[Bug other/58712] [4.9 Regression] issues found by --enable-checking=valgrind

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

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58712

--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
(In reply to Markus Trippelsdorf from comment #0) 
> 3)
> ==1269== Invalid write of size 8
> ==1269==    at 0x63BDBD: iterative_hash_canonical_type(tree_node*, unsigned
> int) (gimple.c:3220)
> ==1269==    by 0xC224F1: htab_find_slot (hashtab.c:704)
> ==1269==    by 0x641818: gimple_register_canonical_type(tree_node*)
> (gimple.c:3488)
> ==1269==    by 0x504AEB: lto_read_decls(lto_file_decl_data*, void const*,
> vec<ld_plugin_symbol_resolution, va_heap, vl_ptr>) (lto.c:2495)
> ==1269==    by 0x505CBB: lto_main() (lto.c:2786)
> ==1269==    by 0x7B3A95: compile_file() (toplev.c:546)
> ==1269==    by 0x7B5619: toplev_main(int, char**) (toplev.c:1893)
> ==1269==    by 0x4ED5A6D: (below main) (in /lib64/libc-2.18.90.so)
> ==1269==  Address 0x40325a0 is not stack'd, malloc'd or (recently) free'd
> 
> This is an older issue, see PR55358 for more info.

By inspection I have created a patch and am bootstrapping/testing it for this
issue.