uma: Avoid excessive per-CPU draining

After commit 389a3fa, uma_reclaim_domain(UMA_RECLAIM_DRAIN_CPU)
calls uma_zone_reclaim_domain(UMA_RECLAIM_DRAIN_CPU) twice on each zone
in addition to globally draining per-CPU caches. This was unintended
and is unnecessarily slow; in particular, draining per-CPU caches
requires binding to each CPU.

Stop draining per-CPU caches when visiting each zone, just do it once in
pcpu_cache_drain_safe() to minimize the amount of expensive sched_bind()
calls.

Fixes:		389a3fa ("uma: Add UMA_ZONE_UNMANAGED")
MFC after:	1 week
Sponsored by:	Klara, Inc.
Sponsored by:	NetApp, Inc.
Reviewed by:	gallatin, kib
Differential Revision:	https://reviews.freebsd.org/D49349

Author: markjdb

sys/vm/uma_core.c

@@ -5339,6 +5339,13 @@ uma_reclaim_domain(int req, int domain)
 		zone_foreach(uma_reclaim_domain_cb, &args);
 		break;
 	case UMA_RECLAIM_DRAIN_CPU:
+		/*
+		 * Reclaim globally visible free items from all zones, then drain
+		 * per-CPU buckets, then reclaim items freed while draining.
+		 * This approach minimizes expensive context switching needed to
+		 * drain each zone's per-CPU buckets.
+		 */
+		args.req = UMA_RECLAIM_DRAIN;
 		zone_foreach(uma_reclaim_domain_cb, &args);
 		pcpu_cache_drain_safe(NULL);
 		zone_foreach(uma_reclaim_domain_cb, &args);