block: Inline blk_integrity in struct gendisk

Up until now the_integrity profile has been dynamically allocated and
attached to struct gendisk after the disk has been made active.

This causes problems because NVMe devices need to register the profile
prior to the partition table being read due to a mandatory metadata
buffer requirement. In addition, DM goes through hoops to deal with
preallocating, but not initializing integrity profiles.

Since the integrity profile is small (4 bytes + a pointer), Christoph
suggested moving it to struct gendisk proper. This requires several
changes:

 - Moving the blk_integrity definition to genhd.h.

 - Inlining blk_integrity in struct gendisk.

 - Removing the dynamic allocation code.

 - Adding helper functions which allow gendisk to set up and tear down
   the integrity sysfs dir when a disk is added/deleted.

 - Adding a blk_integrity_revalidate() callback for updating the stable
   pages bdi setting.

 - The calls that depend on whether a device has an integrity profile or
   not now key off of the bi->profile pointer.

 - Simplifying the integrity support routines in DM (Mike Snitzer).

Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Reported-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagig@mellanox.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Jens Axboe <axboe@fb.com>

4 files changed, 53 insertions, 55 deletions

diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c
index e76ed003769e..061152a43730 100644
--- a/drivers/md/dm-table.c
+++ b/drivers/md/dm-table.c
@@ -1014,15 +1014,16 @@ static int dm_table_build_index(struct dm_table *t)
 	return r;
 }
 
+static bool integrity_profile_exists(struct gendisk *disk)
+{
+	return !!blk_get_integrity(disk);
+}
+
 /*
  * Get a disk whose integrity profile reflects the table's profile.
- * If %match_all is true, all devices' profiles must match.
- * If %match_all is false, all devices must at least have an
- * allocated integrity profile; but uninitialized is ok.
  * Returns NULL if integrity support was inconsistent or unavailable.
  */
-static struct gendisk * dm_table_get_integrity_disk(struct dm_table *t,
-						    bool match_all)
+static struct gendisk * dm_table_get_integrity_disk(struct dm_table *t)
 {
 	struct list_head *devices = dm_table_get_devices(t);
 	struct dm_dev_internal *dd = NULL;
@@ -1030,10 +1031,8 @@ static struct gendisk * dm_table_get_integrity_disk(struct dm_table *t,
 
 	list_for_each_entry(dd, devices, list) {
 		template_disk = dd->dm_dev->bdev->bd_disk;
-		if (!blk_get_integrity(template_disk))
+		if (!integrity_profile_exists(template_disk))
 			goto no_integrity;
-		if (!match_all && !blk_integrity_is_initialized(template_disk))
-			continue; /* skip uninitialized profiles */
 		else if (prev_disk &&
 			 blk_integrity_compare(prev_disk, template_disk) < 0)
 			goto no_integrity;
@@ -1052,34 +1051,40 @@ no_integrity:
 }
 
 /*
- * Register the mapped device for blk_integrity support if
- * the underlying devices have an integrity profile.  But all devices
- * may not have matching profiles (checking all devices isn't reliable
+ * Register the mapped device for blk_integrity support if the
+ * underlying devices have an integrity profile.  But all devices may
+ * not have matching profiles (checking all devices isn't reliable
  * during table load because this table may use other DM device(s) which
- * must be resumed before they will have an initialized integity profile).
- * Stacked DM devices force a 2 stage integrity profile validation:
- * 1 - during load, validate all initialized integrity profiles match
- * 2 - during resume, validate all integrity profiles match
+ * must be resumed before they will have an initialized integity
+ * profile).  Consequently, stacked DM devices force a 2 stage integrity
+ * profile validation: First pass during table load, final pass during
+ * resume.
  */
-static int dm_table_prealloc_integrity(struct dm_table *t, struct mapped_device *md)
+static int dm_table_register_integrity(struct dm_table *t)
 {
+	struct mapped_device *md = t->md;
 	struct gendisk *template_disk = NULL;
 
-	template_disk = dm_table_get_integrity_disk(t, false);
+	template_disk = dm_table_get_integrity_disk(t);
 	if (!template_disk)
 		return 0;
 
-	if (!blk_integrity_is_initialized(dm_disk(md))) {
+	if (!integrity_profile_exists(dm_disk(md))) {
 		t->integrity_supported = 1;
-		return blk_integrity_register(dm_disk(md), NULL);
+		/*
+		 * Register integrity profile during table load; we can do
+		 * this because the final profile must match during resume.
+		 */
+		blk_integrity_register(dm_disk(md),
+				       blk_get_integrity(template_disk));
+		return 0;
 	}
 
 	/*
-	 * If DM device already has an initalized integrity
+	 * If DM device already has an initialized integrity
 	 * profile the new profile should not conflict.
 	 */
-	if (blk_integrity_is_initialized(template_disk) &&
-	    blk_integrity_compare(dm_disk(md), template_disk) < 0) {
+	if (blk_integrity_compare(dm_disk(md), template_disk) < 0) {
 		DMWARN("%s: conflict with existing integrity profile: "
 		       "%s profile mismatch",
 		       dm_device_name(t->md),
@@ -1087,7 +1092,7 @@ static int dm_table_prealloc_integrity(struct dm_table *t, struct mapped_device
 		return 1;
 	}
 
-	/* Preserve existing initialized integrity profile */
+	/* Preserve existing integrity profile */
 	t->integrity_supported = 1;
 	return 0;
 }
@@ -1112,7 +1117,7 @@ int dm_table_complete(struct dm_table *t)
 		return r;
 	}
 
-	r = dm_table_prealloc_integrity(t, t->md);
+	r = dm_table_register_integrity(t);
 	if (r) {
 		DMERR("could not register integrity profile.");
 		return r;
@@ -1278,29 +1283,30 @@ combine_limits:
 }
 
 /*
- * Set the integrity profile for this device if all devices used have
- * matching profiles.  We're quite deep in the resume path but still
- * don't know if all devices (particularly DM devices this device
- * may be stacked on) have matching profiles.  Even if the profiles
- * don't match we have no way to fail (to resume) at this point.
+ * Verify that all devices have an integrity profile that matches the
+ * DM device's registered integrity profile.  If the profiles don't
+ * match then unregister the DM device's integrity profile.
  */
-static void dm_table_set_integrity(struct dm_table *t)
+static void dm_table_verify_integrity(struct dm_table *t)
 {
 	struct gendisk *template_disk = NULL;
 
-	if (!blk_get_integrity(dm_disk(t->md)))
-		return;
+	if (t->integrity_supported) {
+		/*
+		 * Verify that the original integrity profile
+		 * matches all the devices in this table.
+		 */
+		template_disk = dm_table_get_integrity_disk(t);
+		if (template_disk &&
+		    blk_integrity_compare(dm_disk(t->md), template_disk) >= 0)
+			return;
+	}
 
-	template_disk = dm_table_get_integrity_disk(t, true);
-	if (template_disk)
-		blk_integrity_register(dm_disk(t->md),
-				       blk_get_integrity(template_disk));
-	else if (blk_integrity_is_initialized(dm_disk(t->md)))
-		DMWARN("%s: device no longer has a valid integrity profile",
-		       dm_device_name(t->md));
-	else
+	if (integrity_profile_exists(dm_disk(t->md))) {
 		DMWARN("%s: unable to establish an integrity profile",
 		       dm_device_name(t->md));
+		blk_integrity_unregister(dm_disk(t->md));
+	}
 }
 
 static int device_flush_capable(struct dm_target *ti, struct dm_dev *dev,
@@ -1500,7 +1506,7 @@ void dm_table_set_restrictions(struct dm_table *t, struct request_queue *q,
 	else
 		queue_flag_set_unlocked(QUEUE_FLAG_NO_SG_MERGE, q);
 
-	dm_table_set_integrity(t);
+	dm_table_verify_integrity(t);
 
 	/*
 	 * Determine whether or not this queue's I/O timings contribute
diff --git a/drivers/md/md.c b/drivers/md/md.c
index c702de18207a..2af9d590e1a0 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -1962,12 +1962,9 @@ int md_integrity_register(struct mddev *mddev)
 	 * All component devices are integrity capable and have matching
 	 * profiles, register the common profile for the md device.
 	 */
-	if (blk_integrity_register(mddev->gendisk,
-			bdev_get_integrity(reference->bdev)) != 0) {
-		printk(KERN_ERR "md: failed to register integrity for %s\n",
-			mdname(mddev));
-		return -EINVAL;
-	}
+	blk_integrity_register(mddev->gendisk,
+			       bdev_get_integrity(reference->bdev));
+
 	printk(KERN_NOTICE "md: data integrity enabled on %s\n", mdname(mddev));
 	if (bioset_integrity_create(mddev->bio_set, BIO_POOL_SIZE)) {
 		printk(KERN_ERR "md: failed to create integrity pool for %s\n",
diff --git a/drivers/nvdimm/core.c b/drivers/nvdimm/core.c
index 7df89b547ae1..e85848caf8d2 100644
--- a/drivers/nvdimm/core.c
+++ b/drivers/nvdimm/core.c
@@ -405,7 +405,6 @@ int nd_integrity_init(struct gendisk *disk, unsigned long meta_size)
 		.generate_fn = nd_pi_nop_generate_verify,
 		.verify_fn = nd_pi_nop_generate_verify,
 	};
-	int ret;
 
 	if (meta_size == 0)
 		return 0;
@@ -414,10 +413,7 @@ int nd_integrity_init(struct gendisk *disk, unsigned long meta_size)
 	bi.tuple_size = meta_size;
 	bi.tag_size = meta_size;
 
-	ret = blk_integrity_register(disk, &bi);
-	if (ret)
-		return ret;
-
+	blk_integrity_register(disk, &bi);
 	blk_queue_max_integrity_segments(disk->queue, 1);
 
 	return 0;
diff --git a/drivers/nvme/host/pci.c b/drivers/nvme/host/pci.c
index 04e3d60a1e45..8d2aeaaa3895 100644
--- a/drivers/nvme/host/pci.c
+++ b/drivers/nvme/host/pci.c
@@ -538,7 +538,7 @@ static void nvme_dif_remap(struct request *req,
 	virt = bip_get_seed(bip);
 	phys = nvme_block_nr(ns, blk_rq_pos(req));
 	nlb = (blk_rq_bytes(req) >> ns->lba_shift);
-	ts = ns->disk->integrity->tuple_size;
+	ts = ns->disk->integrity.tuple_size;
 
 	for (i = 0; i < nlb; i++, virt++, phys++) {
 		pi = (struct t10_pi_tuple *)p;
@@ -2044,8 +2044,7 @@ static int nvme_revalidate_disk(struct gendisk *disk)
 	ns->pi_type = pi_type;
 	blk_queue_logical_block_size(ns->queue, bs);
 
-	if (ns->ms && !blk_get_integrity(disk) && (disk->flags & GENHD_FL_UP) &&
-								!ns->ext)
+	if (ns->ms && !ns->ext)
 		nvme_init_integrity(ns);
 
 	if (ns->ms && !(ns->ms == 8 && ns->pi_type) && !blk_get_integrity(disk))