diff options
author | Szabolcs Nagy <szabolcs.nagy@arm.com> | 2018-05-11 13:20:04 +0100 |
---|---|---|
committer | Szabolcs Nagy <szabolcs.nagy@arm.com> | 2018-05-16 13:52:13 +0100 |
commit | 5eb9d19608e31a2d9fe5a9e2708e4628524c29d9 (patch) | |
tree | 6d255c36d9ff6b4dacfdeba4554d2524d7d47357 /math/expf.c | |
parent | c65db17340782d647c49e17cbba244862dc38402 (diff) | |
download | arm-optimized-routines-5eb9d19608e31a2d9fe5a9e2708e4628524c29d9.tar.gz |
Add force_eval_ and eval_as_ helper functions
These functions are needed to force a particular behaviour on the
compiler: force the evaluation of an expression for the fenv
side-effects, or evaluate an expression without excess precision.
Currently only eval_as_double is used.
Diffstat (limited to 'math/expf.c')
-rw-r--r-- | math/expf.c | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/math/expf.c b/math/expf.c index 5d42b76..9d32c08 100644 --- a/math/expf.c +++ b/math/expf.c @@ -84,7 +84,7 @@ expf (float x) ki = converttoint (z); #else # define SHIFT __exp2f_data.shift - kd = (double) (z + SHIFT); /* Rounding to double precision is required. */ + kd = eval_as_double (z + SHIFT); ki = asuint64 (kd); kd -= SHIFT; #endif |