MFC r267387:

Allocating new bucket for bucket zone, never take it from the zone itself,
since it will almost certanly fail.  Take next bigger zone instead.

This situation should not happen with original bucket zones configuration:
"32 Bucket" zone uses "64 Bucket" and vice versa.  But if "64 Bucket" zone
lock is congested, zone may grow its bucket size and start biting itself.

sys/vm/uma_core.c

@@ -381,6 +381,8 @@ bucket_alloc(uma_zone_t zone, void *udata, int flags)
 	if ((uintptr_t)udata & UMA_ZFLAG_CACHEONLY)
 		flags |= M_NOVM;
 	ubz = bucket_zone_lookup(zone->uz_count);
+	if (ubz->ubz_zone == zone && (ubz + 1)->ubz_entries != 0)
+		ubz++;
 	bucket = uma_zalloc_arg(ubz->ubz_zone, udata, flags);
 	if (bucket) {
 #ifdef INVARIANTS