Illumos 5818 - zfs {ref}compressratio is incorrect with 4k sector size

5818 zfs {ref}compressratio is incorrect with 4k sector size
Reviewed by: Alex Reece <alex@delphix.com>
Reviewed by: George Wilson <george@delphix.com>
Reviewed by: Richard Elling <richard.elling@richardelling.com>
Reviewed by: Steven Hartland <killing@multiplay.co.uk>
Approved by: Albert Lee <trisk@omniti.com>

References:
  https://www.illumos.org/issues/5818
  illumos/illumos-gate@81cd5c5

Ported-by: Don Brady <don.brady@intel.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3432

cmd/ztest/ztest.c

@@ -20,7 +20,7 @@
  */
 /*
  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2011, 2014 by Delphix. All rights reserved.
+ * Copyright (c) 2011, 2015 by Delphix. All rights reserved.
  * Copyright 2011 Nexenta Systems, Inc.  All rights reserved.
  * Copyright (c) 2013 Steven Hartland. All rights reserved.
  */
@@ -1022,21 +1022,6 @@ ztest_random_spa_version(uint64_t initial_version)
 	return (version);
 }
 
-/*
- * Find the largest ashift used
- */
-static uint64_t
-ztest_spa_get_ashift(void) {
-	uint64_t i;
-	uint64_t ashift = SPA_MINBLOCKSHIFT;
-	vdev_t *rvd = ztest_spa->spa_root_vdev;
-
-	for (i = 0; i < rvd->vdev_children; i++) {
-		ashift = MAX(ashift, rvd->vdev_child[i]->vdev_ashift);
-	}
-	return (ashift);
-}
-
 static int
 ztest_random_blocksize(void)
 {
@@ -1047,7 +1032,8 @@ ztest_random_blocksize(void)
 	int maxbs = SPA_OLD_MAXBLOCKSHIFT;
 	if (spa_maxblocksize(ztest_spa) == SPA_MAXBLOCKSIZE)
 		maxbs = 20;
-	uint64_t block_shift = ztest_random(maxbs - ztest_spa_get_ashift() + 1);
+	uint64_t block_shift =
+	    ztest_random(maxbs - ztest_spa->spa_max_ashift + 1);
 	return (1 << (SPA_MINBLOCKSHIFT + block_shift));
 }
 

include/sys/spa_impl.h

@@ -20,7 +20,7 @@
  */
 /*
  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2011, 2014 by Delphix. All rights reserved.
+ * Copyright (c) 2011, 2015 by Delphix. All rights reserved.
  * Copyright 2011 Nexenta Systems, Inc.  All rights reserved.
  * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
  */
@@ -150,6 +150,8 @@ struct spa {
 	kcondvar_t	spa_evicting_os_cv;	/* Objset Eviction Completion */
 	txg_list_t	spa_vdev_txg_list;	/* per-txg dirty vdev list */
 	vdev_t		*spa_root_vdev;		/* top-level vdev container */
+	int		spa_min_ashift;		/* of vdevs in normal class */
+	int		spa_max_ashift;		/* of vdevs in normal class */
 	uint64_t	spa_config_guid;	/* config pool guid */
 	uint64_t	spa_load_guid;		/* spa_load initialized guid */
 	uint64_t	spa_last_synced_guid;	/* last synced guid */

include/sys/vdev_impl.h

@@ -20,7 +20,7 @@
  */
 /*
  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012, 2014 by Delphix. All rights reserved.
+ * Copyright (c) 2011, 2015 by Delphix. All rights reserved.
  */
 
 #ifndef _SYS_VDEV_IMPL_H
@@ -210,7 +210,7 @@ struct vdev {
 	boolean_t	vdev_isl2cache;	/* was a l2cache device		*/
 	vdev_queue_t	vdev_queue;	/* I/O deadline schedule queue	*/
 	vdev_cache_t	vdev_cache;	/* physical block cache		*/
-	spa_aux_vdev_t	*vdev_aux;	/* for l2cache vdevs		*/
+	spa_aux_vdev_t	*vdev_aux;	/* for l2cache and spares vdevs	*/
 	zio_t		*vdev_probe_zio; /* root of current probe	*/
 	vdev_aux_t	vdev_label_aux;	/* on-disk aux state		*/
 

module/zfs/spa.c

@@ -2245,6 +2245,8 @@ spa_load_impl(spa_t *spa, uint64_t pool_guid, nvlist_t *config,
 		return (error);
 
 	ASSERT(spa->spa_root_vdev == rvd);
+	ASSERT3U(spa->spa_min_ashift, >=, SPA_MINBLOCKSHIFT);
+	ASSERT3U(spa->spa_max_ashift, <=, SPA_MAXBLOCKSHIFT);
 
 	if (type != SPA_IMPORT_ASSEMBLE) {
 		ASSERT(spa_guid(spa) == pool_guid);

module/zfs/spa_misc.c

@@ -20,7 +20,7 @@
  */
 /*
  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2011, 2014 by Delphix. All rights reserved.
+ * Copyright (c) 2011, 2015 by Delphix. All rights reserved.
  * Copyright 2011 Nexenta Systems, Inc.  All rights reserved.
  * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
  */
@@ -623,6 +623,9 @@ spa_add(const char *name, nvlist_t *config, const char *altroot)
 
 	spa->spa_debug = ((zfs_flags & ZFS_DEBUG_SPA) != 0);
 
+	spa->spa_min_ashift = INT_MAX;
+	spa->spa_max_ashift = 0;
+
 	/*
 	 * As a pool is being created, treat all features as disabled by
 	 * setting SPA_FEATURE_DISABLED for all entries in the feature

module/zfs/vdev.c

@@ -22,7 +22,7 @@
 /*
  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
  * Copyright 2011 Nexenta Systems, Inc.  All rights reserved.
- * Copyright (c) 2011, 2014 by Delphix. All rights reserved.
+ * Copyright (c) 2011, 2015 by Delphix. All rights reserved.
  */
 
 #include <sys/zfs_context.h>
@@ -206,8 +206,9 @@ vdev_add_child(vdev_t *pvd, vdev_t *cvd)
 	size_t oldsize, newsize;
 	uint64_t id = cvd->vdev_id;
 	vdev_t **newchild;
+	spa_t *spa = cvd->vdev_spa;
 
-	ASSERT(spa_config_held(cvd->vdev_spa, SCL_ALL, RW_WRITER) == SCL_ALL);
+	ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL);
 	ASSERT(cvd->vdev_parent == NULL);
 
 	cvd->vdev_parent = pvd;
@@ -1336,6 +1337,17 @@ vdev_open(vdev_t *vd)
 		return (error);
 	}
 
+	/*
+	 * Track the min and max ashift values for normal data devices.
+	 */
+	if (vd->vdev_top == vd && vd->vdev_ashift != 0 &&
+	    !vd->vdev_islog && vd->vdev_aux == NULL) {
+		if (vd->vdev_ashift > spa->spa_max_ashift)
+			spa->spa_max_ashift = vd->vdev_ashift;
+		if (vd->vdev_ashift < spa->spa_min_ashift)
+			spa->spa_min_ashift = vd->vdev_ashift;
+	}
+
 	/*
 	 * If a leaf vdev has a DTL, and seems healthy, then kick off a
 	 * resilver.  But don't do this if we are doing a reopen for a scrub,

module/zfs/zio.c

@@ -20,7 +20,7 @@
  */
 /*
  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2011, 2014 by Delphix. All rights reserved.
+ * Copyright (c) 2011, 2015 by Delphix. All rights reserved.
  * Copyright (c) 2011 Nexenta Systems, Inc. All rights reserved.
  */
 
@@ -1195,19 +1195,26 @@ zio_write_bp_init(zio_t *zio)
 			return (ZIO_PIPELINE_CONTINUE);
 		} else {
 			/*
-			 * Round up compressed size to MINBLOCKSIZE and
-			 * zero the tail.
+			 * Round up compressed size up to the ashift
+			 * of the smallest-ashift device, and zero the tail.
+			 * This ensures that the compressed size of the BP
+			 * (and thus compressratio property) are correct,
+			 * in that we charge for the padding used to fill out
+			 * the last sector.
 			 */
-			size_t rounded =
-			    P2ROUNDUP(psize, (size_t)SPA_MINBLOCKSIZE);
-			if (rounded > psize) {
-				bzero((char *)cbuf + psize, rounded - psize);
-				psize = rounded;
-			}
-			if (psize == lsize) {
+			size_t rounded;
+
+			ASSERT3U(spa->spa_min_ashift, >=, SPA_MINBLOCKSHIFT);
+
+			rounded = (size_t)P2ROUNDUP(psize,
+			    1ULL << spa->spa_min_ashift);
+			if (rounded >= lsize) {
 				compress = ZIO_COMPRESS_OFF;
 				zio_buf_free(cbuf, lsize);
+				psize = lsize;
 			} else {
+				bzero((char *)cbuf + psize, rounded - psize);
+				psize = rounded;
 				zio_push_transform(zio, cbuf,
 				    psize, lsize, NULL);
 			}