Comment 56 for bug 569273

from valgrinding the above test (not for the feint of heart I warn you) the following results are observed;

==24310== 181,440 bytes in 180 blocks are possibly lost in loss record 3,196 of 3,198
==24310== at 0x4C2732A: memalign (vg_replace_malloc.c:581)
==24310== by 0x4C27383: posix_memalign (vg_replace_malloc.c:709)
==24310== by 0x7A04397: slab_allocator_alloc_chunk (gslice.c:1152)
==24310== by 0x7A04C58: g_slice_alloc (gslice.c:678)
==24310== by 0x7A04D35: g_slice_alloc0 (gslice.c:848)
==24310== by 0x54BA3DF: gdk_event_new (gdkevents.c:356)
==24310== by 0x54D1793: _gdk_make_event (gdkwindow.c:9882)
==24310== by 0x54D5F48: gdk_window_update_visibility (gdkwindow.c:828)
==24310== by 0x54D6148: recompute_visible_regions_internal (gdkwindow.c:1034)
==24310== by 0x54D64C0: set_viewable (gdkwindow.c:6675)
==24310== by 0x54D6489: set_viewable (gdkwindow.c:6683)
==24310== by 0x54D6489: set_viewable (gdkwindow.c:6683)
==24310==
==24310== 395,760 bytes in 97 blocks are possibly lost in loss record 3,197 of 3,198
==24310== at 0x4C2732A: memalign (vg_replace_malloc.c:581)
==24310== by 0x4C27383: posix_memalign (vg_replace_malloc.c:709)
==24310== by 0x7A04397: slab_allocator_alloc_chunk (gslice.c:1152)
==24310== by 0x7A04C58: g_slice_alloc (gslice.c:678)
==24310== by 0x7A04D35: g_slice_alloc0 (gslice.c:848)
==24310== by 0x7179ACA: g_type_create_instance (gtype.c:1867)
==24310== by 0x715C2D8: g_object_constructor (gobject.c:1482)
==24310== by 0x715B139: g_object_newv (gobject.c:1266)
==24310== by 0x715B86B: g_object_new (gobject.c:1178)
==24310== by 0x54DCEC3: gdk_window_new (gdkwindow.c:1330)
==24310== by 0x4F91561: gtk_menu_item_realize (gtkmenuitem.c:1189)
==24310== by 0x7155AFD: g_closure_invoke (gclosure.c:766)

This points to the problem being located in gslice in slab_allocator_alloc_chunk, there are other leaks in the gtkwindow but that isn't related to this bug.