/* $NetBSD: platform_ldbl128.c,v 1.6 2023/07/05 11:42:14 rillig Exp $ */ # 3 "platform_ldbl128.c" /* * Test features that only apply to platforms that have 128-bit long double. */ /* lint1-extra-flags: -c -h -a -p -b -r -z -X 351 */ /* lint1-only-if: ldbl128 */ /* CONSTCOND */ typedef int bits_per_byte[((unsigned char)-1) == 255 ? 1 : -1]; typedef int bytes_per_long_double[sizeof(long double) == 16 ? 1 : -1]; /* * Platforms with 128-bit 'long double' typically use IEEE 754-2008, which has * 1 bit sign + 15 bit exponent + 112 bit normalized mantissa. This means the * maximum representable value is 1.1111111(bin) * 2^16383, which is about * 1.189e4932. This is in the same range as for 96-bit 'long double', as the * exponent range is the same. */ /* expect+1: warning: floating-point constant out of range [248] */ double larger_than_ldbl = 1e4933; /* expect+1: warning: floating-point constant out of range [248] */ long double larger_than_ldbl_l = 1e4933L; /* expect+1: warning: floating-point constant out of range [248] */ double larger_than_dbl = 1e4932; /* Fits in 'long double' but not in 'double'. */ long double larger_than_dbl_l = 1e4932L;