kmem_cache: Call constructor/destructor on each alloc/free

This has a few benefits. First, it fixes a regression that "Rework
generic memory allocation interfaces" appears to have triggered in
splat's slab_reap and slab_age tests. Second, it makes porting code from
Illumos to ZFSOnLinux easier. Third, it has the side effect of making
reclaim from slab caches that specify reclaim functions an order of
magnitude faster. The splat slab_reap test usually took 30 to 40
seconds. With this change, it takes 3 to 4.

Signed-off-by: Richard Yao <ryao@gentoo.org>

module/spl/spl-kmem.c

@@ -926,9 +926,6 @@ spl_slab_alloc(spl_kmem_cache_t *skc, int flags)
 		list_add_tail(&sko->sko_list, &sks->sks_free_list);
 	}
 
-	list_for_each_entry(sko, &sks->sks_free_list, sko_list)
-		if (skc->skc_ctor)
-			skc->skc_ctor(sko->sko_addr, skc->skc_private, flags);
 out:
 	if (rc) {
 		if (skc->skc_flags & KMC_OFFSLAB)
@@ -1034,9 +1031,6 @@ spl_slab_reclaim(spl_kmem_cache_t *skc, int count, int flag)
 	list_for_each_entry_safe(sko, n, &sko_list, sko_list) {
 		ASSERT(sko->sko_magic == SKO_MAGIC);
 
-		if (skc->skc_dtor)
-			skc->skc_dtor(sko->sko_addr, skc->skc_private);
-
 		if (skc->skc_flags & KMC_OFFSLAB)
 			kv_free(skc, sko->sko_addr, size);
 	}
@@ -1138,9 +1132,6 @@ spl_emergency_alloc(spl_kmem_cache_t *skc, int flags, void **obj)
 		SRETURN(-EINVAL);
 	}
 
-	if (skc->skc_ctor)
-		skc->skc_ctor(ske->ske_obj, skc->skc_private, flags);
-
 	*obj = ske->ske_obj;
 
 	SRETURN(0);
@@ -1167,9 +1158,6 @@ spl_emergency_free(spl_kmem_cache_t *skc, void *obj)
 	if (unlikely(ske == NULL))
 		SRETURN(-ENOENT);
 
-	if (skc->skc_dtor)
-		skc->skc_dtor(ske->ske_obj, skc->skc_private);
-
 	kfree(ske->ske_obj);
 	kfree(ske);
 
@@ -2051,12 +2039,9 @@ spl_kmem_cache_alloc(spl_kmem_cache_t *skc, int flags)
 		do {
 			int lflags = kmem_flags_convert(flags) | __GFP_COMP;
 			obj = kmem_cache_alloc(slc, lflags);
-			if (obj && skc->skc_ctor)
-				skc->skc_ctor(obj, skc->skc_private, flags);
 		} while ((obj == NULL) && !(flags & KM_NOSLEEP));
 
-		atomic_dec(&skc->skc_ref);
-		SRETURN(obj);
+		SGOTO(ret, 0);
 	}
 
 	local_irq_disable();
@@ -2085,12 +2070,20 @@ spl_kmem_cache_alloc(spl_kmem_cache_t *skc, int flags)
 	ASSERT(obj);
 	ASSERT(IS_P2ALIGNED(obj, skc->skc_obj_align));
 
+ret:
 	/* Pre-emptively migrate object to CPU L1 cache */
-	prefetchw(obj);
+	if (obj) {
+		if (obj && skc->skc_ctor)
+			skc->skc_ctor(obj, skc->skc_private, flags);
+		else
+			prefetchw(obj);
+	}
+
 	atomic_dec(&skc->skc_ref);
 
 	SRETURN(obj);
 }
+
 EXPORT_SYMBOL(spl_kmem_cache_alloc);
 
 /*
@@ -2110,13 +2103,16 @@ spl_kmem_cache_free(spl_kmem_cache_t *skc, void *obj)
 	ASSERT(!test_bit(KMC_BIT_DESTROY, &skc->skc_flags));
 	atomic_inc(&skc->skc_ref);
 
+	/*
+	 * Run the destructor
+	 */
+	if (skc->skc_dtor)
+		skc->skc_dtor(obj, skc->skc_private);
+
 	/*
 	 * Free the object from the Linux underlying Linux slab.
 	 */
 	if (skc->skc_flags & KMC_SLAB) {
-		if (skc->skc_dtor)
-			skc->skc_dtor(obj, skc->skc_private);
-
 		kmem_cache_free(skc->skc_linux_cache, obj);
 		goto out;
 	}