Currently, KASAN always poisons/unpoisons large kmalloc allocations (the ones that fall back onto page_alloc) without checking whether the poisoning was excluded due to page_alloc sampling. We need to add checks similar to the one in kasan_mempool_poison_pages to kasan_mempool_poison_object, kasan_kmalloc_large, and kasan_kfree_large.
This becomes obsolete if https://bugzilla.kernel.org/show_bug.cgi?id=218322 is implemented.