[Bug tree-optimization/50693] Loop optimization restricted by GOTOs

["rguenth at gcc dot gnu.org" <gcc-bugzilla at gcc dot gnu dot org>]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50693

--- Comment #15 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-10-11 14:34:47 UTC ---
Note that it doesn't handle memset though, and the convoluted loop wouldn't be
easy to detect either.

    size_t i = 0;
    bool loop_cond = i < n;
    while (loop_cond) {
        goto copy_block;
        loop_back:
            loop_cond = i < n;
    }
    goto end;
    copy_block:
        v->chars[i] = c;
        i++;
        goto loop_back;
    end:
        v->chars[n] = '\0';
        return v;

is simply trying to be too clever.  I can't even understand that source ;)

What probably causes this is that we don't merge the blocks

  # i_29 = PHI <i_10(3), 0(2)>
copy_block:
  D.3506_7 = v_20->chars;
  D.3507_8 = D.3506_7 + i_29;
  *D.3507_8 = c_9(D);
  i_10 = i_29 + 1;
  goto <bb 3> (loop_back);

loop_back:
  loop_cond_11 = i_10 < n_3(D);
  if (loop_cond_11 != 0)
    goto <bb 4> (copy_block);
  else
    goto <bb 5> (end);

even though it's a fallthru edge.  We don't do this to preserve user labels
for debugging (and mind, no code-gen differences between -g0 vs. -g).