[Bug target/66326] Floating point exception with -mfpmath=387 and -fcilkplus.

["ryan.burn at gmail dot com" <gcc-bugzilla at gcc dot gnu dot org>]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66326

ryan.burn at gmail dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |ryan.burn at gmail dot com

--- Comment #4 from ryan.burn at gmail dot com ---
I'm getting a similar problem testing with 6.0. Only the flags -mfpmath doesn't
seem to make a difference for the case I'm testing. But it runs successfully
with -O1 or above but fails with -O0. Not sure which particular option is
causing the problem.

Here's the code I'm testing with

///////////////////////////////////////////////////////////////////////////
#include <cilk/cilk.h>
#include <vector>
#include <random>

template <class T>
void do_not_optimize_away(T&& x) {
  asm volatile("" : "+r"(x));
}

const int N = 1'000'000;

auto compute() {
  std::vector<double> v(N);
  auto rng = std::mt19937{std::random_device{}()};
  std::uniform_real_distribution<double> dist(0, 1);
  for (int i = 0; i < N; ++i) v[i] = std::log(std::sqrt(dist(rng)));
  return v;
}

int main() {
  std::vector<double> v1, v2, v3;
  cilk_spawn [&] { v1 = compute(); }();
  cilk_spawn [&] { v2 = compute(); }();
  v3 = compute();
  do_not_optimize_away(v1.data());
  do_not_optimize_away(v2.data());
  do_not_optimize_away(v3.data());
  return 0;
}
///////////////////////////////////////////////////////////////////////////

// O1 level - good
rnburn@localhost ~/test/gcc_cilk $ /home/rnburn/local/bin/g++ -std=c++1z -O1
-fcilkplus -std=c++14   -march=haswell t2.cpp 
rnburn@localhost ~/test/gcc_cilk $ ./a.out 

// O0 level - fails
rnburn@localhost ~/test/gcc_cilk $ /home/rnburn/local/bin/g++ -std=c++1z -O0
-fcilkplus -std=c++14   -march=haswell t2.cpp 
rnburn@localhost ~/test/gcc_cilk $ ./a.out 
Floating point exception