------- Comment #13 from dominiq at lps dot ens dot fr 2007-03-20 16:08 -------
> You can propose that we only enable sincos transformation
> if TARGET_HAS_SINCOS is set, I wouldn't necessarily object to
> that. (The targets I care for have a sincos)
Sound reasonable: replacing:
return (TARGET_HAS_SINCOS
|| TARGET_C99_FUNCTIONS)
&& optimize;
by
return TARGET_HAS_SINCOS
&& optimize;
in gcc/tree-ssa-math-opts.c, isn't it?
I can even do a preliminary test to check that it
does not break anything.
What's bother me is that i suspect the problem is present
for all non-Linux platforms and to have no feedback from them.
If you have some idea about the way to trigger their interest,
it would be nice.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31249
