Finding Bugs using Path-Sensitive Static Analysis Gábor Horváth Gabor.Horvath@microsoft.com @XazaxHunWelcome to CppCon 2021! Join #visual_studio channel on CppCon Discord https://aka.ms/cppcon/discord the latest announcements Take our survey https://aka.ms/cppconAgenda • Intro to path-sensitive static analysis • Path-sensitive checks in MSVC • A look under the hood • Upcoming features • Lessons Documentation in The Era of Concepts and Ranges – Sy Brand & Christopher Di Bella (Google) • Static Analysis and Program Safety in C+ +: Making it Real – Sunny Chatterjee • In-memory and Persistent

numCols_z>, Layout_z, Accessor_z> z) { static_assert(x.static_extent(0) == dynamic_extent || z.static_extent(0) == dynamic_extent || x.static_extent(0) == z.static_ext ... using size_type = std:: optimized for column-major matricesWhich layout should we use? Our BLAS-like interface specifies one row stride, and elements are assumed to be contiguous in each each row, so the column stride is always should we use? BLAS specifies one row stride, and elements are contiguous in each each row, so the column stride is always 1. Sound like we should use std::layout_stride, right? 24 Wrong! std::layout_stride

elements (memory locality) static dimensions std::vector> matrix; com/griswaldbrooks/spanny2 motivations template struct static_matrix { T& operator()(std::size_t i, std::size_t j) { return data_[i*M + j]; } private: indices to storage locations data = row major<3, 3> = column major<3, 3> = 0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8 0 3 6 1 4 7 2 5 8github.com/griswaldbrooks/spanny2

copy and paste both ways from and interactive SQL terminalAnnotations u Columns u /*:type*/ u The column name is immediately to the left of the comment u Parameters u /*:name:type*/ Type annotation C++ for fields and parameters template struct meta_structs_from_query { static constexpr auto ts = parse_type_specs(); using fields_type = typename meta_struct_from_type_specs< struct type_specs { auto operator<=>(const type_specs &) const = default; type_spec data[N]; static constexpr std::size_t size() { return N; } constexpr auto &operator[](std::size_t i) const { return

using T = typename V::value_type; T x { }; T::static_function(); // ... };  With C++20: [](const vector& vec) { T x { }; T::static_function(); // ... };29 Templated Lambda Expressions ❌73 constinit  Static initialization order fiasco  Variables with static storage and dynamic initializers  bugs due to undefined order of dynamic initializations  ❌  Static variables with constant  Represents information about a specific location in a source code  line, column, file_name, function_name  Construct one using source_location::current()  Example: void LogInfo(string_view

return month_names[month-1]; }Correct Construction byclass Month { int month; public: static bool is_valid(int month) { return month >= 1 && month <= 12; } Month(int month) month) { return month_names[month.value()-1]; }class Month { int month; public: static bool is_valid(int month) { return month >= 1 && month <= 12; } Month(int month) {}", ex.what()); } }struct unchecked{}; class Month { int month; public: static bool is_valid(int month) { return month >= 1 && month <= 12; } Month(int month)

Single-Threaded（static） libc.lib 898,826 Multithreaded（static） libcmt.lib 951,142 Multithreaded DLL msvcrt.lib 5,510,000 Debug Single-Threaded（static） libcd.lib 2,374,542 Debug Multithreaded（static）libcmtd /ML Single-Threaded（static） /MT Multithreaded（static） /MD Multithreaded DLL（dynamic import library） /MLd Debug Single-Threaded（static） /MTd Debug Multithreaded（static） /MDd Debug Multithreaded message, UINT wParam, LONG lParam) { DWORD ThreadID[5]; static DWORD ThreadArg[5] = {HIGHEST_THREAD, // 0x00 ABOVE_AVE_THREAD

aligned_alloc 实现 的。 案例：临时创建的数组 • 临时创建的数组，每次调用 func 都会重 复内存分配一次（进入一次内核态），非 常浪费时间。 解决：手动池化 • 声明为 static 变量，这样第二次进入 func 的时候还是 同一个数组，不需要重复分配内存。 thread_local 表示 如有多个线程，每个线程保留一个 tmp 对象的副本， 防止多线程调用 func Intel Xeon CPU was up to 29 X faster. • By using Morton ordering as an alternative to row-major or column-major data storage, significant speedups can be achieved on the Intel Xeon Phi coprocessor or Intel

to a default value in the recipe. For example, one package can define that its default linkage is static, and this is the linkage that should be used if consumers don’t specify otherwise. Note: You can we can modify the previous example to use dynamic linkage instead of the default one, which was static, by editing the conanfile.txt: [requires] Poco/1.9.0@pocoproject/stable [generators] cmake (continues copying the resulting executables to your bin folder, or for copying assets, images, sounds, test static files, etc. Conan is a generic solution for package management, not only (but focused in) for C/C++

to a default value in the recipe. For example, one package can define that its default linkage is static, and this is the linkage that should be used if consumers don’t specify otherwise. Note: You can we can modify the previous example to use dynamic linkage instead of the default one, which was static, by editing the conanfile.txt: [requires] Poco/1.9.0@pocoproject/stable [generators] cmake (continues copying the resulting executables to your bin folder, or for copying assets, images, sounds, test static files, etc. Conan is a generic solution for package management, not only (but focused in) for C/C++