD_frameLog* fl) +{ + size_t const sizePerFrame = 8 + (fl->checksumFlag?4:0); + size_t const seekTableLen = ZSTD_SKIPPABLEHEADERSIZE + + sizePerFrame * fl->size + + ZSTD_seekTableFooterSize; + + return seekTableLen; +} + +static inline size_t ZSTD_stwrite32(ZSTD_frameLog* fl, + ZSTD_outBuffer* output, U32 const value, + U32 const offset) +{ + if (fl->seekTablePos < offset + 4) { + BYTE tmp[4]; /* so that we can work with buffers too small to write a whole word to */ + size_t const lenWrite = + MIN(output->size - output->pos, offset + 4 - fl->seekTablePos); + MEM_writeLE32(tmp, value); + memcpy((BYTE*)output->dst + output->pos, + tmp + (fl->seekTablePos - offset), lenWrite); + output->pos += lenWrite; + fl->seekTablePos += (U32)lenWrite; + + if (lenWrite < 4) return ZSTD_seekable_seekTableSize(fl) - fl->seekTablePos; + } + return 0; +} + +size_t ZSTD_seekable_writeSeekTable(ZSTD_frameLog* fl, ZSTD_outBuffer* output) +{ + /* seekTableIndex: the current index in the table and + * seekTableSize: the amount of the table written so far + * + * This function is written this way so that if it has to return early + * because of a small buffer, it can keep going where it left off. + */ + + size_t const sizePerFrame = 8 + (fl->checksumFlag?4:0); + size_t const seekTableLen = ZSTD_seekable_seekTableSize(fl); + + CHECK_Z(ZSTD_stwrite32(fl, output, ZSTD_MAGIC_SKIPPABLE_START | 0xE, 0)); + assert(seekTableLen <= (size_t)UINT_MAX); + CHECK_Z(ZSTD_stwrite32(fl, output, (U32)seekTableLen - ZSTD_SKIPPABLEHEADERSIZE, 4)); + + while (fl->seekTableIndex < fl->size) { + unsigned long long const start = ZSTD_SKIPPABLEHEADERSIZE + sizePerFrame * fl->seekTableIndex; + assert(start + 8 <= UINT_MAX); + CHECK_Z(ZSTD_stwrite32(fl, output, + fl->entries[fl->seekTableIndex].cSize, + (U32)start + 0)); + + CHECK_Z(ZSTD_stwrite32(fl, output, + fl->entries[fl->seekTableIndex].dSize, + (U32)start + 4)); + + if (fl->checksumFlag) { + CHECK_Z(ZSTD_stwrite32( + fl, output, fl->entries[fl->seekTableIndex].checksum, + (U32)start + 8)); + } + + fl->seekTableIndex++; + } + + assert(seekTableLen <= UINT_MAX); + CHECK_Z(ZSTD_stwrite32(fl, output, fl->size, + (U32)seekTableLen - ZSTD_seekTableFooterSize)); + + if (output->size - output->pos < 1) return seekTableLen - fl->seekTablePos; + if (fl->seekTablePos < seekTableLen - 4) { + BYTE const sfd = (BYTE)((fl->checksumFlag) << 7); + + ((BYTE*)output->dst)[output->pos] = sfd; + output->pos++; + fl->seekTablePos++; + } + + CHECK_Z(ZSTD_stwrite32(fl, output, ZSTD_SEEKABLE_MAGICNUMBER, + (U32)seekTableLen - 4)); + + if (fl->seekTablePos != seekTableLen) return ERROR(GENERIC); + return 0; +} + +size_t ZSTD_seekable_endStream(ZSTD_seekable_CStream* zcs, ZSTD_outBuffer* output) +{ + if (!zcs->writingSeekTable) { + const size_t endFrame = ZSTD_seekable_endFrame(zcs, output); + if (ZSTD_isError(endFrame)) return endFrame; + /* return an accurate size hint */ + if (endFrame) return endFrame + ZSTD_seekable_seekTableSize(&zcs->framelog); + } + + zcs->writingSeekTable = 1; + + return ZSTD_seekable_writeSeekTable(&zcs->framelog, output); +} diff --git a/3rdparty/zstd/contrib/seekable_format/zstdseek_decompress.c b/3rdparty/zstd/contrib/seekable_format/zstdseek_decompress.c new file mode 100644 index 00000000000..4fefc5f2eb6 --- /dev/null +++ b/3rdparty/zstd/contrib/seekable_format/zstdseek_decompress.c @@ -0,0 +1,595 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * All rights reserved. + * + * This source code is licensed under both the BSD-style license (found in the + * LICENSE file in the root directory of this source tree) and the GPLv2 (found + * in the COPYING file in the root directory of this source tree). + * You may select, at your option, one of the above-listed licenses. + */ + +/* ********************************************************* +* Turn on Large Files support (>4GB) for 32-bit Linux/Unix +***********************************************************/ +#if !defined(__64BIT__) || defined(__MINGW32__) /* No point defining Large file for 64 bit but MinGW-w64 requires it */ +# if !defined(_FILE_OFFSET_BITS) +# define _FILE_OFFSET_BITS 64 /* turn off_t into a 64-bit type for ftello, fseeko */ +# endif +# if !defined(_LARGEFILE_SOURCE) /* obsolete macro, replaced with _FILE_OFFSET_BITS */ +# define _LARGEFILE_SOURCE 1 /* Large File Support extension (LFS) - fseeko, ftello */ +# endif +# if defined(_AIX) || defined(__hpux) +# define _LARGE_FILES /* Large file support on 32-bits AIX and HP-UX */ +# endif +#endif + +/* ************************************************************ +* Detect POSIX version +* PLATFORM_POSIX_VERSION = 0 for non-Unix e.g. Windows +* PLATFORM_POSIX_VERSION = 1 for Unix-like but non-POSIX +* PLATFORM_POSIX_VERSION > 1 is equal to found _POSIX_VERSION +* Value of PLATFORM_POSIX_VERSION can be forced on command line +***************************************************************/ +#ifndef PLATFORM_POSIX_VERSION + +# if (defined(__APPLE__) && defined(__MACH__)) || defined(__SVR4) || defined(_AIX) || defined(__hpux) /* POSIX.1-2001 (SUSv3) conformant */ \ + || defined(__DragonFly__) || defined(__FreeBSD__) || defined(__NetBSD__) || defined(__OpenBSD__) /* BSD distros */ + /* exception rule : force posix version to 200112L, + * note: it's better to use unistd.h's _POSIX_VERSION whenever possible */ +# define PLATFORM_POSIX_VERSION 200112L + +/* try to determine posix version through official unistd.h's _POSIX_VERSION (https://pubs.opengroup.org/onlinepubs/7908799/xsh/unistd.h.html). + * note : there is no simple way to know in advance if <unistd.h> is present or not on target system, + * Posix specification mandates its presence and its content, but target system must respect this spec. + * It's necessary to _not_ #include <unistd.h> whenever target OS is not unix-like + * otherwise it will block preprocessing stage. + * The following list of build macros tries to "guess" if target OS is likely unix-like, and therefore can #include <unistd.h> + */ +# elif !defined(_WIN32) \ + && ( defined(__unix__) || defined(__unix) \ + || defined(__midipix__) || defined(__VMS) || defined(__HAIKU__) ) + +# if defined(__linux__) || defined(__linux) || defined(__CYGWIN__) +# ifndef _POSIX_C_SOURCE +# define _POSIX_C_SOURCE 200809L /* feature test macro : https://www.gnu.org/software/libc/manual/html_node/Feature-Test-Macros.html */ +# endif +# endif +# include <unistd.h> /* declares _POSIX_VERSION */ +# if defined(_POSIX_VERSION) /* POSIX compliant */ +# define PLATFORM_POSIX_VERSION _POSIX_VERSION +# else +# define PLATFORM_POSIX_VERSION 1 +# endif + +# ifdef __UCLIBC__ +# ifndef __USE_MISC +# define __USE_MISC /* enable st_mtim on uclibc */ +# endif +# endif + +# else /* non-unix target platform (like Windows) */ +# define PLATFORM_POSIX_VERSION 0 +# endif + +#endif /* PLATFORM_POSIX_VERSION */ + + +/* ************************************************************ +* Avoid fseek()'s 2GiB barrier with MSVC, macOS, *BSD, MinGW +***************************************************************/ +#if defined(_MSC_VER) && _MSC_VER >= 1400 +# define LONG_SEEK _fseeki64 +#elif !defined(__64BIT__) && (PLATFORM_POSIX_VERSION >= 200112L) /* No point defining Large file for 64 bit */ +# define LONG_SEEK fseeko +#elif defined(__MINGW32__) && !defined(__STRICT_ANSI__) && !defined(__NO_MINGW_LFS) && defined(__MSVCRT__) +# define LONG_SEEK fseeko64 +#elif defined(_WIN32) && !defined(__DJGPP__) +# include <windows.h> + static int LONG_SEEK(FILE* file, __int64 offset, int origin) { + LARGE_INTEGER off; + DWORD method; + off.QuadPart = offset; + if (origin == SEEK_END) + method = FILE_END; + else if (origin == SEEK_CUR) + method = FILE_CURRENT; + else + method = FILE_BEGIN; + + if (SetFilePointerEx((HANDLE) _get_osfhandle(_fileno(file)), off, NULL, method)) + return 0; + else + return -1; + } +#else +# define LONG_SEEK fseek +#endif + +#include <stdlib.h> /* malloc, free */ +#include <stdio.h> /* FILE* */ +#include <limits.h> /* UNIT_MAX */ +#include <assert.h> + +#define XXH_STATIC_LINKING_ONLY +#include "xxhash.h" + +#define ZSTD_STATIC_LINKING_ONLY +#include "zstd.h" +#include "zstd_errors.h" +#include "mem.h" +#include "zstd_seekable.h" + +#undef ERROR +#define ERROR(name) ((size_t)-ZSTD_error_##name) + +#define CHECK_IO(f) { int const errcod = (f); if (errcod < 0) return ERROR(seekableIO); } + +#undef MIN +#undef MAX +#define MIN(a, b) ((a) < (b) ? (a) : (b)) +#define MAX(a, b) ((a) > (b) ? (a) : (b)) + +#define ZSTD_SEEKABLE_NO_OUTPUT_PROGRESS_MAX 16 + +/* Special-case callbacks for FILE* and in-memory modes, so that we can treat + * them the same way as the advanced API */ +static int ZSTD_seekable_read_FILE(void* opaque, void* buffer, size_t n) +{ + size_t const result = fread(buffer, 1, n, (FILE*)opaque); + if (result != n) { + return -1; + } + return 0; +} + +static int ZSTD_seekable_seek_FILE(void* opaque, long long offset, int origin) +{ + int const ret = LONG_SEEK((FILE*)opaque, offset, origin); + if (ret) return ret; + return fflush((FILE*)opaque); +} + +typedef struct { + const void *ptr; + size_t size; + size_t pos; +} buffWrapper_t; + +static int ZSTD_seekable_read_buff(void* opaque, void* buffer, size_t n) +{ + buffWrapper_t* const buff = (buffWrapper_t*)opaque; + assert(buff != NULL); + if (buff->pos + n > buff->size) return -1; + memcpy(buffer, (const BYTE*)buff->ptr + buff->pos, n); + buff->pos += n; + return 0; +} + +static int ZSTD_seekable_seek_buff(void* opaque, long long offset, int origin) +{ + buffWrapper_t* const buff = (buffWrapper_t*) opaque; + unsigned long long newOffset; + assert(buff != NULL); + switch (origin) { + case SEEK_SET: + assert(offset >= 0); + newOffset = (unsigned long long)offset; + break; + case SEEK_CUR: + newOffset = (unsigned long long)((long long)buff->pos + offset); + break; + case SEEK_END: + newOffset = (unsigned long long)((long long)buff->size + offset); + break; + default: + assert(0); /* not possible */ + } + if (newOffset > buff->size) { + return -1; + } + buff->pos = newOffset; + return 0; +} + +typedef struct { + U64 cOffset; + U64 dOffset; + U32 checksum; +} seekEntry_t; + +struct ZSTD_seekTable_s { + seekEntry_t* entries; + size_t tableLen; + + int checksumFlag; +}; + +#define SEEKABLE_BUFF_SIZE ZSTD_BLOCKSIZE_MAX + +struct ZSTD_seekable_s { + ZSTD_DStream* dstream; + ZSTD_seekTable seekTable; + ZSTD_seekable_customFile src; + + U64 decompressedOffset; + U32 curFrame; + + BYTE inBuff[SEEKABLE_BUFF_SIZE]; /* need to do our own input buffering */ + BYTE outBuff[SEEKABLE_BUFF_SIZE]; /* so we can efficiently decompress the + starts of chunks before we get to the + desired section */ + ZSTD_inBuffer in; /* maintain continuity across ZSTD_seekable_decompress operations */ + buffWrapper_t buffWrapper; /* for `src.opaque` in in-memory mode */ + + XXH64_state_t xxhState; +}; + +ZSTD_seekable* ZSTD_seekable_create(void) +{ + ZSTD_seekable* const zs = (ZSTD_seekable*)malloc(sizeof(ZSTD_seekable)); + if (zs == NULL) return NULL; + + /* also initializes stage to zsds_init */ + memset(zs, 0, sizeof(*zs)); + + zs->dstream = ZSTD_createDStream(); + if (zs->dstream == NULL) { + free(zs); + return NULL; + } + + return zs; +} + +size_t ZSTD_seekable_free(ZSTD_seekable* zs) +{ + if (zs == NULL) return 0; /* support free on null */ + ZSTD_freeDStream(zs->dstream); + free(zs->seekTable.entries); + free(zs); + return 0; +} + +ZSTD_seekTable* ZSTD_seekTable_create_fromSeekable(const ZSTD_seekable* zs) +{ + ZSTD_seekTable* const st = (ZSTD_seekTable*)malloc(sizeof(ZSTD_seekTable)); + if (st==NULL) return NULL; + + st->checksumFlag = zs->seekTable.checksumFlag; + st->tableLen = zs->seekTable.tableLen; + + /* Allocate an extra entry at the end to match logic of initial allocation */ + size_t const entriesSize = sizeof(seekEntry_t) * (zs->seekTable.tableLen + 1); + seekEntry_t* const entries = (seekEntry_t*)malloc(entriesSize); + if (entries==NULL) { + free(st); + return NULL; + } + + memcpy(entries, zs->seekTable.entries, entriesSize); + st->entries = entries; + return st; +} + +size_t ZSTD_seekTable_free(ZSTD_seekTable* st) +{ + if (st == NULL) return 0; /* support free on null */ + free(st->entries); + free(st); + return 0; +} + +/** ZSTD_seekable_offsetToFrameIndex() : + * Performs a binary search to find the last frame with a decompressed offset + * <= pos + * @return : the frame's index */ +unsigned ZSTD_seekable_offsetToFrameIndex(const ZSTD_seekable* zs, unsigned long long pos) +{ + return ZSTD_seekTable_offsetToFrameIndex(&zs->seekTable, pos); +} + +unsigned ZSTD_seekTable_offsetToFrameIndex(const ZSTD_seekTable* st, unsigned long long pos) +{ + U32 lo = 0; + U32 hi = (U32)st->tableLen; + assert(st->tableLen <= UINT_MAX); + + if (pos >= st->entries[st->tableLen].dOffset) { + return (unsigned)st->tableLen; + } + + while (lo + 1 < hi) { + U32 const mid = lo + ((hi - lo) >> 1); + if (st->entries[mid].dOffset <= pos) { + lo = mid; + } else { + hi = mid; + } + } + return lo; +} + +unsigned ZSTD_seekable_getNumFrames(const ZSTD_seekable* zs) +{ + return ZSTD_seekTable_getNumFrames(&zs->seekTable); +} + +unsigned ZSTD_seekTable_getNumFrames(const ZSTD_seekTable* st) +{ + assert(st->tableLen <= UINT_MAX); + return (unsigned)st->tableLen; +} + +unsigned long long ZSTD_seekable_getFrameCompressedOffset(const ZSTD_seekable* zs, unsigned frameIndex) +{ + return ZSTD_seekTable_getFrameCompressedOffset(&zs->seekTable, frameIndex); +} + +unsigned long long ZSTD_seekTable_getFrameCompressedOffset(const ZSTD_seekTable* st, unsigned frameIndex) +{ + if (frameIndex >= st->tableLen) return ZSTD_SEEKABLE_FRAMEINDEX_TOOLARGE; + return st->entries[frameIndex].cOffset; +} + +unsigned long long ZSTD_seekable_getFrameDecompressedOffset(const ZSTD_seekable* zs, unsigned frameIndex) +{ + return ZSTD_seekTable_getFrameDecompressedOffset(&zs->seekTable, frameIndex); +} + +unsigned long long ZSTD_seekTable_getFrameDecompressedOffset(const ZSTD_seekTable* st, unsigned frameIndex) +{ + if (frameIndex >= st->tableLen) return ZSTD_SEEKABLE_FRAMEINDEX_TOOLARGE; + return st->entries[frameIndex].dOffset; +} + +size_t ZSTD_seekable_getFrameCompressedSize(const ZSTD_seekable* zs, unsigned frameIndex) +{ + return ZSTD_seekTable_getFrameCompressedSize(&zs->seekTable, frameIndex); +} + +size_t ZSTD_seekTable_getFrameCompressedSize(const ZSTD_seekTable* st, unsigned frameIndex) +{ + if (frameIndex >= st->tableLen) return ERROR(frameIndex_tooLarge); + return st->entries[frameIndex + 1].cOffset - + st->entries[frameIndex].cOffset; +} + +size_t ZSTD_seekable_getFrameDecompressedSize(const ZSTD_seekable* zs, unsigned frameIndex) +{ + return ZSTD_seekTable_getFrameDecompressedSize(&zs->seekTable, frameIndex); +} + +size_t ZSTD_seekTable_getFrameDecompressedSize(const ZSTD_seekTable* st, unsigned frameIndex) +{ + if (frameIndex > st->tableLen) return ERROR(frameIndex_tooLarge); + return st->entries[frameIndex + 1].dOffset - + st->entries[frameIndex].dOffset; +} + +static size_t ZSTD_seekable_loadSeekTable(ZSTD_seekable* zs) +{ + int checksumFlag; + ZSTD_seekable_customFile src = zs->src; + /* read the footer, fixed size */ + CHECK_IO(src.seek(src.opaque, -(int)ZSTD_seekTableFooterSize, SEEK_END)); + CHECK_IO(src.read(src.opaque, zs->inBuff, ZSTD_seekTableFooterSize)); + + if (MEM_readLE32(zs->inBuff + 5) != ZSTD_SEEKABLE_MAGICNUMBER) { + return ERROR(prefix_unknown); + } + + { BYTE const sfd = zs->inBuff[4]; + checksumFlag = sfd >> 7; + + /* check reserved bits */ + if ((sfd >> 2) & 0x1f) { + return ERROR(corruption_detected); + } } + + { U32 const numFrames = MEM_readLE32(zs->inBuff); + U32 const sizePerEntry = 8 + (checksumFlag?4:0); + U32 const tableSize = sizePerEntry * numFrames; + U32 const frameSize = tableSize + ZSTD_seekTableFooterSize + ZSTD_SKIPPABLEHEADERSIZE; + + U32 remaining = frameSize - ZSTD_seekTableFooterSize; /* don't need to re-read footer */ + { U32 const toRead = MIN(remaining, SEEKABLE_BUFF_SIZE); + CHECK_IO(src.seek(src.opaque, -(S64)frameSize, SEEK_END)); + CHECK_IO(src.read(src.opaque, zs->inBuff, toRead)); + remaining -= toRead; + } + + if (MEM_readLE32(zs->inBuff) != (ZSTD_MAGIC_SKIPPABLE_START | 0xE)) { + return ERROR(prefix_unknown); + } + if (MEM_readLE32(zs->inBuff+4) + ZSTD_SKIPPABLEHEADERSIZE != frameSize) { + return ERROR(prefix_unknown); + } + + { /* Allocate an extra entry at the end so that we can do size + * computations on the last element without special case */ + seekEntry_t* const entries = (seekEntry_t*)malloc(sizeof(seekEntry_t) * (numFrames + 1)); + + U32 idx = 0; + U32 pos = 8; + + U64 cOffset = 0; + U64 dOffset = 0; + + if (entries == NULL) return ERROR(memory_allocation); + + /* compute cumulative positions */ + for (; idx < numFrames; idx++) { + if (pos + sizePerEntry > SEEKABLE_BUFF_SIZE) { + U32 const offset = SEEKABLE_BUFF_SIZE - pos; + U32 const toRead = MIN(remaining, SEEKABLE_BUFF_SIZE - offset); + memmove(zs->inBuff, zs->inBuff + pos, offset); /* move any data we haven't read yet */ + CHECK_IO(src.read(src.opaque, zs->inBuff+offset, toRead)); + remaining -= toRead; + pos = 0; + } + entries[idx].cOffset = cOffset; + entries[idx].dOffset = dOffset; + + cOffset += MEM_readLE32(zs->inBuff + pos); + pos += 4; + dOffset += MEM_readLE32(zs->inBuff + pos); + pos += 4; + if (checksumFlag) { + entries[idx].checksum = MEM_readLE32(zs->inBuff + pos); + pos += 4; + } + } + entries[numFrames].cOffset = cOffset; + entries[numFrames].dOffset = dOffset; + + zs->seekTable.entries = entries; + zs->seekTable.tableLen = numFrames; + zs->seekTable.checksumFlag = checksumFlag; + return 0; + } + } +} + +size_t ZSTD_seekable_initBuff(ZSTD_seekable* zs, const void* src, size_t srcSize) +{ + zs->buffWrapper = (buffWrapper_t){src, srcSize, 0}; + { ZSTD_seekable_customFile srcFile = {&zs->buffWrapper, + &ZSTD_seekable_read_buff, + &ZSTD_seekable_seek_buff}; + return ZSTD_seekable_initAdvanced(zs, srcFile); } +} + +size_t ZSTD_seekable_initFile(ZSTD_seekable* zs, FILE* src) +{ + ZSTD_seekable_customFile srcFile = {src, &ZSTD_seekable_read_FILE, + &ZSTD_seekable_seek_FILE}; + return ZSTD_seekable_initAdvanced(zs, srcFile); +} + +size_t ZSTD_seekable_initAdvanced(ZSTD_seekable* zs, ZSTD_seekable_customFile src) +{ + zs->src = src; + + { const size_t seekTableInit = ZSTD_seekable_loadSeekTable(zs); + if (ZSTD_isError(seekTableInit)) return seekTableInit; } + + zs->decompressedOffset = (U64)-1; + zs->curFrame = (U32)-1; + + { const size_t dstreamInit = ZSTD_initDStream(zs->dstream); + if (ZSTD_isError(dstreamInit)) return dstreamInit; } + return 0; +} + +size_t ZSTD_seekable_decompress(ZSTD_seekable* zs, void* dst, size_t len, unsigned long long offset) +{ + unsigned long long const eos = zs->seekTable.entries[zs->seekTable.tableLen].dOffset; + if (offset + len > eos) { + len = eos - offset; + } + + U32 targetFrame = ZSTD_seekable_offsetToFrameIndex(zs, offset); + U32 noOutputProgressCount = 0; + size_t srcBytesRead = 0; + do { + /* check if we can continue from a previous decompress job */ + if (targetFrame != zs->curFrame || offset < zs->decompressedOffset) { + zs->decompressedOffset = zs->seekTable.entries[targetFrame].dOffset; + zs->curFrame = targetFrame; + + assert(zs->seekTable.entries[targetFrame].cOffset < LLONG_MAX); + CHECK_IO(zs->src.seek(zs->src.opaque, + (long long)zs->seekTable.entries[targetFrame].cOffset, + SEEK_SET)); + zs->in = (ZSTD_inBuffer){zs->inBuff, 0, 0}; + XXH64_reset(&zs->xxhState, 0); + ZSTD_DCtx_reset(zs->dstream, ZSTD_reset_session_only); + if (zs->buffWrapper.size && srcBytesRead > zs->buffWrapper.size) { + return ERROR(seekableIO); + } + } + + while (zs->decompressedOffset < offset + len) { + size_t toRead; + ZSTD_outBuffer outTmp; + size_t prevOutPos; + size_t prevInPos; + size_t forwardProgress; + if (zs->decompressedOffset < offset) { + /* dummy decompressions until we get to the target offset */ + outTmp = (ZSTD_outBuffer){zs->outBuff, (size_t) (MIN(SEEKABLE_BUFF_SIZE, offset - zs->decompressedOffset)), 0}; + } else { + outTmp = (ZSTD_outBuffer){dst, len, (size_t) (zs->decompressedOffset - offset)}; + } + + prevOutPos = outTmp.pos; + prevInPos = zs->in.pos; + toRead = ZSTD_decompressStream(zs->dstream, &outTmp, &zs->in); + if (ZSTD_isError(toRead)) { + return toRead; + } + + if (zs->seekTable.checksumFlag) { + XXH64_update(&zs->xxhState, (BYTE*)outTmp.dst + prevOutPos, + outTmp.pos - prevOutPos); + } + forwardProgress = outTmp.pos - prevOutPos; + if (forwardProgress == 0) { + if (noOutputProgressCount++ > ZSTD_SEEKABLE_NO_OUTPUT_PROGRESS_MAX) { + return ERROR(seekableIO); + } + } else { + noOutputProgressCount = 0; + } + zs->decompressedOffset += forwardProgress; + srcBytesRead += zs->in.pos - prevInPos; + + if (toRead == 0) { + /* frame complete */ + + /* verify checksum */ + if (zs->seekTable.checksumFlag && + (XXH64_digest(&zs->xxhState) & 0xFFFFFFFFU) != + zs->seekTable.entries[targetFrame].checksum) { + return ERROR(corruption_detected); + } + + if (zs->decompressedOffset < offset + len) { + /* go back to the start and force a reset of the stream */ + targetFrame = ZSTD_seekable_offsetToFrameIndex(zs, zs->decompressedOffset); + /* in this case it will fail later with corruption_detected, since last block does not have checksum */ + assert(targetFrame != zs->seekTable.tableLen); + } + break; + } + + /* read in more data if we're done with this buffer */ + if (zs->in.pos == zs->in.size) { + toRead = MIN(toRead, SEEKABLE_BUFF_SIZE); + CHECK_IO(zs->src.read(zs->src.opaque, zs->inBuff, toRead)); + zs->in.size = toRead; + zs->in.pos = 0; + } + } /* while (zs->decompressedOffset < offset + len) */ + } while (zs->decompressedOffset != offset + len); + + return len; +} + +size_t ZSTD_seekable_decompressFrame(ZSTD_seekable* zs, void* dst, size_t dstSize, unsigned frameIndex) +{ + if (frameIndex >= zs->seekTable.tableLen) { + return ERROR(frameIndex_tooLarge); + } + + { size_t const decompressedSize = + zs->seekTable.entries[frameIndex + 1].dOffset - + zs->seekTable.entries[frameIndex].dOffset; + if (dstSize < decompressedSize) { + return ERROR(dstSize_tooSmall); + } + return ZSTD_seekable_decompress( + zs, dst, decompressedSize, + zs->seekTable.entries[frameIndex].dOffset); + } +} diff --git a/3rdparty/zstd/contrib/seqBench/Makefile b/3rdparty/zstd/contrib/seqBench/Makefile new file mode 100644 index 00000000000..e7f08a42cc8 --- /dev/null +++ b/3rdparty/zstd/contrib/seqBench/Makefile @@ -0,0 +1,58 @@ +# ################################################################ +# Copyright (c) Meta Platforms, Inc. and affiliates. +# All rights reserved. +# +# This source code is licensed under both the BSD-style license (found in the +# LICENSE file in the root directory of this source tree) and the GPLv2 (found +# in the COPYING file in the root directory of this source tree). +# ################################################################ + +PROGDIR = ../../programs +LIBDIR = ../../lib + +LIBZSTD = $(LIBDIR)/libzstd.a + +CPPFLAGS+= -I$(LIBDIR) -I$(LIBDIR)/common -I$(LIBDIR)/dictBuilder -I$(PROGDIR) + +CFLAGS ?= -O3 -g +CFLAGS += -std=gnu99 +DEBUGFLAGS= -Wall -Wextra -Wcast-qual -Wcast-align -Wshadow \ + -Wstrict-aliasing=1 -Wswitch-enum \ + -Wstrict-prototypes -Wundef -Wpointer-arith \ + -Wvla -Wformat=2 -Winit-self -Wfloat-equal -Wwrite-strings \ + -Wredundant-decls +CFLAGS += $(DEBUGFLAGS) $(MOREFLAGS) + + +default: seqBench + +all : seqBench + +seqBench: util.o timefn.o benchfn.o datagen.o xxhash.o seqBench.c $(LIBZSTD) + $(CC) $(CPPFLAGS) $(CFLAGS) $^ $(LDFLAGS) -o $@ + +.PHONY: $(LIBZSTD) +$(LIBZSTD): + $(MAKE) -C $(LIBDIR) libzstd.a CFLAGS="$(CFLAGS)" + +benchfn.o: $(PROGDIR)/benchfn.c + $(CC) $(CPPFLAGS) $(CFLAGS) $^ -c + +timefn.o: $(PROGDIR)/timefn.c + $(CC) $(CPPFLAGS) $(CFLAGS) $^ -c + +datagen.o: $(PROGDIR)/datagen.c + $(CC) $(CPPFLAGS) $(CFLAGS) $^ -c + +util.o: $(PROGDIR)/util.c + $(CC) $(CPPFLAGS) $(CFLAGS) $^ -c + + +xxhash.o : $(LIBDIR)/common/xxhash.c + $(CC) $(CPPFLAGS) $(CFLAGS) $^ -c + + +clean: + $(RM) *.o + $(MAKE) -C $(LIBDIR) clean > /dev/null + $(RM) seqBench diff --git a/3rdparty/zstd/contrib/seqBench/seqBench.c b/3rdparty/zstd/contrib/seqBench/seqBench.c new file mode 100644 index 00000000000..7efebec7bea --- /dev/null +++ b/3rdparty/zstd/contrib/seqBench/seqBench.c @@ -0,0 +1,53 @@ +#define ZSTD_STATIC_LINKING_ONLY +#include <zstd.h> +#include <stdio.h> +#include <stdlib.h> +#include <assert.h> +#include <string.h> + +int main(int argc, char *argv[]) { + ZSTD_CCtx* zc = ZSTD_createCCtx(); + + if (argc != 2) { + printf("Usage: seqBench <file>\n"); // TODO provide the block delim option here + return 1; + } + + FILE *f = fopen(argv[1], "rb"); + fseek(f, 0, SEEK_END); + long inBufSize = ftell(f); + fseek(f, 0, SEEK_SET); + + char *inBuf = malloc(inBufSize + 1); + fread(inBuf, inBufSize, 1, f); + fclose(f); + + size_t seqsSize = ZSTD_sequenceBound(inBufSize); + ZSTD_Sequence *seqs = (ZSTD_Sequence*)malloc(seqsSize * sizeof(ZSTD_Sequence)); + char *outBuf = malloc(ZSTD_compressBound(inBufSize)); + + ZSTD_generateSequences(zc, seqs, seqsSize, inBuf, inBufSize); + ZSTD_CCtx_setParameter(zc, ZSTD_c_blockDelimiters, ZSTD_sf_explicitBlockDelimiters); + size_t outBufSize = ZSTD_compressSequences(zc, outBuf, inBufSize, seqs, seqsSize, inBuf, inBufSize); + if (ZSTD_isError(outBufSize)) { + printf("ERROR: %lu\n", outBufSize); + return 1; + } + + char *validationBuf = malloc(inBufSize); + ZSTD_decompress(validationBuf, inBufSize, outBuf, outBufSize); + + if (memcmp(inBuf, validationBuf, inBufSize) == 0) { + printf("Compression and decompression were successful!\n"); + } else { + printf("ERROR: input and validation buffers don't match!\n"); + for (int i = 0; i < inBufSize; i++) { + if (inBuf[i] != validationBuf[i]) { + printf("First bad index: %d\n", i); + break; + } + } + } + + return 0; +} diff --git a/3rdparty/zstd/contrib/snap/snapcraft.yaml b/3rdparty/zstd/contrib/snap/snapcraft.yaml new file mode 100644 index 00000000000..0a77946ae0a --- /dev/null +++ b/3rdparty/zstd/contrib/snap/snapcraft.yaml @@ -0,0 +1,28 @@ +name: zstd +version: git +summary: Zstandard - Fast real-time compression algorithm +description: | + Zstandard, or zstd as short version, is a fast lossless compression + algorithm, targeting real-time compression scenarios at zlib-level and better + compression ratios. It's backed by a very fast entropy stage, provided by + Huff0 and FSE library + +grade: devel # must be 'stable' to release into candidate/stable channels +confinement: devmode # use 'strict' once you have the right plugs and slots + +apps: + zstd: + command: usr/local/bin/zstd + plugs: [home, removable-media] + zstdgrep: + command: usr/local/bin/zstdgrep + plugs: [home, removable-media] + zstdless: + command: usr/local/bin/zstdless + plugs: [home, removable-media] + +parts: + zstd: + source: . + plugin: make + build-packages: [g++] |