https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87776

            Bug ID: 87776
           Summary: [9 Regression] Compile time hog during RPO VN
           Product: gcc
           Version: 9.0
            Status: UNCONFIRMED
          Keywords: compile-time-hog
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: asolokha at gmx dot com
  Target Milestone: ---

gcc-9.0.0-alpha20181021 snapshot (r265361) takes indefinite time to compile the
following snippet w/ -O2 (-O3, -Ofast, -Os) --param rpo-vn-max-loop-depth=2:

int wp;

void
ff (void)
{
  int *s3 = ℘

  for (wp = 0; wp < 1; ++wp)
    {
      if (wp != 0)
        {
 o8:
          ;
        }
      else
        {
 b7:
          while (wp < 1)
            {
            }
        }

      for (*s3 = 0; *s3 < 1; ++*s3)
        {
          for (wp = 0; wp < 1; ++wp)
            for (wp = 0; wp < 1; ++wp)
              {
              }

          for (wp = 0; wp < 1; ++wp)
            goto o8;

          for (wp = 0; wp < 1; ++wp)
            goto b7;
        }
    }
}

% timeout 10 gcc-9.0.0-alpha20181021 -O2 --param rpo-vn-max-loop-depth=2 -c
uppnycas.c
zsh: exit 124   timeout 10 gcc-9.0.0-alpha20181021 -O2 --param
rpo-vn-max-loop-depth=2 -c

Or is it something to expect w/ such a small argument for
rpo-vn-max-loop-depth?

Reply via email to