Remove mg_allocators (#17192)

Previous code allowed each metaslab group to have different number
of allocators.  But in practice it worked only for embedded SLOGs,
relying on a number of conditions and creating a significant mine
field if any of those change.  I just stepped on one myself.

This change makes all groups to have spa_alloc_count allocators.
It may cost us extra 192 bytes of memory per normal top-level vdev
on large systems, but I find it a small price for cleaner and more
reliable code.

Signed-off-by:	Alexander Motin <mav@FreeBSD.org>
Sponsored by:	iXsystems, Inc.
Fixes #17188
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Reviewed-by: Paul Dagnelie <pcd@delphix.com>

include/sys/metaslab.h

@@ -118,7 +118,7 @@ uint64_t metaslab_class_get_deferred(metaslab_class_t *);
 void metaslab_space_update(vdev_t *, metaslab_class_t *,
     int64_t, int64_t, int64_t);
 
-metaslab_group_t *metaslab_group_create(metaslab_class_t *, vdev_t *, int);
+metaslab_group_t *metaslab_group_create(metaslab_class_t *, vdev_t *);
 void metaslab_group_destroy(metaslab_group_t *);
 void metaslab_group_activate(metaslab_group_t *);
 void metaslab_group_passivate(metaslab_group_t *);

include/sys/metaslab_impl.h

@@ -269,7 +269,6 @@ struct metaslab_group {
 	kmutex_t		mg_ms_disabled_lock;
 	kcondvar_t		mg_ms_disabled_cv;
 
-	int			mg_allocators;
 	metaslab_group_allocator_t	mg_allocator[];
 };
 

module/zfs/metaslab.c

@@ -1037,12 +1037,13 @@ metaslab_sort_by_flushed(const void *va, const void *vb)
 }
 
 metaslab_group_t *
-metaslab_group_create(metaslab_class_t *mc, vdev_t *vd, int allocators)
+metaslab_group_create(metaslab_class_t *mc, vdev_t *vd)
 {
+	spa_t *spa = mc->mc_spa;
 	metaslab_group_t *mg;
 
 	mg = kmem_zalloc(offsetof(metaslab_group_t,
-	    mg_allocator[allocators]), KM_SLEEP);
+	    mg_allocator[spa->spa_alloc_count]), KM_SLEEP);
 	mutex_init(&mg->mg_lock, NULL, MUTEX_DEFAULT, NULL);
 	mutex_init(&mg->mg_ms_disabled_lock, NULL, MUTEX_DEFAULT, NULL);
 	cv_init(&mg->mg_ms_disabled_cv, NULL, CV_DEFAULT, NULL);
@@ -1053,9 +1054,8 @@ metaslab_group_create(metaslab_class_t *mc, vdev_t *vd, int allocators)
 	mg->mg_activation_count = 0;
 	mg->mg_initialized = B_FALSE;
 	mg->mg_no_free_space = B_TRUE;
-	mg->mg_allocators = allocators;
 
-	for (int i = 0; i < allocators; i++) {
+	for (int i = 0; i < spa->spa_alloc_count; i++) {
 		metaslab_group_allocator_t *mga = &mg->mg_allocator[i];
 		zfs_refcount_create_tracked(&mga->mga_queue_depth);
 	}
@@ -1066,6 +1066,8 @@ metaslab_group_create(metaslab_class_t *mc, vdev_t *vd, int allocators)
 void
 metaslab_group_destroy(metaslab_group_t *mg)
 {
+	spa_t *spa = mg->mg_class->mc_spa;
+
 	ASSERT(mg->mg_prev == NULL);
 	ASSERT(mg->mg_next == NULL);
 	/*
@@ -1080,12 +1082,12 @@ metaslab_group_destroy(metaslab_group_t *mg)
 	mutex_destroy(&mg->mg_ms_disabled_lock);
 	cv_destroy(&mg->mg_ms_disabled_cv);
 
-	for (int i = 0; i < mg->mg_allocators; i++) {
+	for (int i = 0; i < spa->spa_alloc_count; i++) {
 		metaslab_group_allocator_t *mga = &mg->mg_allocator[i];
 		zfs_refcount_destroy(&mga->mga_queue_depth);
 	}
 	kmem_free(mg, offsetof(metaslab_group_t,
-	    mg_allocator[mg->mg_allocators]));
+	    mg_allocator[spa->spa_alloc_count]));
 }
 
 void
@@ -1167,7 +1169,7 @@ metaslab_group_passivate(metaslab_group_t *mg)
 	taskq_wait_outstanding(spa->spa_metaslab_taskq, 0);
 	spa_config_enter(spa, locks & ~(SCL_ZIO - 1), spa, RW_WRITER);
 	metaslab_group_alloc_update(mg);
-	for (int i = 0; i < mg->mg_allocators; i++) {
+	for (int i = 0; i < spa->spa_alloc_count; i++) {
 		metaslab_group_allocator_t *mga = &mg->mg_allocator[i];
 		metaslab_t *msp = mga->mga_primary;
 		if (msp != NULL) {
@@ -3574,7 +3576,7 @@ metaslab_passivate_allocator(metaslab_group_t *mg, metaslab_t *msp,
 	mutex_enter(&mg->mg_lock);
 	ASSERT3P(msp->ms_group, ==, mg);
 	ASSERT3S(0, <=, msp->ms_allocator);
-	ASSERT3U(msp->ms_allocator, <, mg->mg_allocators);
+	ASSERT3U(msp->ms_allocator, <, mg->mg_class->mc_spa->spa_alloc_count);
 
 	metaslab_group_allocator_t *mga = &mg->mg_allocator[msp->ms_allocator];
 	if (msp->ms_primary) {
@@ -3700,7 +3702,7 @@ metaslab_group_preload(metaslab_group_t *mg)
 		}
 
 		VERIFY(taskq_dispatch(spa->spa_metaslab_taskq, metaslab_preload,
-		    msp, TQ_SLEEP | (m <= mg->mg_allocators ? TQ_FRONT : 0))
+		    msp, TQ_SLEEP | (m <= spa->spa_alloc_count ? TQ_FRONT : 0))
 		    != TASKQID_INVALID);
 	}
 	mutex_exit(&mg->mg_lock);
@@ -4896,10 +4898,9 @@ metaslab_group_alloc(metaslab_group_t *mg, zio_alloc_list_t *zal,
 	}
 
 	/*
-	 * If we don't have enough metaslabs active to fill the entire array, we
+	 * If we don't have enough metaslabs active, we just use the 0th slot.
-	 * just use the 0th slot.
 	 */
-	if (mg->mg_ms_ready < mg->mg_allocators * 3)
+	if (allocator >= mg->mg_ms_ready / 3)
 		allocator = 0;
 	metaslab_group_allocator_t *mga = &mg->mg_allocator[allocator];
 

module/zfs/vdev.c

@@ -1490,12 +1490,11 @@ vdev_metaslab_group_create(vdev_t *vd)
 			mc = spa_normal_class(spa);
 		}
 
-		vd->vdev_mg = metaslab_group_create(mc, vd,
+		vd->vdev_mg = metaslab_group_create(mc, vd);
-		    spa->spa_alloc_count);
 
 		if (!vd->vdev_islog) {
 			vd->vdev_log_mg = metaslab_group_create(
-			    spa_embedded_log_class(spa), vd, 1);
+			    spa_embedded_log_class(spa), vd);
 		}
 
 		/*