X-Git-Url: https://git.sesse.net/?p=stockfish;a=blobdiff_plain;f=src%2Fmisc.cpp;h=d9c0679bf4e3704de73bcd99c94ffe54db39351a;hp=6ed0bdd0d9108a5633fde87664559bac1f2c40e3;hb=a24f28be8567c2527b154ef981090368a2bd8f76;hpb=9cb1f9f2655cd378eec9d88fd57a039a04396aae diff --git a/src/misc.cpp b/src/misc.cpp index 6ed0bdd0..d9c0679b 100644 --- a/src/misc.cpp +++ b/src/misc.cpp @@ -47,6 +47,11 @@ typedef bool(*fun3_t)(HANDLE, CONST GROUP_AFFINITY*, PGROUP_AFFINITY); #include #include +#if defined(__linux__) && !defined(__ANDROID__) +#include +#include +#endif + #include "misc.h" #include "thread.h" @@ -56,7 +61,7 @@ namespace { /// Version number. If Version is left empty, then compile date in the format /// DD-MM-YY and show in engine_info. -const string Version = "11"; +const string Version = ""; /// Our fancy logging facility. The trick here is to replace cin.rdbuf() and /// cout.rdbuf() with two Tie objects that tie cin and cout to a file stream. We @@ -156,9 +161,9 @@ const string engine_info(bool to_uci) { const std::string compiler_info() { - #define STRINGIFY2(x) #x - #define STRINGIFY(x) STRINGIFY2(x) - #define VER_STRING(major, minor, patch) STRINGIFY(major) "." STRINGIFY(minor) "." STRINGIFY(patch) + #define stringify2(x) #x + #define stringify(x) stringify2(x) + #define make_version_string(major, minor, patch) stringify(major) "." stringify(minor) "." stringify(patch) /// Predefined macros hell: /// @@ -172,26 +177,26 @@ const std::string compiler_info() { #ifdef __clang__ compiler += "clang++ "; - compiler += VER_STRING(__clang_major__, __clang_minor__, __clang_patchlevel__); + compiler += make_version_string(__clang_major__, __clang_minor__, __clang_patchlevel__); #elif __INTEL_COMPILER compiler += "Intel compiler "; compiler += "(version "; - compiler += STRINGIFY(__INTEL_COMPILER) " update " STRINGIFY(__INTEL_COMPILER_UPDATE); + compiler += stringify(__INTEL_COMPILER) " update " stringify(__INTEL_COMPILER_UPDATE); compiler += ")"; #elif _MSC_VER compiler += "MSVC "; compiler += "(version "; - compiler += STRINGIFY(_MSC_FULL_VER) "." STRINGIFY(_MSC_BUILD); + compiler += stringify(_MSC_FULL_VER) "." stringify(_MSC_BUILD); compiler += ")"; #elif __GNUC__ compiler += "g++ (GNUC) "; - compiler += VER_STRING(__GNUC__, __GNUC_MINOR__, __GNUC_PATCHLEVEL__); + compiler += make_version_string(__GNUC__, __GNUC_MINOR__, __GNUC_PATCHLEVEL__); #else compiler += "Unknown compiler "; compiler += "(unknown version)"; #endif - #if defined(__APPLE__) + #if defined(__APPLE__) compiler += " on Apple"; #elif defined(__CYGWIN__) compiler += " on Cygwin"; @@ -289,6 +294,133 @@ void prefetch(void* addr) { #endif + +/// aligned_ttmem_alloc() will return suitably aligned memory, and if possible use large pages. +/// The returned pointer is the aligned one, while the mem argument is the one that needs +/// to be passed to free. With c++17 some of this functionality could be simplified. + +#if defined(__linux__) && !defined(__ANDROID__) + +void* aligned_ttmem_alloc(size_t allocSize, void*& mem) { + + constexpr size_t alignment = 2 * 1024 * 1024; // assumed 2MB page sizes + size_t size = ((allocSize + alignment - 1) / alignment) * alignment; // multiple of alignment + if (posix_memalign(&mem, alignment, size)) + mem = nullptr; + madvise(mem, allocSize, MADV_HUGEPAGE); + return mem; +} + +#elif defined(_WIN64) + +static void* aligned_ttmem_alloc_large_pages(size_t allocSize) { + + HANDLE hProcessToken { }; + LUID luid { }; + void* mem = nullptr; + + const size_t largePageSize = GetLargePageMinimum(); + if (!largePageSize) + return nullptr; + + // We need SeLockMemoryPrivilege, so try to enable it for the process + if (!OpenProcessToken(GetCurrentProcess(), TOKEN_ADJUST_PRIVILEGES | TOKEN_QUERY, &hProcessToken)) + return nullptr; + + if (LookupPrivilegeValue(NULL, SE_LOCK_MEMORY_NAME, &luid)) + { + TOKEN_PRIVILEGES tp { }; + TOKEN_PRIVILEGES prevTp { }; + DWORD prevTpLen = 0; + + tp.PrivilegeCount = 1; + tp.Privileges[0].Luid = luid; + tp.Privileges[0].Attributes = SE_PRIVILEGE_ENABLED; + + // Try to enable SeLockMemoryPrivilege. Note that even if AdjustTokenPrivileges() succeeds, + // we still need to query GetLastError() to ensure that the privileges were actually obtained. + if (AdjustTokenPrivileges( + hProcessToken, FALSE, &tp, sizeof(TOKEN_PRIVILEGES), &prevTp, &prevTpLen) && + GetLastError() == ERROR_SUCCESS) + { + // Round up size to full pages and allocate + allocSize = (allocSize + largePageSize - 1) & ~size_t(largePageSize - 1); + mem = VirtualAlloc( + NULL, allocSize, MEM_RESERVE | MEM_COMMIT | MEM_LARGE_PAGES, PAGE_READWRITE); + + // Privilege no longer needed, restore previous state + AdjustTokenPrivileges(hProcessToken, FALSE, &prevTp, 0, NULL, NULL); + } + } + + CloseHandle(hProcessToken); + + return mem; +} + +void* aligned_ttmem_alloc(size_t allocSize, void*& mem) { + + static bool firstCall = true; + + // Try to allocate large pages + mem = aligned_ttmem_alloc_large_pages(allocSize); + + // Suppress info strings on the first call. The first call occurs before 'uci' + // is received and in that case this output confuses some GUIs. + if (!firstCall) + { + if (mem) + sync_cout << "info string Hash table allocation: Windows large pages used." << sync_endl; + else + sync_cout << "info string Hash table allocation: Windows large pages not used." << sync_endl; + } + firstCall = false; + + // Fall back to regular, page aligned, allocation if necessary + if (!mem) + mem = VirtualAlloc(NULL, allocSize, MEM_RESERVE | MEM_COMMIT, PAGE_READWRITE); + + return mem; +} + +#else + +void* aligned_ttmem_alloc(size_t allocSize, void*& mem) { + + constexpr size_t alignment = 64; // assumed cache line size + size_t size = allocSize + alignment - 1; // allocate some extra space + mem = malloc(size); + void* ret = reinterpret_cast((uintptr_t(mem) + alignment - 1) & ~uintptr_t(alignment - 1)); + return ret; +} + +#endif + + +/// aligned_ttmem_free() will free the previously allocated ttmem + +#if defined(_WIN64) + +void aligned_ttmem_free(void* mem) { + + if (mem && !VirtualFree(mem, 0, MEM_RELEASE)) + { + DWORD err = GetLastError(); + std::cerr << "Failed to free transposition table. Error code: 0x" << + std::hex << err << std::dec << std::endl; + exit(EXIT_FAILURE); + } +} + +#else + +void aligned_ttmem_free(void *mem) { + free(mem); +} + +#endif + + namespace WinProcGroup { #ifndef _WIN32