https://gcc.gnu.org/bugzilla/show_bug.cgi?id=128081
Roger Sayle <roger at nextmovesoftware dot com> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |roger at nextmovesoftware dot
com
Assignee|unassigned at gcc dot gnu.org |roger at
nextmovesoftware dot com
--- Comment #1 from Roger Sayle <roger at nextmovesoftware dot com> ---
This isn't (or can't be) the fold_const_pow change, but might be a subtle
(unintended) change in semantics of real_isinteger where somewhere is relying
on NaN, Inf or -Inf being considered an integer. The behaviour of
real_isinteger was previously undocumented for qNaN and infinities, and I
thought my change retained (and clarified) the original semantics, but these
failures indicate otherwise. A simple fix would be to revert the
real_isinteger changes (or tweak the sematics), but I'm investigating which
callers are making assumptions about integers. For example, fold-const.cc's
integer_valued_real_single_p is documented to do what the riscv test case is
expecting.
