From 0ba959774e93911caff596de6391f085fb640ac4 Mon Sep 17 00:00:00 2001
From: Guoqing Jiang <gqjiang@suse.com>
Date: Wed, 1 Mar 2017 17:30:29 +0800
Subject: md-cluster: use sync way to handle METADATA_UPDATED msg

Previously, when node received METADATA_UPDATED msg, it just
need to wakeup mddev->thread, then md_reload_sb will be called
eventually.

We taken the asynchronous way to avoid a deadlock issue, the
deadlock issue could happen when one node is receiving the
METADATA_UPDATED msg (wants reconfig_mutex) and trying to run
the path:

md_check_recovery -> mddev_trylock(hold reconfig_mutex)
                  -> md_update_sb-metadata_update_start
		     (want EX on token however token is
		      got by the sending node)

Since we will support resizing for clustered raid, and we
need the metadata update handling to be synchronous so that
the initiating node can detect failure, so we need to change
the way for handling METADATA_UPDATED msg.

But, we obviously need to avoid above deadlock with the
sync way. To make this happen, we considered to not hold
reconfig_mutex to call md_reload_sb, if some other thread
has already taken reconfig_mutex and waiting for the 'token',
then process_recvd_msg() can safely call md_reload_sb()
without taking the mutex. This is because we can be certain
that no other thread will take the mutex, and we also certain
that the actions performed by md_reload_sb() won't interfere
with anything that the other thread is in the middle of.

To make this more concrete, we added a new cinfo->state bit
        MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD

Which is set in lock_token() just before dlm_lock_sync() is
called, and cleared just after. As lock_token() is always
called with reconfig_mutex() held (the specific case is the
resync_info_update which is distinguished well in previous
patch), if process_recvd_msg() finds that the new bit is set,
then the mutex must be held by some other thread, and it will
keep waiting.

So process_metadata_update() can call md_reload_sb() if either
mddev_trylock() succeeds, or if MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD
is set. The tricky bit is what to do if neither of these apply.
We need to wait. Fortunately mddev_unlock() always calls wake_up()
on mddev->thread->wqueue. So we can get lock_token() to call
wake_up() on that when it sets the bit.

There are also some related changes inside this commit:
1. remove RELOAD_SB related codes since there are not valid anymore.
2. mddev is added into md_cluster_info then we can get mddev inside
   lock_token.
3. add new parameter for lock_token to distinguish reconfig_mutex
   is held or not.

And, we need to set MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD in below:
1. set it before unregister thread, otherwise a deadlock could
   appear if stop a resyncing array.
   This is because md_unregister_thread(&cinfo->recv_thread) is
   blocked by recv_daemon -> process_recvd_msg
			  -> process_metadata_update.
   To resolve the issue, MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD is
   also need to be set before unregister thread.
2. set it in metadata_update_start to fix another deadlock.
	a. Node A sends METADATA_UPDATED msg (held Token lock).
	b. Node B wants to do resync, and is blocked since it can't
	   get Token lock, but MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD is
	   not set since the callchain
	   (md_do_sync -> sync_request
        	       -> resync_info_update
		       -> sendmsg
		       -> lock_comm -> lock_token)
	   doesn't hold reconfig_mutex.
	c. Node B trys to update sb (held reconfig_mutex), but stopped
	   at wait_event() in metadata_update_start since we have set
	   MD_CLUSTER_SEND_LOCK flag in lock_comm (step 2).
	d. Then Node B receives METADATA_UPDATED msg from A, of course
	   recv_daemon is blocked forever.
   Since metadata_update_start always calls lock_token with reconfig_mutex,
   we need to set MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD here as well, and
   lock_token don't need to set it twice unless lock_token is invoked from
   lock_comm.

Finally, thanks to Neil for his great idea and help!

Reviewed-by: NeilBrown <neilb@suse.com>
Signed-off-by: Guoqing Jiang <gqjiang@suse.com>
Signed-off-by: Shaohua Li <shli@fb.com>
---
 drivers/md/md-cluster.c | 82 +++++++++++++++++++++++++++++++++++++++----------
 drivers/md/md.c         |  4 ---
 drivers/md/md.h         |  3 --
 3 files changed, 66 insertions(+), 23 deletions(-)

(limited to 'drivers')

diff --git a/drivers/md/md-cluster.c b/drivers/md/md-cluster.c
index 321ecac23027..5cf0a9d29bf0 100644
--- a/drivers/md/md-cluster.c
+++ b/drivers/md/md-cluster.c
@@ -67,9 +67,10 @@ struct resync_info {
  * set up all the related infos such as bitmap and personality */
 #define		MD_CLUSTER_ALREADY_IN_CLUSTER		6
 #define		MD_CLUSTER_PENDING_RECV_EVENT		7
-
+#define 	MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD		8
 
 struct md_cluster_info {
+	struct mddev *mddev; /* the md device which md_cluster_info belongs to */
 	/* dlm lock space and resources for clustered raid. */
 	dlm_lockspace_t *lockspace;
 	int slot_number;
@@ -523,11 +524,17 @@ static void process_add_new_disk(struct mddev *mddev, struct cluster_msg *cmsg)
 
 static void process_metadata_update(struct mddev *mddev, struct cluster_msg *msg)
 {
+	int got_lock = 0;
 	struct md_cluster_info *cinfo = mddev->cluster_info;
 	mddev->good_device_nr = le32_to_cpu(msg->raid_slot);
-	set_bit(MD_RELOAD_SB, &mddev->flags);
+
 	dlm_lock_sync(cinfo->no_new_dev_lockres, DLM_LOCK_CR);
-	md_wakeup_thread(mddev->thread);
+	wait_event(mddev->thread->wqueue,
+		   (got_lock = mddev_trylock(mddev)) ||
+		    test_bit(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state));
+	md_reload_sb(mddev, mddev->good_device_nr);
+	if (got_lock)
+		mddev_unlock(mddev);
 }
 
 static void process_remove_disk(struct mddev *mddev, struct cluster_msg *msg)
@@ -646,11 +653,29 @@ out:
  * Takes the lock on the TOKEN lock resource so no other
  * node can communicate while the operation is underway.
  */
-static int lock_token(struct md_cluster_info *cinfo)
+static int lock_token(struct md_cluster_info *cinfo, bool mddev_locked)
 {
-	int error;
+	int error, set_bit = 0;
+	struct mddev *mddev = cinfo->mddev;
 
+	/*
+	 * If resync thread run after raid1d thread, then process_metadata_update
+	 * could not continue if raid1d held reconfig_mutex (and raid1d is blocked
+	 * since another node already got EX on Token and waitting the EX of Ack),
+	 * so let resync wake up thread in case flag is set.
+	 */
+	if (mddev_locked && !test_bit(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD,
+				      &cinfo->state)) {
+		error = test_and_set_bit_lock(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD,
+					      &cinfo->state);
+		WARN_ON_ONCE(error);
+		md_wakeup_thread(mddev->thread);
+		set_bit = 1;
+	}
 	error = dlm_lock_sync(cinfo->token_lockres, DLM_LOCK_EX);
+	if (set_bit)
+		clear_bit_unlock(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state);
+
 	if (error)
 		pr_err("md-cluster(%s:%d): failed to get EX on TOKEN (%d)\n",
 				__func__, __LINE__, error);
@@ -663,12 +688,12 @@ static int lock_token(struct md_cluster_info *cinfo)
 /* lock_comm()
  * Sets the MD_CLUSTER_SEND_LOCK bit to lock the send channel.
  */
-static int lock_comm(struct md_cluster_info *cinfo)
+static int lock_comm(struct md_cluster_info *cinfo, bool mddev_locked)
 {
 	wait_event(cinfo->wait,
 		   !test_and_set_bit(MD_CLUSTER_SEND_LOCK, &cinfo->state));
 
-	return lock_token(cinfo);
+	return lock_token(cinfo, mddev_locked);
 }
 
 static void unlock_comm(struct md_cluster_info *cinfo)
@@ -743,11 +768,12 @@ failed_message:
 	return error;
 }
 
-static int sendmsg(struct md_cluster_info *cinfo, struct cluster_msg *cmsg)
+static int sendmsg(struct md_cluster_info *cinfo, struct cluster_msg *cmsg,
+		   bool mddev_locked)
 {
 	int ret;
 
-	lock_comm(cinfo);
+	lock_comm(cinfo, mddev_locked);
 	ret = __sendmsg(cinfo, cmsg);
 	unlock_comm(cinfo);
 	return ret;
@@ -834,6 +860,7 @@ static int join(struct mddev *mddev, int nodes)
 	mutex_init(&cinfo->recv_mutex);
 
 	mddev->cluster_info = cinfo;
+	cinfo->mddev = mddev;
 
 	memset(str, 0, 64);
 	sprintf(str, "%pU", mddev->uuid);
@@ -908,6 +935,7 @@ static int join(struct mddev *mddev, int nodes)
 
 	return 0;
 err:
+	set_bit(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state);
 	md_unregister_thread(&cinfo->recovery_thread);
 	md_unregister_thread(&cinfo->recv_thread);
 	lockres_free(cinfo->message_lockres);
@@ -943,7 +971,7 @@ static void resync_bitmap(struct mddev *mddev)
 	int err;
 
 	cmsg.type = cpu_to_le32(BITMAP_NEEDS_SYNC);
-	err = sendmsg(cinfo, &cmsg);
+	err = sendmsg(cinfo, &cmsg, 1);
 	if (err)
 		pr_err("%s:%d: failed to send BITMAP_NEEDS_SYNC message (%d)\n",
 			__func__, __LINE__, err);
@@ -963,6 +991,7 @@ static int leave(struct mddev *mddev)
 	if (cinfo->slot_number > 0 && mddev->recovery_cp != MaxSector)
 		resync_bitmap(mddev);
 
+	set_bit(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state);
 	md_unregister_thread(&cinfo->recovery_thread);
 	md_unregister_thread(&cinfo->recv_thread);
 	lockres_free(cinfo->message_lockres);
@@ -997,16 +1026,30 @@ static int slot_number(struct mddev *mddev)
 static int metadata_update_start(struct mddev *mddev)
 {
 	struct md_cluster_info *cinfo = mddev->cluster_info;
+	int ret;
+
+	/*
+	 * metadata_update_start is always called with the protection of
+	 * reconfig_mutex, so set WAITING_FOR_TOKEN here.
+	 */
+	ret = test_and_set_bit_lock(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD,
+				    &cinfo->state);
+	WARN_ON_ONCE(ret);
+	md_wakeup_thread(mddev->thread);
 
 	wait_event(cinfo->wait,
 		   !test_and_set_bit(MD_CLUSTER_SEND_LOCK, &cinfo->state) ||
 		   test_and_clear_bit(MD_CLUSTER_SEND_LOCKED_ALREADY, &cinfo->state));
 
 	/* If token is already locked, return 0 */
-	if (cinfo->token_lockres->mode == DLM_LOCK_EX)
+	if (cinfo->token_lockres->mode == DLM_LOCK_EX) {
+		clear_bit_unlock(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state);
 		return 0;
+	}
 
-	return lock_token(cinfo);
+	ret = lock_token(cinfo, 1);
+	clear_bit_unlock(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state);
+	return ret;
 }
 
 static int metadata_update_finish(struct mddev *mddev)
@@ -1069,7 +1112,14 @@ static int resync_info_update(struct mddev *mddev, sector_t lo, sector_t hi)
 	cmsg.low = cpu_to_le64(lo);
 	cmsg.high = cpu_to_le64(hi);
 
-	return sendmsg(cinfo, &cmsg);
+	/*
+	 * mddev_lock is held if resync_info_update is called from
+	 * resync_finish (md_reap_sync_thread -> resync_finish)
+	 */
+	if (lo == 0 && hi == 0)
+		return sendmsg(cinfo, &cmsg, 1);
+	else
+		return sendmsg(cinfo, &cmsg, 0);
 }
 
 static int resync_finish(struct mddev *mddev)
@@ -1119,7 +1169,7 @@ static int add_new_disk(struct mddev *mddev, struct md_rdev *rdev)
 	cmsg.type = cpu_to_le32(NEWDISK);
 	memcpy(cmsg.uuid, uuid, 16);
 	cmsg.raid_slot = cpu_to_le32(rdev->desc_nr);
-	lock_comm(cinfo);
+	lock_comm(cinfo, 1);
 	ret = __sendmsg(cinfo, &cmsg);
 	if (ret)
 		return ret;
@@ -1179,7 +1229,7 @@ static int remove_disk(struct mddev *mddev, struct md_rdev *rdev)
 	struct md_cluster_info *cinfo = mddev->cluster_info;
 	cmsg.type = cpu_to_le32(REMOVE);
 	cmsg.raid_slot = cpu_to_le32(rdev->desc_nr);
-	return sendmsg(cinfo, &cmsg);
+	return sendmsg(cinfo, &cmsg, 1);
 }
 
 static int lock_all_bitmaps(struct mddev *mddev)
@@ -1243,7 +1293,7 @@ static int gather_bitmaps(struct md_rdev *rdev)
 
 	cmsg.type = cpu_to_le32(RE_ADD);
 	cmsg.raid_slot = cpu_to_le32(rdev->desc_nr);
-	err = sendmsg(cinfo, &cmsg);
+	err = sendmsg(cinfo, &cmsg, 1);
 	if (err)
 		goto out;
 
diff --git a/drivers/md/md.c b/drivers/md/md.c
index f6ae1d67bcd0..83f325077dc8 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -8385,7 +8385,6 @@ void md_check_recovery(struct mddev *mddev)
 		(mddev->sb_flags & ~ (1<<MD_SB_CHANGE_PENDING)) ||
 		test_bit(MD_RECOVERY_NEEDED, &mddev->recovery) ||
 		test_bit(MD_RECOVERY_DONE, &mddev->recovery) ||
-		test_bit(MD_RELOAD_SB, &mddev->flags) ||
 		(mddev->external == 0 && mddev->safemode == 1) ||
 		(mddev->safemode == 2 && ! atomic_read(&mddev->writes_pending)
 		 && !mddev->in_sync && mddev->recovery_cp == MaxSector)
@@ -8434,9 +8433,6 @@ void md_check_recovery(struct mddev *mddev)
 						rdev->raid_disk < 0)
 					md_kick_rdev_from_array(rdev);
 			}
-
-			if (test_and_clear_bit(MD_RELOAD_SB, &mddev->flags))
-				md_reload_sb(mddev, mddev->good_device_nr);
 		}
 
 		if (!mddev->external) {
diff --git a/drivers/md/md.h b/drivers/md/md.h
index dde8ecb760c8..1c00160b09f9 100644
--- a/drivers/md/md.h
+++ b/drivers/md/md.h
@@ -219,9 +219,6 @@ enum mddev_flags {
 				 * it then */
 	MD_JOURNAL_CLEAN,	/* A raid with journal is already clean */
 	MD_HAS_JOURNAL,		/* The raid array has journal feature set */
-	MD_RELOAD_SB,		/* Reload the superblock because another node
-				 * updated it.
-				 */
 	MD_CLUSTER_RESYNC_LOCKED, /* cluster raid only, which means node
 				   * already took resync lock, need to
 				   * release the lock */
-- 
cgit v1.2.3


From 7da3d203330305695f38302c076ee0d32c005ede Mon Sep 17 00:00:00 2001
From: Guoqing Jiang <gqjiang@suse.com>
Date: Wed, 1 Mar 2017 16:42:38 +0800
Subject: md-cluster: add CHANGE_CAPACITY message type

The msg type CHANGE_CAPACITY is introduced to support
resize clustered raid in later patch, and it is sent
after all the nodes have the same sync_size, receiver
node just need to set new capacity once received this
msg.

Reviewed-by: NeilBrown <neilb@suse.com>
Signed-off-by: Guoqing Jiang <gqjiang@suse.com>
Signed-off-by: Shaohua Li <shli@fb.com>
---
 drivers/md/md-cluster.c | 5 +++++
 1 file changed, 5 insertions(+)

(limited to 'drivers')

diff --git a/drivers/md/md-cluster.c b/drivers/md/md-cluster.c
index 5cf0a9d29bf0..8b7d55bf5aa2 100644
--- a/drivers/md/md-cluster.c
+++ b/drivers/md/md-cluster.c
@@ -104,6 +104,7 @@ enum msg_type {
 	REMOVE,
 	RE_ADD,
 	BITMAP_NEEDS_SYNC,
+	CHANGE_CAPACITY,
 };
 
 struct cluster_msg {
@@ -579,6 +580,10 @@ static int process_recvd_msg(struct mddev *mddev, struct cluster_msg *msg)
 	case METADATA_UPDATED:
 		process_metadata_update(mddev, msg);
 		break;
+	case CHANGE_CAPACITY:
+		set_capacity(mddev->gendisk, mddev->array_sectors);
+		revalidate_disk(mddev->gendisk);
+		break;
 	case RESYNCING:
 		process_suspend_info(mddev, le32_to_cpu(msg->slot),
 				     le64_to_cpu(msg->low),
-- 
cgit v1.2.3


From b98938d16a108d0201d54b8db1337dac71aebfdd Mon Sep 17 00:00:00 2001
From: Guoqing Jiang <gqjiang@suse.com>
Date: Wed, 1 Mar 2017 16:42:39 +0800
Subject: md-cluster: introduce cluster_check_sync_size

Support resize is a little complex for clustered
raid, since we need to ensure all the nodes share
the same knowledge about the size of raid.

We achieve the goal by check the sync_size which
is in each node's bitmap, we can only change the
capacity after cluster_check_sync_size returns 0.

Also, get_bitmap_from_slot is added to get a slot's
bitmap. And we exported some funcs since they are
used in cluster_check_sync_size().

We can also reuse get_bitmap_from_slot to remove
redundant code existed in bitmap_copy_from_slot.

Reviewed-by: NeilBrown <neilb@suse.com>
Signed-off-by: Guoqing Jiang <gqjiang@suse.com>
Signed-off-by: Shaohua Li <shli@fb.com>
---
 drivers/md/bitmap.c     | 41 ++++++++++++++++++++++++---------
 drivers/md/bitmap.h     |  2 ++
 drivers/md/md-cluster.c | 60 +++++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 93 insertions(+), 10 deletions(-)

(limited to 'drivers')

diff --git a/drivers/md/bitmap.c b/drivers/md/bitmap.c
index 9fb2ccac958a..b6fa55a3cff8 100644
--- a/drivers/md/bitmap.c
+++ b/drivers/md/bitmap.c
@@ -471,6 +471,7 @@ void bitmap_update_sb(struct bitmap *bitmap)
 	kunmap_atomic(sb);
 	write_page(bitmap, bitmap->storage.sb_page, 1);
 }
+EXPORT_SYMBOL(bitmap_update_sb);
 
 /* print out the bitmap file superblock */
 void bitmap_print_sb(struct bitmap *bitmap)
@@ -1727,7 +1728,7 @@ void bitmap_flush(struct mddev *mddev)
 /*
  * free memory that was allocated
  */
-static void bitmap_free(struct bitmap *bitmap)
+void bitmap_free(struct bitmap *bitmap)
 {
 	unsigned long k, pages;
 	struct bitmap_page *bp;
@@ -1761,6 +1762,7 @@ static void bitmap_free(struct bitmap *bitmap)
 	kfree(bp);
 	kfree(bitmap);
 }
+EXPORT_SYMBOL(bitmap_free);
 
 void bitmap_destroy(struct mddev *mddev)
 {
@@ -1920,6 +1922,27 @@ out:
 }
 EXPORT_SYMBOL_GPL(bitmap_load);
 
+struct bitmap *get_bitmap_from_slot(struct mddev *mddev, int slot)
+{
+	int rv = 0;
+	struct bitmap *bitmap;
+
+	bitmap = bitmap_create(mddev, slot);
+	if (IS_ERR(bitmap)) {
+		rv = PTR_ERR(bitmap);
+		return ERR_PTR(rv);
+	}
+
+	rv = bitmap_init_from_disk(bitmap, 0);
+	if (rv) {
+		bitmap_free(bitmap);
+		return ERR_PTR(rv);
+	}
+
+	return bitmap;
+}
+EXPORT_SYMBOL(get_bitmap_from_slot);
+
 /* Loads the bitmap associated with slot and copies the resync information
  * to our bitmap
  */
@@ -1929,14 +1952,13 @@ int bitmap_copy_from_slot(struct mddev *mddev, int slot,
 	int rv = 0, i, j;
 	sector_t block, lo = 0, hi = 0;
 	struct bitmap_counts *counts;
-	struct bitmap *bitmap = bitmap_create(mddev, slot);
-
-	if (IS_ERR(bitmap))
-		return PTR_ERR(bitmap);
+	struct bitmap *bitmap;
 
-	rv = bitmap_init_from_disk(bitmap, 0);
-	if (rv)
-		goto err;
+	bitmap = get_bitmap_from_slot(mddev, slot);
+	if (IS_ERR(bitmap)) {
+		pr_err("%s can't get bitmap from slot %d\n", __func__, slot);
+		return -1;
+	}
 
 	counts = &bitmap->counts;
 	for (j = 0; j < counts->chunks; j++) {
@@ -1963,8 +1985,7 @@ int bitmap_copy_from_slot(struct mddev *mddev, int slot,
 	bitmap_unplug(mddev->bitmap);
 	*low = lo;
 	*high = hi;
-err:
-	bitmap_free(bitmap);
+
 	return rv;
 }
 EXPORT_SYMBOL_GPL(bitmap_copy_from_slot);
diff --git a/drivers/md/bitmap.h b/drivers/md/bitmap.h
index 5b6dd63dda91..9f761097aab2 100644
--- a/drivers/md/bitmap.h
+++ b/drivers/md/bitmap.h
@@ -267,8 +267,10 @@ void bitmap_daemon_work(struct mddev *mddev);
 
 int bitmap_resize(struct bitmap *bitmap, sector_t blocks,
 		  int chunksize, int init);
+struct bitmap *get_bitmap_from_slot(struct mddev *mddev, int slot);
 int bitmap_copy_from_slot(struct mddev *mddev, int slot,
 				sector_t *lo, sector_t *hi, bool clear_bits);
+void bitmap_free(struct bitmap *bitmap);
 #endif
 
 #endif
diff --git a/drivers/md/md-cluster.c b/drivers/md/md-cluster.c
index 8b7d55bf5aa2..92c3c51ede4d 100644
--- a/drivers/md/md-cluster.c
+++ b/drivers/md/md-cluster.c
@@ -1091,6 +1091,66 @@ static void metadata_update_cancel(struct mddev *mddev)
 	unlock_comm(cinfo);
 }
 
+/*
+ * return 0 if all the bitmaps have the same sync_size
+ */
+int cluster_check_sync_size(struct mddev *mddev)
+{
+	int i, rv;
+	bitmap_super_t *sb;
+	unsigned long my_sync_size, sync_size = 0;
+	int node_num = mddev->bitmap_info.nodes;
+	int current_slot = md_cluster_ops->slot_number(mddev);
+	struct bitmap *bitmap = mddev->bitmap;
+	char str[64];
+	struct dlm_lock_resource *bm_lockres;
+
+	sb = kmap_atomic(bitmap->storage.sb_page);
+	my_sync_size = sb->sync_size;
+	kunmap_atomic(sb);
+
+	for (i = 0; i < node_num; i++) {
+		if (i == current_slot)
+			continue;
+
+		bitmap = get_bitmap_from_slot(mddev, i);
+		if (IS_ERR(bitmap)) {
+			pr_err("can't get bitmap from slot %d\n", i);
+			return -1;
+		}
+
+		/*
+		 * If we can hold the bitmap lock of one node then
+		 * the slot is not occupied, update the sb.
+		 */
+		snprintf(str, 64, "bitmap%04d", i);
+		bm_lockres = lockres_init(mddev, str, NULL, 1);
+		if (!bm_lockres) {
+			pr_err("md-cluster: Cannot initialize %s\n", str);
+			lockres_free(bm_lockres);
+			return -1;
+		}
+		bm_lockres->flags |= DLM_LKF_NOQUEUE;
+		rv = dlm_lock_sync(bm_lockres, DLM_LOCK_PW);
+		if (!rv)
+			bitmap_update_sb(bitmap);
+		lockres_free(bm_lockres);
+
+		sb = kmap_atomic(bitmap->storage.sb_page);
+		if (sync_size == 0)
+			sync_size = sb->sync_size;
+		else if (sync_size != sb->sync_size) {
+			kunmap_atomic(sb);
+			bitmap_free(bitmap);
+			return -1;
+		}
+		kunmap_atomic(sb);
+		bitmap_free(bitmap);
+	}
+
+	return (my_sync_size == sync_size) ? 0 : -1;
+}
+
 static int resync_start(struct mddev *mddev)
 {
 	struct md_cluster_info *cinfo = mddev->cluster_info;
-- 
cgit v1.2.3


From 818da59f97d6764edb91fa9d1f63d333d54eb999 Mon Sep 17 00:00:00 2001
From: Guoqing Jiang <gqjiang@suse.com>
Date: Wed, 1 Mar 2017 16:42:40 +0800
Subject: md-cluster: add the support for resize

To update size for cluster raid, we need to make
sure all nodes can perform the change successfully.
However, it is possible that some of them can't do
it due to failure (bitmap_resize could fail). So
we need to consider the issue before we set the
capacity unconditionally, and we use below steps
to perform sanity check.

1. A change the size, then broadcast METADATA_UPDATED
   msg.
2. B and C receive METADATA_UPDATED change the size
   excepts call set_capacity, sync_size is not update
   if the change failed. Also call bitmap_update_sb
   to sync sb to disk.
3. A checks other node's sync_size, if sync_size has
   been updated in all nodes, then send CHANGE_CAPACITY
   msg otherwise send msg to revert previous change.
4. B and C call set_capacity if receive CHANGE_CAPACITY
   msg, otherwise pers->resize will be called to restore
   the old value.

Reviewed-by: NeilBrown <neilb@suse.com>
Signed-off-by: Guoqing Jiang <gqjiang@suse.com>
Signed-off-by: Shaohua Li <shli@fb.com>
---
 drivers/md/md-cluster.c | 76 +++++++++++++++++++++++++++++++++++++++++++++++++
 drivers/md/md-cluster.h |  1 +
 drivers/md/md.c         | 21 ++++++++++----
 3 files changed, 93 insertions(+), 5 deletions(-)

(limited to 'drivers')

diff --git a/drivers/md/md-cluster.c b/drivers/md/md-cluster.c
index 92c3c51ede4d..b21ef58819f6 100644
--- a/drivers/md/md-cluster.c
+++ b/drivers/md/md-cluster.c
@@ -1151,6 +1151,81 @@ int cluster_check_sync_size(struct mddev *mddev)
 	return (my_sync_size == sync_size) ? 0 : -1;
 }
 
+/*
+ * Update the size for cluster raid is a little more complex, we perform it
+ * by the steps:
+ * 1. hold token lock and update superblock in initiator node.
+ * 2. send METADATA_UPDATED msg to other nodes.
+ * 3. The initiator node continues to check each bitmap's sync_size, if all
+ *    bitmaps have the same value of sync_size, then we can set capacity and
+ *    let other nodes to perform it. If one node can't update sync_size
+ *    accordingly, we need to revert to previous value.
+ */
+static void update_size(struct mddev *mddev, sector_t old_dev_sectors)
+{
+	struct md_cluster_info *cinfo = mddev->cluster_info;
+	struct cluster_msg cmsg;
+	struct md_rdev *rdev;
+	int ret = 0;
+	int raid_slot = -1;
+
+	md_update_sb(mddev, 1);
+	lock_comm(cinfo, 1);
+
+	memset(&cmsg, 0, sizeof(cmsg));
+	cmsg.type = cpu_to_le32(METADATA_UPDATED);
+	rdev_for_each(rdev, mddev)
+		if (rdev->raid_disk >= 0 && !test_bit(Faulty, &rdev->flags)) {
+			raid_slot = rdev->desc_nr;
+			break;
+		}
+	if (raid_slot >= 0) {
+		cmsg.raid_slot = cpu_to_le32(raid_slot);
+		/*
+		 * We can only change capiticy after all the nodes can do it,
+		 * so need to wait after other nodes already received the msg
+		 * and handled the change
+		 */
+		ret = __sendmsg(cinfo, &cmsg);
+		if (ret) {
+			pr_err("%s:%d: failed to send METADATA_UPDATED msg\n",
+			       __func__, __LINE__);
+			unlock_comm(cinfo);
+			return;
+		}
+	} else {
+		pr_err("md-cluster: No good device id found to send\n");
+		unlock_comm(cinfo);
+		return;
+	}
+
+	/*
+	 * check the sync_size from other node's bitmap, if sync_size
+	 * have already updated in other nodes as expected, send an
+	 * empty metadata msg to permit the change of capacity
+	 */
+	if (cluster_check_sync_size(mddev) == 0) {
+		memset(&cmsg, 0, sizeof(cmsg));
+		cmsg.type = cpu_to_le32(CHANGE_CAPACITY);
+		ret = __sendmsg(cinfo, &cmsg);
+		if (ret)
+			pr_err("%s:%d: failed to send CHANGE_CAPACITY msg\n",
+			       __func__, __LINE__);
+		set_capacity(mddev->gendisk, mddev->array_sectors);
+		revalidate_disk(mddev->gendisk);
+	} else {
+		/* revert to previous sectors */
+		ret = mddev->pers->resize(mddev, old_dev_sectors);
+		if (!ret)
+			revalidate_disk(mddev->gendisk);
+		ret = __sendmsg(cinfo, &cmsg);
+		if (ret)
+			pr_err("%s:%d: failed to send METADATA_UPDATED msg\n",
+			       __func__, __LINE__);
+	}
+	unlock_comm(cinfo);
+}
+
 static int resync_start(struct mddev *mddev)
 {
 	struct md_cluster_info *cinfo = mddev->cluster_info;
@@ -1396,6 +1471,7 @@ static struct md_cluster_operations cluster_ops = {
 	.gather_bitmaps = gather_bitmaps,
 	.lock_all_bitmaps = lock_all_bitmaps,
 	.unlock_all_bitmaps = unlock_all_bitmaps,
+	.update_size = update_size,
 };
 
 static int __init cluster_init(void)
diff --git a/drivers/md/md-cluster.h b/drivers/md/md-cluster.h
index e765499ba591..274016177983 100644
--- a/drivers/md/md-cluster.h
+++ b/drivers/md/md-cluster.h
@@ -27,6 +27,7 @@ struct md_cluster_operations {
 	int (*gather_bitmaps)(struct md_rdev *rdev);
 	int (*lock_all_bitmaps)(struct mddev *mddev);
 	void (*unlock_all_bitmaps)(struct mddev *mddev);
+	void (*update_size)(struct mddev *mddev, sector_t old_dev_sectors);
 };
 
 #endif /* _MD_CLUSTER_H */
diff --git a/drivers/md/md.c b/drivers/md/md.c
index 83f325077dc8..72ef3f18ac9a 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -6493,10 +6493,7 @@ static int update_size(struct mddev *mddev, sector_t num_sectors)
 	struct md_rdev *rdev;
 	int rv;
 	int fit = (num_sectors == 0);
-
-	/* cluster raid doesn't support update size */
-	if (mddev_is_clustered(mddev))
-		return -EINVAL;
+	sector_t old_dev_sectors = mddev->dev_sectors;
 
 	if (mddev->pers->resize == NULL)
 		return -EINVAL;
@@ -6525,7 +6522,9 @@ static int update_size(struct mddev *mddev, sector_t num_sectors)
 	}
 	rv = mddev->pers->resize(mddev, num_sectors);
 	if (!rv) {
-		if (mddev->queue) {
+		if (mddev_is_clustered(mddev))
+			md_cluster_ops->update_size(mddev, old_dev_sectors);
+		else if (mddev->queue) {
 			set_capacity(mddev->gendisk, mddev->array_sectors);
 			revalidate_disk(mddev->gendisk);
 		}
@@ -8743,6 +8742,18 @@ static void check_sb_changes(struct mddev *mddev, struct md_rdev *rdev)
 	int role, ret;
 	char b[BDEVNAME_SIZE];
 
+	/*
+	 * If size is changed in another node then we need to
+	 * do resize as well.
+	 */
+	if (mddev->dev_sectors != le64_to_cpu(sb->size)) {
+		ret = mddev->pers->resize(mddev, le64_to_cpu(sb->size));
+		if (ret)
+			pr_info("md-cluster: resize failed\n");
+		else
+			bitmap_update_sb(mddev->bitmap);
+	}
+
 	/* Check for change of roles in the active devices */
 	rdev_for_each(rdev2, mddev) {
 		if (test_bit(Faulty, &rdev2->flags))
-- 
cgit v1.2.3


From 535ae4eb1225f19e1d1848c65eafea8b7e9112f4 Mon Sep 17 00:00:00 2001
From: Shaohua Li <shli@fb.com>
Date: Wed, 15 Feb 2017 19:37:32 -0800
Subject: md/raid5: prioritize stripes for writeback

In raid5-cache writeback mode, we have two types of stripes to handle.
- stripes which aren't cached yet
- stripes which are cached and flushing out to raid disks

Upperlayer is more sensistive to latency of the first type of stripes
generally. But we only one handle list for all these stripes, where the
two types of stripes are mixed together. When reclaim flushes a lot of
stripes, the first type of stripes could be noticeably delayed. On the
other hand, if the log space is tight, we'd like to handle the second
type of stripes faster and free log space.

This patch destinguishes the two types stripes. They are added into
different handle list. When we try to get a stripe to handl, we prefer
the first type of stripes unless log space is tight.

This should have no impact for !writeback case.

Signed-off-by: Shaohua Li <shli@fb.com>
---
 drivers/md/raid5.c | 48 +++++++++++++++++++++++++++++++++++++++---------
 drivers/md/raid5.h |  2 ++
 2 files changed, 41 insertions(+), 9 deletions(-)

(limited to 'drivers')

diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index ed5cd705b985..5a28bd9b5b5f 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -176,6 +176,13 @@ static int stripe_operations_active(struct stripe_head *sh)
 	       test_bit(STRIPE_COMPUTE_RUN, &sh->state);
 }
 
+static bool stripe_is_lowprio(struct stripe_head *sh)
+{
+	return (test_bit(STRIPE_R5C_FULL_STRIPE, &sh->state) ||
+		test_bit(STRIPE_R5C_PARTIAL_STRIPE, &sh->state)) &&
+	       !test_bit(STRIPE_R5C_CACHING, &sh->state);
+}
+
 static void raid5_wakeup_stripe_thread(struct stripe_head *sh)
 {
 	struct r5conf *conf = sh->raid_conf;
@@ -191,7 +198,10 @@ static void raid5_wakeup_stripe_thread(struct stripe_head *sh)
 	if (list_empty(&sh->lru)) {
 		struct r5worker_group *group;
 		group = conf->worker_groups + cpu_to_group(cpu);
-		list_add_tail(&sh->lru, &group->handle_list);
+		if (stripe_is_lowprio(sh))
+			list_add_tail(&sh->lru, &group->loprio_list);
+		else
+			list_add_tail(&sh->lru, &group->handle_list);
 		group->stripes_cnt++;
 		sh->group = group;
 	}
@@ -254,7 +264,12 @@ static void do_release_stripe(struct r5conf *conf, struct stripe_head *sh,
 			clear_bit(STRIPE_DELAYED, &sh->state);
 			clear_bit(STRIPE_BIT_DELAY, &sh->state);
 			if (conf->worker_cnt_per_group == 0) {
-				list_add_tail(&sh->lru, &conf->handle_list);
+				if (stripe_is_lowprio(sh))
+					list_add_tail(&sh->lru,
+							&conf->loprio_list);
+				else
+					list_add_tail(&sh->lru,
+							&conf->handle_list);
 			} else {
 				raid5_wakeup_stripe_thread(sh);
 				return;
@@ -5172,19 +5187,27 @@ static struct bio *chunk_aligned_read(struct mddev *mddev, struct bio *raid_bio)
  */
 static struct stripe_head *__get_priority_stripe(struct r5conf *conf, int group)
 {
-	struct stripe_head *sh = NULL, *tmp;
+	struct stripe_head *sh, *tmp;
 	struct list_head *handle_list = NULL;
-	struct r5worker_group *wg = NULL;
+	struct r5worker_group *wg;
+	bool second_try = !r5c_is_writeback(conf->log);
+	bool try_loprio = test_bit(R5C_LOG_TIGHT, &conf->cache_state);
 
+again:
+	wg = NULL;
+	sh = NULL;
 	if (conf->worker_cnt_per_group == 0) {
-		handle_list = &conf->handle_list;
+		handle_list = try_loprio ? &conf->loprio_list :
+					&conf->handle_list;
 	} else if (group != ANY_GROUP) {
-		handle_list = &conf->worker_groups[group].handle_list;
+		handle_list = try_loprio ? &conf->worker_groups[group].loprio_list :
+				&conf->worker_groups[group].handle_list;
 		wg = &conf->worker_groups[group];
 	} else {
 		int i;
 		for (i = 0; i < conf->group_cnt; i++) {
-			handle_list = &conf->worker_groups[i].handle_list;
+			handle_list = try_loprio ? &conf->worker_groups[i].loprio_list :
+				&conf->worker_groups[i].handle_list;
 			wg = &conf->worker_groups[i];
 			if (!list_empty(handle_list))
 				break;
@@ -5235,8 +5258,13 @@ static struct stripe_head *__get_priority_stripe(struct r5conf *conf, int group)
 		wg = NULL;
 	}
 
-	if (!sh)
-		return NULL;
+	if (!sh) {
+		if (second_try)
+			return NULL;
+		second_try = true;
+		try_loprio = !try_loprio;
+		goto again;
+	}
 
 	if (wg) {
 		wg->stripes_cnt--;
@@ -6546,6 +6574,7 @@ static int alloc_thread_groups(struct r5conf *conf, int cnt,
 
 		group = &(*worker_groups)[i];
 		INIT_LIST_HEAD(&group->handle_list);
+		INIT_LIST_HEAD(&group->loprio_list);
 		group->conf = conf;
 		group->workers = workers + i * cnt;
 
@@ -6773,6 +6802,7 @@ static struct r5conf *setup_conf(struct mddev *mddev)
 	init_waitqueue_head(&conf->wait_for_stripe);
 	init_waitqueue_head(&conf->wait_for_overlap);
 	INIT_LIST_HEAD(&conf->handle_list);
+	INIT_LIST_HEAD(&conf->loprio_list);
 	INIT_LIST_HEAD(&conf->hold_list);
 	INIT_LIST_HEAD(&conf->delayed_list);
 	INIT_LIST_HEAD(&conf->bitmap_list);
diff --git a/drivers/md/raid5.h b/drivers/md/raid5.h
index 4bb27b97bf6b..6b9d2e839e6d 100644
--- a/drivers/md/raid5.h
+++ b/drivers/md/raid5.h
@@ -542,6 +542,7 @@ struct r5worker {
 
 struct r5worker_group {
 	struct list_head handle_list;
+	struct list_head loprio_list;
 	struct r5conf *conf;
 	struct r5worker *workers;
 	int stripes_cnt;
@@ -608,6 +609,7 @@ struct r5conf {
 						  */
 
 	struct list_head	handle_list; /* stripes needing handling */
+	struct list_head	loprio_list; /* low priority stripes */
 	struct list_head	hold_list; /* preread ready stripes */
 	struct list_head	delayed_list; /* stripes that have plugged requests */
 	struct list_head	bitmap_list; /* stripes delaying awaiting bitmap update */
-- 
cgit v1.2.3


From 84890c03b6c5d7e8d76ea5e20b6aaf7e7ad410f0 Mon Sep 17 00:00:00 2001
From: Shaohua Li <shli@fb.com>
Date: Wed, 15 Feb 2017 19:58:05 -0800
Subject: md/raid5-cache: bump flush stripe batch size

Bump the flush stripe batch size to 2048. For my 12 disks raid
array, the stripes takes:
12 * 4k * 2048 = 96MB

This is still quite small. A hardware raid card generally has 1GB size,
which we suggest the raid5-cache has similar cache size.

The advantage of a big batch size is we can dispatch a lot of IO in the
same time, then we can do some scheduling to make better IO pattern.

Last patch prioritizes stripes, so we don't worry about a big flush
stripe batch will starve normal stripes.

Signed-off-by: Shaohua Li <shli@fb.com>
---
 drivers/md/raid5-cache.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

(limited to 'drivers')

diff --git a/drivers/md/raid5-cache.c b/drivers/md/raid5-cache.c
index 3f307be01b10..738e180d3740 100644
--- a/drivers/md/raid5-cache.c
+++ b/drivers/md/raid5-cache.c
@@ -43,7 +43,7 @@
 /* wake up reclaim thread periodically */
 #define R5C_RECLAIM_WAKEUP_INTERVAL (30 * HZ)
 /* start flush with these full stripes */
-#define R5C_FULL_STRIPE_FLUSH_BATCH 256
+#define R5C_FULL_STRIPE_FLUSH_BATCH(conf) (conf->max_nr_stripes / 4)
 /* reclaim stripes in groups */
 #define R5C_RECLAIM_STRIPE_GROUP (NR_STRIPE_HASH_LOCKS * 2)
 
@@ -381,7 +381,7 @@ void r5c_check_cached_full_stripe(struct r5conf *conf)
 	 * or a full stripe (chunk size / 4k stripes).
 	 */
 	if (atomic_read(&conf->r5c_cached_full_stripes) >=
-	    min(R5C_FULL_STRIPE_FLUSH_BATCH,
+	    min(R5C_FULL_STRIPE_FLUSH_BATCH(conf),
 		conf->chunk_sectors >> STRIPE_SHIFT))
 		r5l_wake_reclaim(conf->log, 0);
 }
@@ -1393,7 +1393,7 @@ static void r5c_do_reclaim(struct r5conf *conf)
 		stripes_to_flush = R5C_RECLAIM_STRIPE_GROUP;
 	else if (total_cached > conf->min_nr_stripes * 1 / 2 ||
 		 atomic_read(&conf->r5c_cached_full_stripes) - flushing_full >
-		 R5C_FULL_STRIPE_FLUSH_BATCH)
+		 R5C_FULL_STRIPE_FLUSH_BATCH(conf))
 		/*
 		 * if stripe cache pressure moderate, or if there is many full
 		 * stripes,flush all full stripes
-- 
cgit v1.2.3


From aaf9f12ebfafd1ea603d61ead6dbcf456a86e0f3 Mon Sep 17 00:00:00 2001
From: Shaohua Li <shli@fb.com>
Date: Fri, 3 Mar 2017 22:06:12 -0800
Subject: md/raid5: sort bios

Previous patch (raid5: only dispatch IO from raid5d for harddisk raid)
defers IO dispatching. The goal is to create better IO pattern. At that
time, we don't sort the deffered IO and hope the block layer can do IO
merge and sort. Now the raid5-cache writeback could create large amount
of bios. And if we enable muti-thread for stripe handling, we can't
control when to dispatch IO to raid disks. In a lot of time, we are
dispatching IO which block layer can't do merge effectively.

This patch moves further for the IO dispatching defer. We accumulate
bios, but we don't dispatch all the bios after a threshold is met. This
'dispatch partial portion of bios' stragety allows bios coming in a
large time window are sent to disks together. At the dispatching time,
there is large chance the block layer can merge the bios. To make this
more effective, we dispatch IO in ascending order. This increases
request merge chance and reduces disk seek.

Signed-off-by: Shaohua Li <shli@fb.com>
---
 drivers/md/raid5.c | 138 +++++++++++++++++++++++++++++++++++++++++++----------
 drivers/md/raid5.h |  14 +++++-
 2 files changed, 126 insertions(+), 26 deletions(-)

(limited to 'drivers')

diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index 5a28bd9b5b5f..013398ce2080 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -58,6 +58,7 @@
 #include <linux/sched/signal.h>
 
 #include <trace/events/block.h>
+#include <linux/list_sort.h>
 
 #include "md.h"
 #include "raid5.h"
@@ -878,41 +879,107 @@ static int use_new_offset(struct r5conf *conf, struct stripe_head *sh)
 	return 1;
 }
 
-static void flush_deferred_bios(struct r5conf *conf)
+static void dispatch_bio_list(struct bio_list *tmp)
 {
-	struct bio_list tmp;
 	struct bio *bio;
 
-	if (!conf->batch_bio_dispatch || !conf->group_cnt)
+	while ((bio = bio_list_pop(tmp)))
+		generic_make_request(bio);
+}
+
+static int cmp_stripe(void *priv, struct list_head *a, struct list_head *b)
+{
+	const struct r5pending_data *da = list_entry(a,
+				struct r5pending_data, sibling);
+	const struct r5pending_data *db = list_entry(b,
+				struct r5pending_data, sibling);
+	if (da->sector > db->sector)
+		return 1;
+	if (da->sector < db->sector)
+		return -1;
+	return 0;
+}
+
+static void dispatch_defer_bios(struct r5conf *conf, int target,
+				struct bio_list *list)
+{
+	struct r5pending_data *data;
+	struct list_head *first, *next = NULL;
+	int cnt = 0;
+
+	if (conf->pending_data_cnt == 0)
+		return;
+
+	list_sort(NULL, &conf->pending_list, cmp_stripe);
+
+	first = conf->pending_list.next;
+
+	/* temporarily move the head */
+	if (conf->next_pending_data)
+		list_move_tail(&conf->pending_list,
+				&conf->next_pending_data->sibling);
+
+	while (!list_empty(&conf->pending_list)) {
+		data = list_first_entry(&conf->pending_list,
+			struct r5pending_data, sibling);
+		if (&data->sibling == first)
+			first = data->sibling.next;
+		next = data->sibling.next;
+
+		bio_list_merge(list, &data->bios);
+		list_move(&data->sibling, &conf->free_list);
+		cnt++;
+		if (cnt >= target)
+			break;
+	}
+	conf->pending_data_cnt -= cnt;
+	BUG_ON(conf->pending_data_cnt < 0 || cnt < target);
+
+	if (next != &conf->pending_list)
+		conf->next_pending_data = list_entry(next,
+				struct r5pending_data, sibling);
+	else
+		conf->next_pending_data = NULL;
+	/* list isn't empty */
+	if (first != &conf->pending_list)
+		list_move_tail(&conf->pending_list, first);
+}
+
+static void flush_deferred_bios(struct r5conf *conf)
+{
+	struct bio_list tmp = BIO_EMPTY_LIST;
+
+	if (conf->pending_data_cnt == 0)
 		return;
 
-	bio_list_init(&tmp);
 	spin_lock(&conf->pending_bios_lock);
-	bio_list_merge(&tmp, &conf->pending_bios);
-	bio_list_init(&conf->pending_bios);
+	dispatch_defer_bios(conf, conf->pending_data_cnt, &tmp);
+	BUG_ON(conf->pending_data_cnt != 0);
 	spin_unlock(&conf->pending_bios_lock);
 
-	while ((bio = bio_list_pop(&tmp)))
-		generic_make_request(bio);
+	dispatch_bio_list(&tmp);
 }
 
-static void defer_bio_issue(struct r5conf *conf, struct bio *bio)
+static void defer_issue_bios(struct r5conf *conf, sector_t sector,
+				struct bio_list *bios)
 {
-	/*
-	 * change group_cnt will drain all bios, so this is safe
-	 *
-	 * A read generally means a read-modify-write, which usually means a
-	 * randwrite, so we don't delay it
-	 */
-	if (!conf->batch_bio_dispatch || !conf->group_cnt ||
-	    bio_op(bio) == REQ_OP_READ) {
-		generic_make_request(bio);
-		return;
-	}
+	struct bio_list tmp = BIO_EMPTY_LIST;
+	struct r5pending_data *ent;
+
 	spin_lock(&conf->pending_bios_lock);
-	bio_list_add(&conf->pending_bios, bio);
+	ent = list_first_entry(&conf->free_list, struct r5pending_data,
+							sibling);
+	list_move_tail(&ent->sibling, &conf->pending_list);
+	ent->sector = sector;
+	bio_list_init(&ent->bios);
+	bio_list_merge(&ent->bios, bios);
+	conf->pending_data_cnt++;
+	if (conf->pending_data_cnt >= PENDING_IO_MAX)
+		dispatch_defer_bios(conf, PENDING_IO_ONE_FLUSH, &tmp);
+
 	spin_unlock(&conf->pending_bios_lock);
-	md_wakeup_thread(conf->mddev->thread);
+
+	dispatch_bio_list(&tmp);
 }
 
 static void
@@ -925,6 +992,8 @@ static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s)
 	struct r5conf *conf = sh->raid_conf;
 	int i, disks = sh->disks;
 	struct stripe_head *head_sh = sh;
+	struct bio_list pending_bios = BIO_EMPTY_LIST;
+	bool should_defer;
 
 	might_sleep();
 
@@ -941,6 +1010,8 @@ static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s)
 		}
 	}
 
+	should_defer = conf->batch_bio_dispatch && conf->group_cnt;
+
 	for (i = disks; i--; ) {
 		int op, op_flags = 0;
 		int replace_only = 0;
@@ -1095,7 +1166,10 @@ again:
 				trace_block_bio_remap(bdev_get_queue(bi->bi_bdev),
 						      bi, disk_devt(conf->mddev->gendisk),
 						      sh->dev[i].sector);
-			defer_bio_issue(conf, bi);
+			if (should_defer && op_is_write(op))
+				bio_list_add(&pending_bios, bi);
+			else
+				generic_make_request(bi);
 		}
 		if (rrdev) {
 			if (s->syncing || s->expanding || s->expanded
@@ -1140,7 +1214,10 @@ again:
 				trace_block_bio_remap(bdev_get_queue(rbi->bi_bdev),
 						      rbi, disk_devt(conf->mddev->gendisk),
 						      sh->dev[i].sector);
-			defer_bio_issue(conf, rbi);
+			if (should_defer && op_is_write(op))
+				bio_list_add(&pending_bios, rbi);
+			else
+				generic_make_request(rbi);
 		}
 		if (!rdev && !rrdev) {
 			if (op_is_write(op))
@@ -1158,6 +1235,9 @@ again:
 		if (sh != head_sh)
 			goto again;
 	}
+
+	if (should_defer && !bio_list_empty(&pending_bios))
+		defer_issue_bios(conf, head_sh->sector, &pending_bios);
 }
 
 static struct dma_async_tx_descriptor *
@@ -6678,6 +6758,7 @@ static void free_conf(struct r5conf *conf)
 			put_page(conf->disks[i].extra_page);
 	kfree(conf->disks);
 	kfree(conf->stripe_hashtbl);
+	kfree(conf->pending_data);
 	kfree(conf);
 }
 
@@ -6787,6 +6868,14 @@ static struct r5conf *setup_conf(struct mddev *mddev)
 	conf = kzalloc(sizeof(struct r5conf), GFP_KERNEL);
 	if (conf == NULL)
 		goto abort;
+	INIT_LIST_HEAD(&conf->free_list);
+	INIT_LIST_HEAD(&conf->pending_list);
+	conf->pending_data = kzalloc(sizeof(struct r5pending_data) *
+		PENDING_IO_MAX, GFP_KERNEL);
+	if (!conf->pending_data)
+		goto abort;
+	for (i = 0; i < PENDING_IO_MAX; i++)
+		list_add(&conf->pending_data[i].sibling, &conf->free_list);
 	/* Don't enable multi-threading by default*/
 	if (!alloc_thread_groups(conf, 0, &group_cnt, &worker_cnt_per_group,
 				 &new_group)) {
@@ -6811,7 +6900,6 @@ static struct r5conf *setup_conf(struct mddev *mddev)
 	atomic_set(&conf->active_stripes, 0);
 	atomic_set(&conf->preread_active_stripes, 0);
 	atomic_set(&conf->active_aligned_reads, 0);
-	bio_list_init(&conf->pending_bios);
 	spin_lock_init(&conf->pending_bios_lock);
 	conf->batch_bio_dispatch = true;
 	rdev_for_each(rdev, mddev) {
diff --git a/drivers/md/raid5.h b/drivers/md/raid5.h
index 6b9d2e839e6d..985cdc4850c2 100644
--- a/drivers/md/raid5.h
+++ b/drivers/md/raid5.h
@@ -572,6 +572,14 @@ enum r5_cache_state {
 				 */
 };
 
+#define PENDING_IO_MAX 512
+#define PENDING_IO_ONE_FLUSH 128
+struct r5pending_data {
+	struct list_head sibling;
+	sector_t sector; /* stripe sector */
+	struct bio_list bios;
+};
+
 struct r5conf {
 	struct hlist_head	*stripe_hashtbl;
 	/* only protect corresponding hash list and inactive_list */
@@ -689,9 +697,13 @@ struct r5conf {
 	int			worker_cnt_per_group;
 	struct r5l_log		*log;
 
-	struct bio_list		pending_bios;
 	spinlock_t		pending_bios_lock;
 	bool			batch_bio_dispatch;
+	struct r5pending_data	*pending_data;
+	struct list_head	free_list;
+	struct list_head	pending_list;
+	int			pending_data_cnt;
+	struct r5pending_data	*next_pending_data;
 };
 
 
-- 
cgit v1.2.3


From effe6ee7523aa50d0517bd7da141e112b44d89fa Mon Sep 17 00:00:00 2001
From: Song Liu <songliubraving@fb.com>
Date: Tue, 7 Mar 2017 16:49:17 -0800
Subject: md/r5cache: improve recovery with read ahead page pool

In r5cache recovery, the journal device is scanned page by page.
Currently, we use sync_page_io() to read journal device. This is
not efficient when we have to recovery many stripes from the journal.

To improve the speed of recovery, this patch introduces a read ahead
page pool (ra_pool) to recovery_ctx. With ra_pool, multiple consecutive
pages are read in one IO. Then the recovery code read the journal from
ra_pool.

With ra_pool, r5l_recovery_ctx has become much bigger. Therefore,
r5l_recovery_log() is refactored so r5l_recovery_ctx is not using
stack space.

Signed-off-by: Song Liu <songliubraving@fb.com>
Signed-off-by: Shaohua Li <shli@fb.com>
---
 drivers/md/raid5-cache.c | 221 +++++++++++++++++++++++++++++++++++++----------
 1 file changed, 175 insertions(+), 46 deletions(-)

(limited to 'drivers')

diff --git a/drivers/md/raid5-cache.c b/drivers/md/raid5-cache.c
index 738e180d3740..5c8640c86b90 100644
--- a/drivers/md/raid5-cache.c
+++ b/drivers/md/raid5-cache.c
@@ -30,6 +30,7 @@
  * underneath hardware sector size. only works with PAGE_SIZE == 4096
  */
 #define BLOCK_SECTORS (8)
+#define BLOCK_SECTOR_SHIFT (3)
 
 /*
  * log->max_free_space is min(1/4 disk size, 10G reclaimable space).
@@ -1552,6 +1553,8 @@ bool r5l_log_disk_error(struct r5conf *conf)
 	return ret;
 }
 
+#define R5L_RECOVERY_PAGE_POOL_SIZE 256
+
 struct r5l_recovery_ctx {
 	struct page *meta_page;		/* current meta */
 	sector_t meta_total_blocks;	/* total size of current meta and data */
@@ -1560,18 +1563,131 @@ struct r5l_recovery_ctx {
 	int data_parity_stripes;	/* number of data_parity stripes */
 	int data_only_stripes;		/* number of data_only stripes */
 	struct list_head cached_list;
+
+	/*
+	 * read ahead page pool (ra_pool)
+	 * in recovery, log is read sequentially. It is not efficient to
+	 * read every page with sync_page_io(). The read ahead page pool
+	 * reads multiple pages with one IO, so further log read can
+	 * just copy data from the pool.
+	 */
+	struct page *ra_pool[R5L_RECOVERY_PAGE_POOL_SIZE];
+	sector_t pool_offset;	/* offset of first page in the pool */
+	int total_pages;	/* total allocated pages */
+	int valid_pages;	/* pages with valid data */
+	struct bio *ra_bio;	/* bio to do the read ahead */
 };
 
+static int r5l_recovery_allocate_ra_pool(struct r5l_log *log,
+					    struct r5l_recovery_ctx *ctx)
+{
+	struct page *page;
+
+	ctx->ra_bio = bio_alloc_bioset(GFP_KERNEL, BIO_MAX_PAGES, log->bs);
+	if (!ctx->ra_bio)
+		return -ENOMEM;
+
+	ctx->valid_pages = 0;
+	ctx->total_pages = 0;
+	while (ctx->total_pages < R5L_RECOVERY_PAGE_POOL_SIZE) {
+		page = alloc_page(GFP_KERNEL);
+
+		if (!page)
+			break;
+		ctx->ra_pool[ctx->total_pages] = page;
+		ctx->total_pages += 1;
+	}
+
+	if (ctx->total_pages == 0) {
+		bio_put(ctx->ra_bio);
+		return -ENOMEM;
+	}
+
+	ctx->pool_offset = 0;
+	return 0;
+}
+
+static void r5l_recovery_free_ra_pool(struct r5l_log *log,
+					struct r5l_recovery_ctx *ctx)
+{
+	int i;
+
+	for (i = 0; i < ctx->total_pages; ++i)
+		put_page(ctx->ra_pool[i]);
+	bio_put(ctx->ra_bio);
+}
+
+/*
+ * fetch ctx->valid_pages pages from offset
+ * In normal cases, ctx->valid_pages == ctx->total_pages after the call.
+ * However, if the offset is close to the end of the journal device,
+ * ctx->valid_pages could be smaller than ctx->total_pages
+ */
+static int r5l_recovery_fetch_ra_pool(struct r5l_log *log,
+				      struct r5l_recovery_ctx *ctx,
+				      sector_t offset)
+{
+	bio_reset(ctx->ra_bio);
+	ctx->ra_bio->bi_bdev = log->rdev->bdev;
+	bio_set_op_attrs(ctx->ra_bio, REQ_OP_READ, 0);
+	ctx->ra_bio->bi_iter.bi_sector = log->rdev->data_offset + offset;
+
+	ctx->valid_pages = 0;
+	ctx->pool_offset = offset;
+
+	while (ctx->valid_pages < ctx->total_pages) {
+		bio_add_page(ctx->ra_bio,
+			     ctx->ra_pool[ctx->valid_pages], PAGE_SIZE, 0);
+		ctx->valid_pages += 1;
+
+		offset = r5l_ring_add(log, offset, BLOCK_SECTORS);
+
+		if (offset == 0)  /* reached end of the device */
+			break;
+	}
+
+	return submit_bio_wait(ctx->ra_bio);
+}
+
+/*
+ * try read a page from the read ahead page pool, if the page is not in the
+ * pool, call r5l_recovery_fetch_ra_pool
+ */
+static int r5l_recovery_read_page(struct r5l_log *log,
+				  struct r5l_recovery_ctx *ctx,
+				  struct page *page,
+				  sector_t offset)
+{
+	int ret;
+
+	if (offset < ctx->pool_offset ||
+	    offset >= ctx->pool_offset + ctx->valid_pages * BLOCK_SECTORS) {
+		ret = r5l_recovery_fetch_ra_pool(log, ctx, offset);
+		if (ret)
+			return ret;
+	}
+
+	BUG_ON(offset < ctx->pool_offset ||
+	       offset >= ctx->pool_offset + ctx->valid_pages * BLOCK_SECTORS);
+
+	memcpy(page_address(page),
+	       page_address(ctx->ra_pool[(offset - ctx->pool_offset) >>
+					 BLOCK_SECTOR_SHIFT]),
+	       PAGE_SIZE);
+	return 0;
+}
+
 static int r5l_recovery_read_meta_block(struct r5l_log *log,
 					struct r5l_recovery_ctx *ctx)
 {
 	struct page *page = ctx->meta_page;
 	struct r5l_meta_block *mb;
 	u32 crc, stored_crc;
+	int ret;
 
-	if (!sync_page_io(log->rdev, ctx->pos, PAGE_SIZE, page, REQ_OP_READ, 0,
-			  false))
-		return -EIO;
+	ret = r5l_recovery_read_page(log, ctx, page, ctx->pos);
+	if (ret != 0)
+		return ret;
 
 	mb = page_address(page);
 	stored_crc = le32_to_cpu(mb->checksum);
@@ -1653,8 +1769,7 @@ static void r5l_recovery_load_data(struct r5l_log *log,
 	raid5_compute_sector(conf,
 			     le64_to_cpu(payload->location), 0,
 			     &dd_idx, sh);
-	sync_page_io(log->rdev, log_offset, PAGE_SIZE,
-		     sh->dev[dd_idx].page, REQ_OP_READ, 0, false);
+	r5l_recovery_read_page(log, ctx, sh->dev[dd_idx].page, log_offset);
 	sh->dev[dd_idx].log_checksum =
 		le32_to_cpu(payload->checksum[0]);
 	ctx->meta_total_blocks += BLOCK_SECTORS;
@@ -1673,17 +1788,15 @@ static void r5l_recovery_load_parity(struct r5l_log *log,
 	struct r5conf *conf = mddev->private;
 
 	ctx->meta_total_blocks += BLOCK_SECTORS * conf->max_degraded;
-	sync_page_io(log->rdev, log_offset, PAGE_SIZE,
-		     sh->dev[sh->pd_idx].page, REQ_OP_READ, 0, false);
+	r5l_recovery_read_page(log, ctx, sh->dev[sh->pd_idx].page, log_offset);
 	sh->dev[sh->pd_idx].log_checksum =
 		le32_to_cpu(payload->checksum[0]);
 	set_bit(R5_Wantwrite, &sh->dev[sh->pd_idx].flags);
 
 	if (sh->qd_idx >= 0) {
-		sync_page_io(log->rdev,
-			     r5l_ring_add(log, log_offset, BLOCK_SECTORS),
-			     PAGE_SIZE, sh->dev[sh->qd_idx].page,
-			     REQ_OP_READ, 0, false);
+		r5l_recovery_read_page(
+			log, ctx, sh->dev[sh->qd_idx].page,
+			r5l_ring_add(log, log_offset, BLOCK_SECTORS));
 		sh->dev[sh->qd_idx].log_checksum =
 			le32_to_cpu(payload->checksum[1]);
 		set_bit(R5_Wantwrite, &sh->dev[sh->qd_idx].flags);
@@ -1814,14 +1927,15 @@ r5c_recovery_replay_stripes(struct list_head *cached_stripe_list,
 
 /* if matches return 0; otherwise return -EINVAL */
 static int
-r5l_recovery_verify_data_checksum(struct r5l_log *log, struct page *page,
+r5l_recovery_verify_data_checksum(struct r5l_log *log,
+				  struct r5l_recovery_ctx *ctx,
+				  struct page *page,
 				  sector_t log_offset, __le32 log_checksum)
 {
 	void *addr;
 	u32 checksum;
 
-	sync_page_io(log->rdev, log_offset, PAGE_SIZE,
-		     page, REQ_OP_READ, 0, false);
+	r5l_recovery_read_page(log, ctx, page, log_offset);
 	addr = kmap_atomic(page);
 	checksum = crc32c_le(log->uuid_checksum, addr, PAGE_SIZE);
 	kunmap_atomic(addr);
@@ -1853,17 +1967,17 @@ r5l_recovery_verify_data_checksum_for_mb(struct r5l_log *log,
 
 		if (payload->header.type == R5LOG_PAYLOAD_DATA) {
 			if (r5l_recovery_verify_data_checksum(
-				    log, page, log_offset,
+				    log, ctx, page, log_offset,
 				    payload->checksum[0]) < 0)
 				goto mismatch;
 		} else if (payload->header.type == R5LOG_PAYLOAD_PARITY) {
 			if (r5l_recovery_verify_data_checksum(
-				    log, page, log_offset,
+				    log, ctx, page, log_offset,
 				    payload->checksum[0]) < 0)
 				goto mismatch;
 			if (conf->max_degraded == 2 && /* q for RAID 6 */
 			    r5l_recovery_verify_data_checksum(
-				    log, page,
+				    log, ctx, page,
 				    r5l_ring_add(log, log_offset,
 						 BLOCK_SECTORS),
 				    payload->checksum[1]) < 0)
@@ -2241,55 +2355,70 @@ static void r5c_recovery_flush_data_only_stripes(struct r5l_log *log,
 static int r5l_recovery_log(struct r5l_log *log)
 {
 	struct mddev *mddev = log->rdev->mddev;
-	struct r5l_recovery_ctx ctx;
+	struct r5l_recovery_ctx *ctx;
 	int ret;
 	sector_t pos;
 
-	ctx.pos = log->last_checkpoint;
-	ctx.seq = log->last_cp_seq;
-	ctx.meta_page = alloc_page(GFP_KERNEL);
-	ctx.data_only_stripes = 0;
-	ctx.data_parity_stripes = 0;
-	INIT_LIST_HEAD(&ctx.cached_list);
-
-	if (!ctx.meta_page)
+	ctx = kzalloc(sizeof(*ctx), GFP_KERNEL);
+	if (!ctx)
 		return -ENOMEM;
 
-	ret = r5c_recovery_flush_log(log, &ctx);
-	__free_page(ctx.meta_page);
+	ctx->pos = log->last_checkpoint;
+	ctx->seq = log->last_cp_seq;
+	INIT_LIST_HEAD(&ctx->cached_list);
+	ctx->meta_page = alloc_page(GFP_KERNEL);
 
-	if (ret)
-		return ret;
+	if (!ctx->meta_page) {
+		ret =  -ENOMEM;
+		goto meta_page;
+	}
+
+	if (r5l_recovery_allocate_ra_pool(log, ctx) != 0) {
+		ret = -ENOMEM;
+		goto ra_pool;
+	}
 
-	pos = ctx.pos;
-	ctx.seq += 10000;
+	ret = r5c_recovery_flush_log(log, ctx);
 
+	if (ret)
+		goto error;
 
-	if ((ctx.data_only_stripes == 0) && (ctx.data_parity_stripes == 0))
+	pos = ctx->pos;
+	ctx->seq += 10000;
+
+	if ((ctx->data_only_stripes == 0) && (ctx->data_parity_stripes == 0))
 		pr_debug("md/raid:%s: starting from clean shutdown\n",
 			 mdname(mddev));
 	else
 		pr_debug("md/raid:%s: recovering %d data-only stripes and %d data-parity stripes\n",
-			 mdname(mddev), ctx.data_only_stripes,
-			 ctx.data_parity_stripes);
-
-	if (ctx.data_only_stripes == 0) {
-		log->next_checkpoint = ctx.pos;
-		r5l_log_write_empty_meta_block(log, ctx.pos, ctx.seq++);
-		ctx.pos = r5l_ring_add(log, ctx.pos, BLOCK_SECTORS);
-	} else if (r5c_recovery_rewrite_data_only_stripes(log, &ctx)) {
+			 mdname(mddev), ctx->data_only_stripes,
+			 ctx->data_parity_stripes);
+
+	if (ctx->data_only_stripes == 0) {
+		log->next_checkpoint = ctx->pos;
+		r5l_log_write_empty_meta_block(log, ctx->pos, ctx->seq++);
+		ctx->pos = r5l_ring_add(log, ctx->pos, BLOCK_SECTORS);
+	} else if (r5c_recovery_rewrite_data_only_stripes(log, ctx)) {
 		pr_err("md/raid:%s: failed to rewrite stripes to journal\n",
 		       mdname(mddev));
-		return -EIO;
+		ret =  -EIO;
+		goto error;
 	}
 
-	log->log_start = ctx.pos;
-	log->seq = ctx.seq;
+	log->log_start = ctx->pos;
+	log->seq = ctx->seq;
 	log->last_checkpoint = pos;
 	r5l_write_super(log, pos);
 
-	r5c_recovery_flush_data_only_stripes(log, &ctx);
-	return 0;
+	r5c_recovery_flush_data_only_stripes(log, ctx);
+	ret = 0;
+error:
+	r5l_recovery_free_ra_pool(log, ctx);
+ra_pool:
+	__free_page(ctx->meta_page);
+meta_page:
+	kfree(ctx);
+	return ret;
 }
 
 static void r5l_write_super(struct r5l_log *log, sector_t cp)
-- 
cgit v1.2.3


From ea0213e0c7cc1c1b52badf27bd7db4f50a67baaa Mon Sep 17 00:00:00 2001
From: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
Date: Thu, 9 Mar 2017 09:59:57 +0100
Subject: md: superblock changes for PPL

Include information about PPL location and size into mdp_superblock_1
and copy it to/from rdev. Because PPL is mutually exclusive with bitmap,
put it in place of 'bitmap_offset'. Add a new flag MD_FEATURE_PPL for
'feature_map', analogically to MD_FEATURE_BITMAP_OFFSET. Add MD_HAS_PPL
to mddev->flags to indicate that PPL is enabled on an array.

Signed-off-by: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
Signed-off-by: Shaohua Li <shli@fb.com>
---
 drivers/md/md.c    | 19 +++++++++++++++++++
 drivers/md/md.h    |  8 ++++++++
 drivers/md/raid0.c |  3 ++-
 drivers/md/raid1.c |  3 ++-
 4 files changed, 31 insertions(+), 2 deletions(-)

(limited to 'drivers')

diff --git a/drivers/md/md.c b/drivers/md/md.c
index 72ef3f18ac9a..d57045996d35 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -1507,6 +1507,12 @@ static int super_1_load(struct md_rdev *rdev, struct md_rdev *refdev, int minor_
 	} else if (sb->bblog_offset != 0)
 		rdev->badblocks.shift = 0;
 
+	if (le32_to_cpu(sb->feature_map) & MD_FEATURE_PPL) {
+		rdev->ppl.offset = (__s16)le16_to_cpu(sb->ppl.offset);
+		rdev->ppl.size = le16_to_cpu(sb->ppl.size);
+		rdev->ppl.sector = rdev->sb_start + rdev->ppl.offset;
+	}
+
 	if (!refdev) {
 		ret = 1;
 	} else {
@@ -1619,6 +1625,13 @@ static int super_1_validate(struct mddev *mddev, struct md_rdev *rdev)
 
 		if (le32_to_cpu(sb->feature_map) & MD_FEATURE_JOURNAL)
 			set_bit(MD_HAS_JOURNAL, &mddev->flags);
+
+		if (le32_to_cpu(sb->feature_map) & MD_FEATURE_PPL) {
+			if (le32_to_cpu(sb->feature_map) &
+			    (MD_FEATURE_BITMAP_OFFSET | MD_FEATURE_JOURNAL))
+				return -EINVAL;
+			set_bit(MD_HAS_PPL, &mddev->flags);
+		}
 	} else if (mddev->pers == NULL) {
 		/* Insist of good event counter while assembling, except for
 		 * spares (which don't need an event count) */
@@ -1832,6 +1845,12 @@ retry:
 	if (test_bit(MD_HAS_JOURNAL, &mddev->flags))
 		sb->feature_map |= cpu_to_le32(MD_FEATURE_JOURNAL);
 
+	if (test_bit(MD_HAS_PPL, &mddev->flags)) {
+		sb->feature_map |= cpu_to_le32(MD_FEATURE_PPL);
+		sb->ppl.offset = cpu_to_le16(rdev->ppl.offset);
+		sb->ppl.size = cpu_to_le16(rdev->ppl.size);
+	}
+
 	rdev_for_each(rdev2, mddev) {
 		i = rdev2->desc_nr;
 		if (test_bit(Faulty, &rdev2->flags))
diff --git a/drivers/md/md.h b/drivers/md/md.h
index 1c00160b09f9..a7b2f16452c4 100644
--- a/drivers/md/md.h
+++ b/drivers/md/md.h
@@ -122,6 +122,13 @@ struct md_rdev {
 					   * sysfs entry */
 
 	struct badblocks badblocks;
+
+	struct {
+		short offset;	/* Offset from superblock to start of PPL.
+				 * Not used by external metadata. */
+		unsigned int size;	/* Size in sectors of the PPL space */
+		sector_t sector;	/* First sector of the PPL space */
+	} ppl;
 };
 enum flag_bits {
 	Faulty,			/* device is known to have a fault */
@@ -226,6 +233,7 @@ enum mddev_flags {
 				 * supported as calls to md_error() will
 				 * never cause the array to become failed.
 				 */
+	MD_HAS_PPL,		/* The raid array has PPL feature set */
 };
 
 enum mddev_sb_flags {
diff --git a/drivers/md/raid0.c b/drivers/md/raid0.c
index 93347ca7c7a6..56f70c3ad37c 100644
--- a/drivers/md/raid0.c
+++ b/drivers/md/raid0.c
@@ -29,7 +29,8 @@
 #define UNSUPPORTED_MDDEV_FLAGS		\
 	((1L << MD_HAS_JOURNAL) |	\
 	 (1L << MD_JOURNAL_CLEAN) |	\
-	 (1L << MD_FAILFAST_SUPPORTED))
+	 (1L << MD_FAILFAST_SUPPORTED) |\
+	 (1L << MD_HAS_PPL))
 
 static int raid0_congested(struct mddev *mddev, int bits)
 {
diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c
index a34f58772022..730e57259af9 100644
--- a/drivers/md/raid1.c
+++ b/drivers/md/raid1.c
@@ -47,7 +47,8 @@
 
 #define UNSUPPORTED_MDDEV_FLAGS		\
 	((1L << MD_HAS_JOURNAL) |	\
-	 (1L << MD_JOURNAL_CLEAN))
+	 (1L << MD_JOURNAL_CLEAN) |	\
+	 (1L << MD_HAS_PPL))
 
 /*
  * Number of guaranteed r1bios in case of extreme VM load:
-- 
cgit v1.2.3


From ff875738edd44e3bc892d378deacc50bccc9d70c Mon Sep 17 00:00:00 2001
From: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
Date: Thu, 9 Mar 2017 09:59:58 +0100
Subject: raid5: separate header for log functions

Move raid5-cache declarations from raid5.h to raid5-log.h, add inline
wrappers for functions which will be shared with ppl and use them in
raid5 core instead of direct calls to raid5-cache.

Remove unused parameter from r5c_cache_data(), move two duplicated
pr_debug() calls to r5l_init_log().

Signed-off-by: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
Signed-off-by: Shaohua Li <shli@fb.com>
---
 drivers/md/raid5-cache.c | 17 +++++++---
 drivers/md/raid5-log.h   | 81 ++++++++++++++++++++++++++++++++++++++++++++++++
 drivers/md/raid5.c       | 48 ++++++++--------------------
 drivers/md/raid5.h       | 30 ------------------
 4 files changed, 107 insertions(+), 69 deletions(-)
 create mode 100644 drivers/md/raid5-log.h

(limited to 'drivers')

diff --git a/drivers/md/raid5-cache.c b/drivers/md/raid5-cache.c
index 5c8640c86b90..5f82dabdda6f 100644
--- a/drivers/md/raid5-cache.c
+++ b/drivers/md/raid5-cache.c
@@ -344,6 +344,8 @@ void r5c_handle_cached_data_endio(struct r5conf *conf,
 	}
 }
 
+void r5l_wake_reclaim(struct r5l_log *log, sector_t space);
+
 /* Check whether we should flush some stripes to free up stripe cache */
 void r5c_check_stripe_cache_usage(struct r5conf *conf)
 {
@@ -2749,9 +2751,7 @@ void r5c_finish_stripe_write_out(struct r5conf *conf,
 	}
 }
 
-int
-r5c_cache_data(struct r5l_log *log, struct stripe_head *sh,
-	       struct stripe_head_state *s)
+int r5c_cache_data(struct r5l_log *log, struct stripe_head *sh)
 {
 	struct r5conf *conf = sh->raid_conf;
 	int pages = 0;
@@ -2914,6 +2914,10 @@ int r5l_init_log(struct r5conf *conf, struct md_rdev *rdev)
 {
 	struct request_queue *q = bdev_get_queue(rdev->bdev);
 	struct r5l_log *log;
+	char b[BDEVNAME_SIZE];
+
+	pr_debug("md/raid:%s: using device %s as journal\n",
+		 mdname(conf->mddev), bdevname(rdev->bdev, b));
 
 	if (PAGE_SIZE != 4096)
 		return -EINVAL;
@@ -3016,8 +3020,13 @@ io_kc:
 	return -EINVAL;
 }
 
-void r5l_exit_log(struct r5l_log *log)
+void r5l_exit_log(struct r5conf *conf)
 {
+	struct r5l_log *log = conf->log;
+
+	conf->log = NULL;
+	synchronize_rcu();
+
 	flush_work(&log->disable_writeback_work);
 	md_unregister_thread(&log->reclaim_thread);
 	mempool_destroy(log->meta_pool);
diff --git a/drivers/md/raid5-log.h b/drivers/md/raid5-log.h
new file mode 100644
index 000000000000..2da4bd3bbd79
--- /dev/null
+++ b/drivers/md/raid5-log.h
@@ -0,0 +1,81 @@
+#ifndef _RAID5_LOG_H
+#define _RAID5_LOG_H
+
+extern int r5l_init_log(struct r5conf *conf, struct md_rdev *rdev);
+extern void r5l_exit_log(struct r5conf *conf);
+extern int r5l_write_stripe(struct r5l_log *log, struct stripe_head *head_sh);
+extern void r5l_write_stripe_run(struct r5l_log *log);
+extern void r5l_flush_stripe_to_raid(struct r5l_log *log);
+extern void r5l_stripe_write_finished(struct stripe_head *sh);
+extern int r5l_handle_flush_request(struct r5l_log *log, struct bio *bio);
+extern void r5l_quiesce(struct r5l_log *log, int state);
+extern bool r5l_log_disk_error(struct r5conf *conf);
+extern bool r5c_is_writeback(struct r5l_log *log);
+extern int
+r5c_try_caching_write(struct r5conf *conf, struct stripe_head *sh,
+		      struct stripe_head_state *s, int disks);
+extern void
+r5c_finish_stripe_write_out(struct r5conf *conf, struct stripe_head *sh,
+			    struct stripe_head_state *s);
+extern void r5c_release_extra_page(struct stripe_head *sh);
+extern void r5c_use_extra_page(struct stripe_head *sh);
+extern void r5l_wake_reclaim(struct r5l_log *log, sector_t space);
+extern void r5c_handle_cached_data_endio(struct r5conf *conf,
+	struct stripe_head *sh, int disks, struct bio_list *return_bi);
+extern int r5c_cache_data(struct r5l_log *log, struct stripe_head *sh);
+extern void r5c_make_stripe_write_out(struct stripe_head *sh);
+extern void r5c_flush_cache(struct r5conf *conf, int num);
+extern void r5c_check_stripe_cache_usage(struct r5conf *conf);
+extern void r5c_check_cached_full_stripe(struct r5conf *conf);
+extern struct md_sysfs_entry r5c_journal_mode;
+extern void r5c_update_on_rdev_error(struct mddev *mddev);
+extern bool r5c_big_stripe_cached(struct r5conf *conf, sector_t sect);
+
+static inline int log_stripe(struct stripe_head *sh, struct stripe_head_state *s)
+{
+	struct r5conf *conf = sh->raid_conf;
+
+	if (conf->log) {
+		if (!test_bit(STRIPE_R5C_CACHING, &sh->state)) {
+			/* writing out phase */
+			if (s->waiting_extra_page)
+				return 0;
+			return r5l_write_stripe(conf->log, sh);
+		} else if (test_bit(STRIPE_LOG_TRAPPED, &sh->state)) {
+			/* caching phase */
+			return r5c_cache_data(conf->log, sh);
+		}
+	}
+
+	return -EAGAIN;
+}
+
+static inline void log_stripe_write_finished(struct stripe_head *sh)
+{
+	struct r5conf *conf = sh->raid_conf;
+
+	if (conf->log)
+		r5l_stripe_write_finished(sh);
+}
+
+static inline void log_write_stripe_run(struct r5conf *conf)
+{
+	if (conf->log)
+		r5l_write_stripe_run(conf->log);
+}
+
+static inline void log_exit(struct r5conf *conf)
+{
+	if (conf->log)
+		r5l_exit_log(conf);
+}
+
+static inline int log_init(struct r5conf *conf, struct md_rdev *journal_dev)
+{
+	if (journal_dev)
+		return r5l_init_log(conf, journal_dev);
+
+	return 0;
+}
+
+#endif
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index 013398ce2080..f575f40d2acb 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -64,6 +64,7 @@
 #include "raid5.h"
 #include "raid0.h"
 #include "bitmap.h"
+#include "raid5-log.h"
 
 #define UNSUPPORTED_MDDEV_FLAGS	(1L << MD_FAILFAST_SUPPORTED)
 
@@ -997,18 +998,8 @@ static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s)
 
 	might_sleep();
 
-	if (!test_bit(STRIPE_R5C_CACHING, &sh->state)) {
-		/* writing out phase */
-		if (s->waiting_extra_page)
-			return;
-		if (r5l_write_stripe(conf->log, sh) == 0)
-			return;
-	} else {  /* caching phase */
-		if (test_bit(STRIPE_LOG_TRAPPED, &sh->state)) {
-			r5c_cache_data(conf->log, sh, s);
-			return;
-		}
-	}
+	if (log_stripe(sh, s) == 0)
+		return;
 
 	should_defer = conf->batch_bio_dispatch && conf->group_cnt;
 
@@ -3345,7 +3336,7 @@ handle_failed_stripe(struct r5conf *conf, struct stripe_head *sh,
 		if (bi)
 			bitmap_end = 1;
 
-		r5l_stripe_write_finished(sh);
+		log_stripe_write_finished(sh);
 
 		if (test_and_clear_bit(R5_Overlap, &sh->dev[i].flags))
 			wake_up(&conf->wait_for_overlap);
@@ -3764,7 +3755,7 @@ returnbi:
 				discard_pending = 1;
 		}
 
-	r5l_stripe_write_finished(sh);
+	log_stripe_write_finished(sh);
 
 	if (!discard_pending &&
 	    test_bit(R5_Discard, &sh->dev[sh->pd_idx].flags)) {
@@ -4754,7 +4745,7 @@ static void handle_stripe(struct stripe_head *sh)
 
 	if (s.just_cached)
 		r5c_handle_cached_data_endio(conf, sh, disks, &s.return_bi);
-	r5l_stripe_write_finished(sh);
+	log_stripe_write_finished(sh);
 
 	/* Now we