/* This D file is implicitly imported by all ImportC source files. * It provides definitions for C compiler builtin functions and declarations. * The purpose is to make it unnecessary to hardwire them into the compiler. * As the leading double underscore suggests, this is for internal use only. * * Copyright: Copyright Digital Mars 2022 * License: $(HTTP www.boost.org/LICENSE_1_0.txt, Boost License 1.0). * Authors: Walter Bright * Source: $(DRUNTIMESRC __builtins.d) */ module __builtins; /* gcc relies on internal __builtin_xxxx functions and templates to * accomplish . D does the same thing with templates in core.stdc.stdarg. * Here, we redirect the gcc builtin declarations to the equivalent * ones in core.stdc.stdarg, thereby avoiding having to hardware them * into the D compiler. */ import core.stdc.stdarg; alias va_list = core.stdc.stdarg.va_list; version (Posix) { version (X86_64) alias __va_list_tag = core.stdc.stdarg.__va_list_tag; } alias __builtin_va_start = core.stdc.stdarg.va_start; alias __builtin_va_end = core.stdc.stdarg.va_end; alias __builtin_va_copy = core.stdc.stdarg.va_copy; /* dmd's ImportC rewrites __builtin_va_arg into an instantiation of va_arg */ alias va_arg = core.stdc.stdarg.va_arg; version (CRuntime_Microsoft) { //https://docs.microsoft.com/en-us/cpp/cpp/int8-int16-int32-int64?view=msvc-170 alias __int8 = byte; alias __int16 = short; alias __int32 = int; alias __int64 = long; } /*********** floating point *************/ /* https://gcc.gnu.org/onlinedocs/gcc/Other-Builtins.html */ version (DigitalMars) { double __builtin_inf()() { return double.infinity; } float __builtin_inff()() { return float.infinity; } real __builtin_infl()() { return real.infinity; } alias __builtin_huge_val = __builtin_inf; alias __builtin_huge_valf = __builtin_inff; alias __builtin_huge_vall = __builtin_infl; import core.stdc.math; alias __builtin_fabs = core.stdc.math.fabs; alias __builtin_fabsf = core.stdc.math.fabsf; alias __builtin_fabsl = core.stdc.math.fabsl; ushort __builtin_bswap16()(ushort value) { import core.bitop; return core.bitop.byteswap(value); } uint __builtin_bswap32()(uint value) { import core.bitop; return core.bitop.bswap(value); } ulong __builtin_bswap64()(ulong value) { import core.bitop; return core.bitop.bswap(value); } // Stub these out to no-ops int __builtin_constant_p(T)(T exp) { return 0; } // should be something like __traits(compiles, enum X = expr) long __builtin_expect()(long exp, long c) { return exp; } void* __builtin_assume_aligned()(const void* p, size_t align_, ...) { return cast(void*)p; } // https://releases.llvm.org/13.0.0/tools/clang/docs/LanguageExtensions.html#builtin-assume void __builtin_assume(T)(lazy T arg) { } /* Header on macOS for arm64 references this. * Don't need to implement it, it just needs to compile */ align (16) struct __uint128_t { ulong a, b; } }