diff options
| author | Kito Cheng <kito.cheng@sifive.com> | 2021-07-12 12:17:21 +0300 |
|---|---|---|
| committer | Corinna Vinschen <corinna@vinschen.de> | 2021-07-12 22:07:43 +0300 |
| commit | 91f99d323b39dc0c06c40038791db9861d4b76b9 (patch) | |
| tree | 18bdd072eb491f115d013ae9242ebff2f4e0c5c1 /newlib/libc | |
| parent | 9a231073a5978189902183ee7d1ea7390106a318 (diff) | |
Minimal support for ISO/IEC TS 18661-3.
- GCC will set __FLT_EVAL_METHOD__ to 16 if __fp16 supported, e.g.
cortex-a55/aarch64.
- $ aarch64-unknown-elf-gcc -v 2>&1 |grep version
gcc version 9.2.0 (GCC)
- $ aarch64-unknown-elf-gcc -E -dM -mcpu=cortex-a55 - < /dev/null |grep FLT_EVAL_METHOD
#define __FLT_EVAL_METHOD__ 16
#define __FLT_EVAL_METHOD_TS_18661_3__ 16
#define __FLT_EVAL_METHOD_C99__ 16
- The behavior of __FLT_EVAL_METHOD__ == 16 is same as
__FLT_EVAL_METHOD__ == 0 except for float16_t, but newlib didn't
support float16_t.
ISO/IEC TS 18661-3:
http://www.open-std.org/jtc1/sc22/wg14/www/docs/n2405.pdf
V2 Changes:
- List Howland, Craig D as co-author since he provide the draft of comment
in math.h.
Co-authored-by: "Howland, Craig D" <howland@LGSInnovations.com>
Diffstat (limited to 'newlib/libc')
| -rw-r--r-- | newlib/libc/include/math.h | 12 |
1 files changed, 11 insertions, 1 deletions
diff --git a/newlib/libc/include/math.h b/newlib/libc/include/math.h index ba1a8a17e..0b6494e6a 100644 --- a/newlib/libc/include/math.h +++ b/newlib/libc/include/math.h @@ -146,7 +146,17 @@ extern int isnan (double); #define __TMP_FLT_EVAL_METHOD #endif /* FLT_EVAL_METHOD */ #if defined FLT_EVAL_METHOD - #if FLT_EVAL_METHOD == 0 +/* FLT_EVAL_METHOD == 16 has meaning as defined in ISO/IEC TS 18661-3, + * which provides non-compliant extensions to C and POSIX (by adding + * additional positive values for FLT_EVAL_METHOD). It effectively has + * same meaning as the C99 and C11 definitions for value 0, while also + * serving as a flag that the _Float16 (float16_t) type exists. + * + * FLT_EVAL_METHOD could be any number of bits of supported floating point + * format (e.g. 32, 64, 128), but currently only AArch64 and few other targets + * might define that as 16. */ + #if (FLT_EVAL_METHOD == 0) \ + || (FLT_EVAL_METHOD == 16) typedef float float_t; typedef double double_t; #elif FLT_EVAL_METHOD == 1 |