[Bug c/52977] [4.8 Regression] internal compiler error: Segmentation fault with `-x c-header' or `-x cxx-header' option

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52977

--- Comment #3 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-04-16 
10:15:35 UTC ---
ugh...

#0  0x00000000004d7255 in gt_pch_p_14lang_tree_node (this_obj=0x7ffff5bf64c0, 
    x_p=0x7ffff5bf64c0, op=0x8b1a27 <relocate_ptrs>, cookie=0x7fffffffdb20)
    at ./gt-c-decl.h:1448

        case TS_VECTOR:
          if ((void *)(x) == this_obj)
            op (&((*x).generic.vector.typed.type), cookie);
          {
            size_t i0;
            size_t l0 = (size_t)(TYPE_VECTOR_SUBPARTS (TREE_TYPE
((tree)&((*x).generic.vector))));
            for (i0 = 0; i0 != l0; i0++) {
              if ((void *)(x) == this_obj)
                op (&((*x).generic.vector.elts[i0]), cookie);
            }
          }
          break;

so we relocate TREE_TYPE before accessing it to find the number of elements.

Other than trying to teach gengtype to compute all required lengths before
changing pointers I only see the less efficient solution of repeating the
element count in-line :/