1 files changed, 709 insertions, 96 deletions

diff --git a/kernel/fs/nfs/flexfilelayout/flexfilelayout.c b/kernel/fs/nfs/flexfilelayout/flexfilelayout.c
index fecd9201d..2a2e2d8dd 100644
--- a/kernel/fs/nfs/flexfilelayout/flexfilelayout.c
+++ b/kernel/fs/nfs/flexfilelayout/flexfilelayout.c
@@ -20,6 +20,7 @@
 #include "../nfs4trace.h"
 #include "../iostat.h"
 #include "../nfs.h"
+#include "../nfs42.h"
 
 #define NFSDBG_FACILITY         NFSDBG_PNFS_LD
 
@@ -33,6 +34,7 @@ ff_layout_alloc_layout_hdr(struct inode *inode, gfp_t gfp_flags)
 	ffl = kzalloc(sizeof(*ffl), gfp_flags);
 	if (ffl) {
 		INIT_LIST_HEAD(&ffl->error_list);
+		INIT_LIST_HEAD(&ffl->mirrors);
 		return &ffl->generic_hdr;
 	} else
 		return NULL;
@@ -134,6 +136,95 @@ decode_name(struct xdr_stream *xdr, u32 *id)
 	return 0;
 }
 
+static bool ff_mirror_match_fh(const struct nfs4_ff_layout_mirror *m1,
+		const struct nfs4_ff_layout_mirror *m2)
+{
+	int i, j;
+
+	if (m1->fh_versions_cnt != m2->fh_versions_cnt)
+		return false;
+	for (i = 0; i < m1->fh_versions_cnt; i++) {
+		bool found_fh = false;
+		for (j = 0; j < m2->fh_versions_cnt; j++) {
+			if (nfs_compare_fh(&m1->fh_versions[i],
+					&m2->fh_versions[j]) == 0) {
+				found_fh = true;
+				break;
+			}
+		}
+		if (!found_fh)
+			return false;
+	}
+	return true;
+}
+
+static struct nfs4_ff_layout_mirror *
+ff_layout_add_mirror(struct pnfs_layout_hdr *lo,
+		struct nfs4_ff_layout_mirror *mirror)
+{
+	struct nfs4_flexfile_layout *ff_layout = FF_LAYOUT_FROM_HDR(lo);
+	struct nfs4_ff_layout_mirror *pos;
+	struct inode *inode = lo->plh_inode;
+
+	spin_lock(&inode->i_lock);
+	list_for_each_entry(pos, &ff_layout->mirrors, mirrors) {
+		if (mirror->mirror_ds != pos->mirror_ds)
+			continue;
+		if (!ff_mirror_match_fh(mirror, pos))
+			continue;
+		if (atomic_inc_not_zero(&pos->ref)) {
+			spin_unlock(&inode->i_lock);
+			return pos;
+		}
+	}
+	list_add(&mirror->mirrors, &ff_layout->mirrors);
+	mirror->layout = lo;
+	spin_unlock(&inode->i_lock);
+	return mirror;
+}
+
+static void
+ff_layout_remove_mirror(struct nfs4_ff_layout_mirror *mirror)
+{
+	struct inode *inode;
+	if (mirror->layout == NULL)
+		return;
+	inode = mirror->layout->plh_inode;
+	spin_lock(&inode->i_lock);
+	list_del(&mirror->mirrors);
+	spin_unlock(&inode->i_lock);
+	mirror->layout = NULL;
+}
+
+static struct nfs4_ff_layout_mirror *ff_layout_alloc_mirror(gfp_t gfp_flags)
+{
+	struct nfs4_ff_layout_mirror *mirror;
+
+	mirror = kzalloc(sizeof(*mirror), gfp_flags);
+	if (mirror != NULL) {
+		spin_lock_init(&mirror->lock);
+		atomic_set(&mirror->ref, 1);
+		INIT_LIST_HEAD(&mirror->mirrors);
+	}
+	return mirror;
+}
+
+static void ff_layout_free_mirror(struct nfs4_ff_layout_mirror *mirror)
+{
+	ff_layout_remove_mirror(mirror);
+	kfree(mirror->fh_versions);
+	if (mirror->cred)
+		put_rpccred(mirror->cred);
+	nfs4_ff_layout_put_deviceid(mirror->mirror_ds);
+	kfree(mirror);
+}
+
+static void ff_layout_put_mirror(struct nfs4_ff_layout_mirror *mirror)
+{
+	if (mirror != NULL && atomic_dec_and_test(&mirror->ref))
+		ff_layout_free_mirror(mirror);
+}
+
 static void ff_layout_free_mirror_array(struct nfs4_ff_layout_segment *fls)
 {
 	int i;
@@ -143,11 +234,7 @@ static void ff_layout_free_mirror_array(struct nfs4_ff_layout_segment *fls)
 			/* normally mirror_ds is freed in
 			 * .free_deviceid_node but we still do it here
 			 * for .alloc_lseg error path */
-			if (fls->mirror_array[i]) {
-				kfree(fls->mirror_array[i]->fh_versions);
-				nfs4_ff_layout_put_deviceid(fls->mirror_array[i]->mirror_ds);
-				kfree(fls->mirror_array[i]);
-			}
+			ff_layout_put_mirror(fls->mirror_array[i]);
 		}
 		kfree(fls->mirror_array);
 		fls->mirror_array = NULL;
@@ -180,19 +267,88 @@ static void _ff_layout_free_lseg(struct nfs4_ff_layout_segment *fls)
 	}
 }
 
+static bool
+ff_lseg_range_is_after(const struct pnfs_layout_range *l1,
+		const struct pnfs_layout_range *l2)
+{
+	u64 end1, end2;
+
+	if (l1->iomode != l2->iomode)
+		return l1->iomode != IOMODE_READ;
+	end1 = pnfs_calc_offset_end(l1->offset, l1->length);
+	end2 = pnfs_calc_offset_end(l2->offset, l2->length);
+	if (end1 < l2->offset)
+		return false;
+	if (end2 < l1->offset)
+		return true;
+	return l2->offset <= l1->offset;
+}
+
+static bool
+ff_lseg_merge(struct pnfs_layout_segment *new,
+		struct pnfs_layout_segment *old)
+{
+	u64 new_end, old_end;
+
+	if (new->pls_range.iomode != old->pls_range.iomode)
+		return false;
+	old_end = pnfs_calc_offset_end(old->pls_range.offset,
+			old->pls_range.length);
+	if (old_end < new->pls_range.offset)
+		return false;
+	new_end = pnfs_calc_offset_end(new->pls_range.offset,
+			new->pls_range.length);
+	if (new_end < old->pls_range.offset)
+		return false;
+
+	/* Mergeable: copy info from 'old' to 'new' */
+	if (new_end < old_end)
+		new_end = old_end;
+	if (new->pls_range.offset < old->pls_range.offset)
+		new->pls_range.offset = old->pls_range.offset;
+	new->pls_range.length = pnfs_calc_offset_length(new->pls_range.offset,
+			new_end);
+	if (test_bit(NFS_LSEG_ROC, &old->pls_flags))
+		set_bit(NFS_LSEG_ROC, &new->pls_flags);
+	if (test_bit(NFS_LSEG_LAYOUTRETURN, &old->pls_flags))
+		set_bit(NFS_LSEG_LAYOUTRETURN, &new->pls_flags);
+	return true;
+}
+
+static void
+ff_layout_add_lseg(struct pnfs_layout_hdr *lo,
+		struct pnfs_layout_segment *lseg,
+		struct list_head *free_me)
+{
+	pnfs_generic_layout_insert_lseg(lo, lseg,
+			ff_lseg_range_is_after,
+			ff_lseg_merge,
+			free_me);
+}
+
 static void ff_layout_sort_mirrors(struct nfs4_ff_layout_segment *fls)
 {
-	struct nfs4_ff_layout_mirror *tmp;
 	int i, j;
 
 	for (i = 0; i < fls->mirror_array_cnt - 1; i++) {
 		for (j = i + 1; j < fls->mirror_array_cnt; j++)
 			if (fls->mirror_array[i]->efficiency <
-			    fls->mirror_array[j]->efficiency) {
-				tmp = fls->mirror_array[i];
-				fls->mirror_array[i] = fls->mirror_array[j];
-				fls->mirror_array[j] = tmp;
-			}
+			    fls->mirror_array[j]->efficiency)
+				swap(fls->mirror_array[i],
+				     fls->mirror_array[j]);
+	}
+}
+
+static void ff_layout_mark_devices_valid(struct nfs4_ff_layout_segment *fls)
+{
+	struct nfs4_deviceid_node *node;
+	int i;
+
+	if (!(fls->flags & FF_FLAGS_NO_IO_THRU_MDS))
+		return;
+	for (i = 0; i < fls->mirror_array_cnt; i++) {
+		node = &fls->mirror_array[i]->mirror_ds->id_node;
+		clear_bit(NFS_DEVICEID_UNAVAILABLE, &node->flags);
 	}
 }
 
@@ -248,6 +404,7 @@ ff_layout_alloc_lseg(struct pnfs_layout_hdr *lh,
 		goto out_err_free;
 
 	for (i = 0; i < fls->mirror_array_cnt; i++) {
+		struct nfs4_ff_layout_mirror *mirror;
 		struct nfs4_deviceid devid;
 		struct nfs4_deviceid_node *idnode;
 		u32 ds_count;
@@ -264,15 +421,12 @@ ff_layout_alloc_lseg(struct pnfs_layout_hdr *lh,
 		if (ds_count != 1)
 			goto out_err_free;
 
-		fls->mirror_array[i] =
-			kzalloc(sizeof(struct nfs4_ff_layout_mirror),
-				gfp_flags);
+		fls->mirror_array[i] = ff_layout_alloc_mirror(gfp_flags);
 		if (fls->mirror_array[i] == NULL) {
 			rc = -ENOMEM;
 			goto out_err_free;
 		}
 
-		spin_lock_init(&fls->mirror_array[i]->lock);
 		fls->mirror_array[i]->ds_count = ds_count;
 
 		/* deviceid */
@@ -339,15 +493,26 @@ ff_layout_alloc_lseg(struct pnfs_layout_hdr *lh,
 		if (rc)
 			goto out_err_free;
 
+		mirror = ff_layout_add_mirror(lh, fls->mirror_array[i]);
+		if (mirror != fls->mirror_array[i]) {
+			ff_layout_free_mirror(fls->mirror_array[i]);
+			fls->mirror_array[i] = mirror;
+		}
+
 		dprintk("%s: uid %d gid %d\n", __func__,
 			fls->mirror_array[i]->uid,
 			fls->mirror_array[i]->gid);
 	}
 
+	p = xdr_inline_decode(&stream, 4);
+	if (p)
+		fls->flags = be32_to_cpup(p);
+
 	ff_layout_sort_mirrors(fls);
 	rc = ff_layout_check_layout(lgr);
 	if (rc)
 		goto out_err_free;
+	ff_layout_mark_devices_valid(fls);
 
 	ret = &fls->generic_hdr;
 	dprintk("<-- %s (success)\n", __func__);
@@ -376,21 +541,9 @@ static void
 ff_layout_free_lseg(struct pnfs_layout_segment *lseg)
 {
 	struct nfs4_ff_layout_segment *fls = FF_LAYOUT_LSEG(lseg);
-	int i;
 
 	dprintk("--> %s\n", __func__);
 
-	for (i = 0; i < fls->mirror_array_cnt; i++) {
-		if (fls->mirror_array[i]) {
-			nfs4_ff_layout_put_deviceid(fls->mirror_array[i]->mirror_ds);
-			fls->mirror_array[i]->mirror_ds = NULL;
-			if (fls->mirror_array[i]->cred) {
-				put_rpccred(fls->mirror_array[i]->cred);
-				fls->mirror_array[i]->cred = NULL;
-			}
-		}
-	}
-
 	if (lseg->pls_range.iomode == IOMODE_RW) {
 		struct nfs4_flexfile_layout *ffl;
 		struct inode *inode;
@@ -415,6 +568,146 @@ ff_layout_get_lseg_count(struct nfs4_ff_layout_segment *fls)
 	return 1;
 }
 
+static void
+nfs4_ff_start_busy_timer(struct nfs4_ff_busy_timer *timer, ktime_t now)
+{
+	/* first IO request? */
+	if (atomic_inc_return(&timer->n_ops) == 1) {
+		timer->start_time = now;
+	}
+}
+
+static ktime_t
+nfs4_ff_end_busy_timer(struct nfs4_ff_busy_timer *timer, ktime_t now)
+{
+	ktime_t start;
+
+	if (atomic_dec_return(&timer->n_ops) < 0)
+		WARN_ON_ONCE(1);
+
+	start = timer->start_time;
+	timer->start_time = now;
+	return ktime_sub(now, start);
+}
+
+static bool
+nfs4_ff_layoutstat_start_io(struct nfs4_ff_layout_mirror *mirror,
+			    struct nfs4_ff_layoutstat *layoutstat,
+			    ktime_t now)
+{
+	static const ktime_t notime = {0};
+	s64 report_interval = FF_LAYOUTSTATS_REPORT_INTERVAL;
+
+	nfs4_ff_start_busy_timer(&layoutstat->busy_timer, now);
+	if (ktime_equal(mirror->start_time, notime))
+		mirror->start_time = now;
+	if (ktime_equal(mirror->last_report_time, notime))
+		mirror->last_report_time = now;
+	if (layoutstats_timer != 0)
+		report_interval = (s64)layoutstats_timer * 1000LL;
+	if (ktime_to_ms(ktime_sub(now, mirror->last_report_time)) >=
+			report_interval) {
+		mirror->last_report_time = now;
+		return true;
+	}
+
+	return false;
+}
+
+static void
+nfs4_ff_layout_stat_io_update_requested(struct nfs4_ff_layoutstat *layoutstat,
+		__u64 requested)
+{
+	struct nfs4_ff_io_stat *iostat = &layoutstat->io_stat;
+
+	iostat->ops_requested++;
+	iostat->bytes_requested += requested;
+}
+
+static void
+nfs4_ff_layout_stat_io_update_completed(struct nfs4_ff_layoutstat *layoutstat,
+		__u64 requested,
+		__u64 completed,
+		ktime_t time_completed,
+		ktime_t time_started)
+{
+	struct nfs4_ff_io_stat *iostat = &layoutstat->io_stat;
+	ktime_t completion_time = ktime_sub(time_completed, time_started);
+	ktime_t timer;
+
+	iostat->ops_completed++;
+	iostat->bytes_completed += completed;
+	iostat->bytes_not_delivered += requested - completed;
+
+	timer = nfs4_ff_end_busy_timer(&layoutstat->busy_timer, time_completed);
+	iostat->total_busy_time =
+			ktime_add(iostat->total_busy_time, timer);
+	iostat->aggregate_completion_time =
+			ktime_add(iostat->aggregate_completion_time,
+					completion_time);
+}
+
+static void
+nfs4_ff_layout_stat_io_start_read(struct inode *inode,
+		struct nfs4_ff_layout_mirror *mirror,
+		__u64 requested, ktime_t now)
+{
+	bool report;
+
+	spin_lock(&mirror->lock);
+	report = nfs4_ff_layoutstat_start_io(mirror, &mirror->read_stat, now);
+	nfs4_ff_layout_stat_io_update_requested(&mirror->read_stat, requested);
+	spin_unlock(&mirror->lock);
+
+	if (report)
+		pnfs_report_layoutstat(inode, GFP_KERNEL);
+}
+
+static void
+nfs4_ff_layout_stat_io_end_read(struct rpc_task *task,
+		struct nfs4_ff_layout_mirror *mirror,
+		__u64 requested,
+		__u64 completed)
+{
+	spin_lock(&mirror->lock);
+	nfs4_ff_layout_stat_io_update_completed(&mirror->read_stat,
+			requested, completed,
+			ktime_get(), task->tk_start);
+	spin_unlock(&mirror->lock);
+}
+
+static void
+nfs4_ff_layout_stat_io_start_write(struct inode *inode,
+		struct nfs4_ff_layout_mirror *mirror,
+		__u64 requested, ktime_t now)
+{
+	bool report;
+
+	spin_lock(&mirror->lock);
+	report = nfs4_ff_layoutstat_start_io(mirror , &mirror->write_stat, now);
+	nfs4_ff_layout_stat_io_update_requested(&mirror->write_stat, requested);
+	spin_unlock(&mirror->lock);
+
+	if (report)
+		pnfs_report_layoutstat(inode, GFP_NOIO);
+}
+
+static void
+nfs4_ff_layout_stat_io_end_write(struct rpc_task *task,
+		struct nfs4_ff_layout_mirror *mirror,
+		__u64 requested,
+		__u64 completed,
+		enum nfs3_stable_how committed)
+{
+	if (committed == NFS_UNSTABLE)
+		requested = completed = 0;
+
+	spin_lock(&mirror->lock);
+	nfs4_ff_layout_stat_io_update_completed(&mirror->write_stat,
+			requested, completed, ktime_get(), task->tk_start);
+	spin_unlock(&mirror->lock);
+}
+
 static int
 ff_layout_alloc_commit_info(struct pnfs_layout_segment *lseg,
 			    struct nfs_commit_info *cinfo,
@@ -462,17 +755,17 @@ ff_layout_alloc_commit_info(struct pnfs_layout_segment *lseg,
 }
 
 static struct nfs4_pnfs_ds *
-ff_layout_choose_best_ds_for_read(struct nfs_pageio_descriptor *pgio,
+ff_layout_choose_best_ds_for_read(struct pnfs_layout_segment *lseg,
+				  int start_idx,
 				  int *best_idx)
 {
-	struct nfs4_ff_layout_segment *fls;
+	struct nfs4_ff_layout_segment *fls = FF_LAYOUT_LSEG(lseg);
 	struct nfs4_pnfs_ds *ds;
 	int idx;
 
-	fls = FF_LAYOUT_LSEG(pgio->pg_lseg);
 	/* mirrors are sorted by efficiency */
-	for (idx = 0; idx < fls->mirror_array_cnt; idx++) {
-		ds = nfs4_ff_layout_prepare_ds(pgio->pg_lseg, idx, false);
+	for (idx = start_idx; idx < fls->mirror_array_cnt; idx++) {
+		ds = nfs4_ff_layout_prepare_ds(lseg, idx, false);
 		if (ds) {
 			*best_idx = idx;
 			return ds;
@@ -503,7 +796,7 @@ ff_layout_pg_init_read(struct nfs_pageio_descriptor *pgio,
 	if (pgio->pg_lseg == NULL)
 		goto out_mds;
 
-	ds = ff_layout_choose_best_ds_for_read(pgio, &ds_idx);
+	ds = ff_layout_choose_best_ds_for_read(pgio->pg_lseg, 0, &ds_idx);
 	if (!ds)
 		goto out_mds;
 	mirror = FF_LAYOUT_COMP(pgio->pg_lseg, ds_idx);
@@ -585,8 +878,6 @@ ff_layout_pg_get_mirror_count_write(struct nfs_pageio_descriptor *pgio,
 		return FF_LAYOUT_MIRROR_COUNT(pgio->pg_lseg);
 
 	/* no lseg means that pnfs is not in use, so no mirroring here */
-	pnfs_put_lseg(pgio->pg_lseg);
-	pgio->pg_lseg = NULL;
 	nfs_pageio_reset_write_mds(pgio);
 	return 1;
 }
@@ -758,7 +1049,8 @@ static int ff_layout_async_handle_error_v4(struct rpc_task *task,
 		rpc_wake_up(&tbl->slot_tbl_waitq);
 		/* fall through */
 	default:
-		if (ff_layout_has_available_ds(lseg))
+		if (ff_layout_no_fallback_to_mds(lseg) ||
+		    ff_layout_has_available_ds(lseg))
 			return -NFS4ERR_RESET_TO_PNFS;
 reset:
 		dprintk("%s Retry through MDS. Error %d\n", __func__,
@@ -788,18 +1080,26 @@ static int ff_layout_async_handle_error_v3(struct rpc_task *task,
 	if (task->tk_status >= 0)
 		return 0;
 
-	if (task->tk_status != -EJUKEBOX) {
+	switch (task->tk_status) {
+	/* File access problems. Don't mark the device as unavailable */
+	case -EACCES:
+	case -ESTALE:
+	case -EISDIR:
+	case -EBADHANDLE:
+	case -ELOOP:
+	case -ENOSPC:
+		break;
+	case -EJUKEBOX:
+		nfs_inc_stats(lseg->pls_layout->plh_inode, NFSIOS_DELAY);
+		goto out_retry;
+	default:
 		dprintk("%s DS connection error %d\n", __func__,
 			task->tk_status);
 		nfs4_mark_deviceid_unavailable(devid);
-		if (ff_layout_has_available_ds(lseg))
-			return -NFS4ERR_RESET_TO_PNFS;
-		else
-			return -NFS4ERR_RESET_TO_MDS;
 	}
-
-	if (task->tk_status == -EJUKEBOX)
-		nfs_inc_stats(lseg->pls_layout->plh_inode, NFSIOS_DELAY);
+	/* FIXME: Need to prevent infinite looping here. */
+	return -NFS4ERR_RESET_TO_PNFS;
+out_retry:
 	task->tk_status = 0;
 	rpc_restart_call(task);
 	rpc_delay(task, NFS_JUKEBOX_RETRY_TIME);
@@ -829,54 +1129,87 @@ static int ff_layout_async_handle_error(struct rpc_task *task,
 
 static void ff_layout_io_track_ds_error(struct pnfs_layout_segment *lseg,
 					int idx, u64 offset, u64 length,
-					u32 status, int opnum)
+					u32 status, int opnum, int error)
 {
 	struct nfs4_ff_layout_mirror *mirror;
 	int err;
 
+	if (status == 0) {
+		switch (error) {
+		case -ETIMEDOUT:
+		case -EPFNOSUPPORT:
+		case -EPROTONOSUPPORT:
+		case -EOPNOTSUPP:
+		case -ECONNREFUSED:
+		case -ECONNRESET:
+		case -EHOSTDOWN:
+		case -EHOSTUNREACH:
+		case -ENETUNREACH:
+		case -EADDRINUSE:
+		case -ENOBUFS:
+		case -EPIPE:
+		case -EPERM:
+			status = NFS4ERR_NXIO;
+			break;
+		case -EACCES:
+			status = NFS4ERR_ACCESS;
+			break;
+		default:
+			return;
+		}
+	}
+
 	mirror = FF_LAYOUT_COMP(lseg, idx);
 	err = ff_layout_track_ds_error(FF_LAYOUT_FROM_HDR(lseg->pls_layout),
 				       mirror, offset, length, status, opnum,
 				       GFP_NOIO);
+	pnfs_error_mark_layout_for_return(lseg->pls_layout->plh_inode, lseg);
 	dprintk("%s: err %d op %d status %u\n", __func__, err, opnum, status);
 }
 
 /* NFS_PROTO call done callback routines */
-
 static int ff_layout_read_done_cb(struct rpc_task *task,
 				struct nfs_pgio_header *hdr)
 {
-	struct inode *inode;
 	int err;
 
 	trace_nfs4_pnfs_read(hdr, task->tk_status);
-	if (task->tk_status == -ETIMEDOUT && !hdr->res.op_status)
-		hdr->res.op_status = NFS4ERR_NXIO;
-	if (task->tk_status < 0 && hdr->res.op_status)
+	if (task->tk_status < 0)
 		ff_layout_io_track_ds_error(hdr->lseg, hdr->pgio_mirror_idx,
 					    hdr->args.offset, hdr->args.count,
-					    hdr->res.op_status, OP_READ);
+					    hdr->res.op_status, OP_READ,
+					    task->tk_status);
 	err = ff_layout_async_handle_error(task, hdr->args.context->state,
 					   hdr->ds_clp, hdr->lseg,
 					   hdr->pgio_mirror_idx);
 
 	switch (err) {
 	case -NFS4ERR_RESET_TO_PNFS:
+		if (ff_layout_choose_best_ds_for_read(hdr->lseg,
+					hdr->pgio_mirror_idx + 1,
+					&hdr->pgio_mirror_idx))
+			goto out_eagain;
 		set_bit(NFS_LAYOUT_RETURN_BEFORE_CLOSE,
 			&hdr->lseg->pls_layout->plh_flags);
 		pnfs_read_resend_pnfs(hdr);
 		return task->tk_status;
 	case -NFS4ERR_RESET_TO_MDS:
-		inode = hdr->lseg->pls_layout->plh_inode;
-		pnfs_error_mark_layout_for_return(inode, hdr->lseg);
 		ff_layout_reset_read(hdr);
 		return task->tk_status;
 	case -EAGAIN:
-		rpc_restart_call_prepare(task);
-		return -EAGAIN;
+		goto out_eagain;
 	}
 
 	return 0;
+out_eagain:
+	rpc_restart_call_prepare(task);
+	return -EAGAIN;
+}
+
+static bool
+ff_layout_need_layoutcommit(struct pnfs_layout_segment *lseg)
+{
+	return !(FF_LAYOUT_LSEG(lseg)->flags & FF_FLAGS_NO_LAYOUTCOMMIT);
 }
 
 /*
@@ -891,6 +1224,9 @@ static int ff_layout_read_done_cb(struct rpc_task *task,
 static void
 ff_layout_set_layoutcommit(struct nfs_pgio_header *hdr)
 {
+	if (!ff_layout_need_layoutcommit(hdr->lseg))
+		return;
+
 	pnfs_set_layoutcommit(hdr->inode, hdr->lseg,
 			hdr->mds_offset + hdr->res.count);
 	dprintk("%s inode %lu pls_end_pos %lu\n", __func__, hdr->inode->i_ino,
@@ -909,6 +1245,11 @@ ff_layout_reset_to_mds(struct pnfs_layout_segment *lseg, int idx)
 static int ff_layout_read_prepare_common(struct rpc_task *task,
 					 struct nfs_pgio_header *hdr)
 {
+	nfs4_ff_layout_stat_io_start_read(hdr->inode,
+			FF_LAYOUT_COMP(hdr->lseg, hdr->pgio_mirror_idx),
+			hdr->args.count,
+			task->tk_start);
+
 	if (unlikely(test_bit(NFS_CONTEXT_BAD, &hdr->args.context->flags))) {
 		rpc_exit(task, -EIO);
 		return -EIO;
@@ -962,15 +1303,15 @@ static void ff_layout_read_prepare_v4(struct rpc_task *task, void *data)
 {
 	struct nfs_pgio_header *hdr = data;
 
-	if (ff_layout_read_prepare_common(task, hdr))
-		return;
-
 	if (ff_layout_setup_sequence(hdr->ds_clp,
 				     &hdr->args.seq_args,
 				     &hdr->res.seq_res,
 				     task))
 		return;
 
+	if (ff_layout_read_prepare_common(task, hdr))
+		return;
+
 	if (nfs4_set_rw_stateid(&hdr->args.stateid, hdr->args.context,
 			hdr->args.lock_context, FMODE_READ) == -EIO)
 		rpc_exit(task, -EIO); /* lost lock, terminate I/O */
@@ -982,6 +1323,10 @@ static void ff_layout_read_call_done(struct rpc_task *task, void *data)
 
 	dprintk("--> %s task->tk_status %d\n", __func__, task->tk_status);
 
+	nfs4_ff_layout_stat_io_end_read(task,
+			FF_LAYOUT_COMP(hdr->lseg, hdr->pgio_mirror_idx),
+			hdr->args.count, hdr->res.count);
+
 	if (test_bit(NFS_IOHDR_REDO, &hdr->flags) &&
 	    task->tk_status == 0) {
 		nfs4_sequence_done(task, &hdr->res.seq_res);
@@ -1003,32 +1348,26 @@ static void ff_layout_read_count_stats(struct rpc_task *task, void *data)
 static int ff_layout_write_done_cb(struct rpc_task *task,
 				struct nfs_pgio_header *hdr)
 {
-	struct inode *inode;
 	int err;
 
 	trace_nfs4_pnfs_write(hdr, task->tk_status);
-	if (task->tk_status == -ETIMEDOUT && !hdr->res.op_status)
-		hdr->res.op_status = NFS4ERR_NXIO;
-	if (task->tk_status < 0 && hdr->res.op_status)
+	if (task->tk_status < 0)
 		ff_layout_io_track_ds_error(hdr->lseg, hdr->pgio_mirror_idx,
 					    hdr->args.offset, hdr->args.count,
-					    hdr->res.op_status, OP_WRITE);
+					    hdr->res.op_status, OP_WRITE,
+					    task->tk_status);
 	err = ff_layout_async_handle_error(task, hdr->args.context->state,
 					   hdr->ds_clp, hdr->lseg,
 					   hdr->pgio_mirror_idx);
 
 	switch (err) {
 	case -NFS4ERR_RESET_TO_PNFS:
+		pnfs_set_retry_layoutget(hdr->lseg->pls_layout);
+		ff_layout_reset_write(hdr, true);
+		return task->tk_status;
 	case -NFS4ERR_RESET_TO_MDS:
-		inode = hdr->lseg->pls_layout->plh_inode;
-		pnfs_error_mark_layout_for_return(inode, hdr->lseg);
-		if (err == -NFS4ERR_RESET_TO_PNFS) {
-			pnfs_set_retry_layoutget(hdr->lseg->pls_layout);
-			ff_layout_reset_write(hdr, true);
-		} else {
-			pnfs_clear_retry_layoutget(hdr->lseg->pls_layout);
-			ff_layout_reset_write(hdr, false);
-		}
+		pnfs_clear_retry_layoutget(hdr->lseg->pls_layout);
+		ff_layout_reset_write(hdr, false);
 		return task->tk_status;
 	case -EAGAIN:
 		rpc_restart_call_prepare(task);
@@ -1050,28 +1389,24 @@ static int ff_layout_write_done_cb(struct rpc_task *task,
 static int ff_layout_commit_done_cb(struct rpc_task *task,
 				     struct nfs_commit_data *data)
 {
-	struct inode *inode;
 	int err;
 
 	trace_nfs4_pnfs_commit_ds(data, task->tk_status);
-	if (task->tk_status == -ETIMEDOUT && !data->res.op_status)
-		data->res.op_status = NFS4ERR_NXIO;
-	if (task->tk_status < 0 && data->res.op_status)
+	if (task->tk_status < 0)
 		ff_layout_io_track_ds_error(data->lseg, data->ds_commit_index,
 					    data->args.offset, data->args.count,
-					    data->res.op_status, OP_COMMIT);
+					    data->res.op_status, OP_COMMIT,
+					    task->tk_status);
 	err = ff_layout_async_handle_error(task, NULL, data->ds_clp,
 					   data->lseg, data->ds_commit_index);
 
 	switch (err) {
 	case -NFS4ERR_RESET_TO_PNFS:
+		pnfs_set_retry_layoutget(data->lseg->pls_layout);
+		pnfs_generic_prepare_to_resend_writes(data);
+		return -EAGAIN;
 	case -NFS4ERR_RESET_TO_MDS:
-		inode = data->lseg->pls_layout->plh_inode;
-		pnfs_error_mark_layout_for_return(inode, data->lseg);
-		if (err == -NFS4ERR_RESET_TO_PNFS)
-			pnfs_set_retry_layoutget(data->lseg->pls_layout);
-		else
-			pnfs_clear_retry_layoutget(data->lseg->pls_layout);
+		pnfs_clear_retry_layoutget(data->lseg->pls_layout);
 		pnfs_generic_prepare_to_resend_writes(data);
 		return -EAGAIN;
 	case -EAGAIN:
@@ -1079,7 +1414,8 @@ static int ff_layout_commit_done_cb(struct rpc_task *task,
 		return -EAGAIN;
 	}
 
-	if (data->verf.committed == NFS_UNSTABLE)
+	if (data->verf.committed == NFS_UNSTABLE
+	    && ff_layout_need_layoutcommit(data->lseg))
 		pnfs_set_layoutcommit(data->inode, data->lseg, data->lwb);
 
 	return 0;
@@ -1088,6 +1424,11 @@ static int ff_layout_commit_done_cb(struct rpc_task *task,
 static int ff_layout_write_prepare_common(struct rpc_task *task,
 					  struct nfs_pgio_header *hdr)
 {
+	nfs4_ff_layout_stat_io_start_write(hdr->inode,
+			FF_LAYOUT_COMP(hdr->lseg, hdr->pgio_mirror_idx),
+			hdr->args.count,
+			task->tk_start);
+
 	if (unlikely(test_bit(NFS_CONTEXT_BAD, &hdr->args.context->flags))) {
 		rpc_exit(task, -EIO);
 		return -EIO;
@@ -1121,15 +1462,15 @@ static void ff_layout_write_prepare_v4(struct rpc_task *task, void *data)
 {
 	struct nfs_pgio_header *hdr = data;
 
-	if (ff_layout_write_prepare_common(task, hdr))
-		return;
-
 	if (ff_layout_setup_sequence(hdr->ds_clp,
 				     &hdr->args.seq_args,
 				     &hdr->res.seq_res,
 				     task))
 		return;
 
+	if (ff_layout_write_prepare_common(task, hdr))
+		return;
+
 	if (nfs4_set_rw_stateid(&hdr->args.stateid, hdr->args.context,
 			hdr->args.lock_context, FMODE_WRITE) == -EIO)
 		rpc_exit(task, -EIO); /* lost lock, terminate I/O */
@@ -1139,6 +1480,11 @@ static void ff_layout_write_call_done(struct rpc_task *task, void *data)
 {
 	struct nfs_pgio_header *hdr = data;
 
+	nfs4_ff_layout_stat_io_end_write(task,
+			FF_LAYOUT_COMP(hdr->lseg, hdr->pgio_mirror_idx),
+			hdr->args.count, hdr->res.count,
+			hdr->res.verf->committed);
+
 	if (test_bit(NFS_IOHDR_REDO, &hdr->flags) &&
 	    task->tk_status == 0) {
 		nfs4_sequence_done(task, &hdr->res.seq_res);
@@ -1157,8 +1503,17 @@ static void ff_layout_write_count_stats(struct rpc_task *task, void *data)
 	    &NFS_CLIENT(hdr->inode)->cl_metrics[NFSPROC4_CLNT_WRITE]);
 }
 
+static void ff_layout_commit_prepare_common(struct rpc_task *task,
+		struct nfs_commit_data *cdata)
+{
+	nfs4_ff_layout_stat_io_start_write(cdata->inode,
+			FF_LAYOUT_COMP(cdata->lseg, cdata->ds_commit_index),
+			0, task->tk_start);
+}
+
 static void ff_layout_commit_prepare_v3(struct rpc_task *task, void *data)
 {
+	ff_layout_commit_prepare_common(task, data);
 	rpc_call_start(task);
 }
 
@@ -1166,10 +1521,30 @@ static void ff_layout_commit_prepare_v4(struct rpc_task *task, void *data)
 {
 	struct nfs_commit_data *wdata = data;
 
-	ff_layout_setup_sequence(wdata->ds_clp,
+	if (ff_layout_setup_sequence(wdata->ds_clp,
 				 &wdata->args.seq_args,
 				 &wdata->res.seq_res,
-				 task);
+				 task))
+		return;
+	ff_layout_commit_prepare_common(task, data);
+}
+
+static void ff_layout_commit_done(struct rpc_task *task, void *data)
+{
+	struct nfs_commit_data *cdata = data;
+	struct nfs_page *req;
+	__u64 count = 0;
+
+	if (task->tk_status == 0) {
+		list_for_each_entry(req, &cdata->pages, wb_list)
+			count += req->wb_bytes;
+	}
+
+	nfs4_ff_layout_stat_io_end_write(task,
+			FF_LAYOUT_COMP(cdata->lseg, cdata->ds_commit_index),
+			count, count, NFS_FILE_SYNC);
+
+	pnfs_generic_write_commit_done(task, data);
 }
 
 static void ff_layout_commit_count_stats(struct rpc_task *task, void *data)
@@ -1210,14 +1585,14 @@ static const struct rpc_call_ops ff_layout_write_call_ops_v4 = {
 
 static const struct rpc_call_ops ff_layout_commit_call_ops_v3 = {
 	.rpc_call_prepare = ff_layout_commit_prepare_v3,
-	.rpc_call_done = pnfs_generic_write_commit_done,
+	.rpc_call_done = ff_layout_commit_done,
 	.rpc_count_stats = ff_layout_commit_count_stats,
 	.rpc_release = pnfs_generic_commit_release,
 };
 
 static const struct rpc_call_ops ff_layout_commit_call_ops_v4 = {
 	.rpc_call_prepare = ff_layout_commit_prepare_v4,
-	.rpc_call_done = pnfs_generic_write_commit_done,
+	.rpc_call_done = ff_layout_commit_done,
 	.rpc_count_stats = ff_layout_commit_count_stats,
 	.rpc_release = pnfs_generic_commit_release,
 };
@@ -1261,7 +1636,6 @@ ff_layout_read_pagelist(struct nfs_pgio_header *hdr)
 	fh = nfs4_ff_layout_select_ds_fh(lseg, idx);
 	if (fh)
 		hdr->args.fh = fh;
-
 	/*
 	 * Note that if we ever decide to split across DSes,
 	 * then we may need to handle dense-like offsets.
@@ -1390,6 +1764,7 @@ static int ff_layout_initiate_commit(struct nfs_commit_data *data, int how)
 	fh = select_ds_fh_from_commit(lseg, data->ds_commit_index);
 	if (fh)
 		data->args.fh = fh;
+
 	return nfs_initiate_commit(ds_clnt, data, ds->ds_clp->rpc_ops,
 				   vers == 3 ? &ff_layout_commit_call_ops_v3 :
 					       &ff_layout_commit_call_ops_v4,
@@ -1484,15 +1859,250 @@ ff_layout_encode_layoutreturn(struct pnfs_layout_hdr *lo,
 	start = xdr_reserve_space(xdr, 4);
 	BUG_ON(!start);
 
-	if (ff_layout_encode_ioerr(flo, xdr, args))
-		goto out;
-
+	ff_layout_encode_ioerr(flo, xdr, args);
 	ff_layout_encode_iostats(flo, xdr, args);
-out:
+
 	*start = cpu_to_be32((xdr->p - start - 1) * 4);
 	dprintk("%s: Return\n", __func__);
 }
 
+static int
+ff_layout_ntop4(const struct sockaddr *sap, char *buf, const size_t buflen)
+{
+	const struct sockaddr_in *sin = (struct sockaddr_in *)sap;
+
+	return snprintf(buf, buflen, "%pI4", &sin->sin_addr);
+}
+
+static size_t
+ff_layout_ntop6_noscopeid(const struct sockaddr *sap, char *buf,
+			  const int buflen)
+{
+	const struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)sap;
+	const struct in6_addr *addr = &sin6->sin6_addr;
+
+	/*
+	 * RFC 4291, Section 2.2.2
+	 *
+	 * Shorthanded ANY address
+	 */
+	if (ipv6_addr_any(addr))
+		return snprintf(buf, buflen, "::");
+
+	/*
+	 * RFC 4291, Section 2.2.2
+	 *
+	 * Shorthanded loopback address
+	 */
+	if (ipv6_addr_loopback(addr))
+		return snprintf(buf, buflen, "::1");
+
+	/*
+	 * RFC 4291, Section 2.2.3
+	 *
+	 * Special presentation address format for mapped v4
+	 * addresses.
+	 */
+	if (ipv6_addr_v4mapped(addr))
+		return snprintf(buf, buflen, "::ffff:%pI4",
+					&addr->s6_addr32[3]);
+
+	/*
+	 * RFC 4291, Section 2.2.1
+	 */
+	return snprintf(buf, buflen, "%pI6c", addr);
+}
+
+/* Derived from rpc_sockaddr2uaddr */
+static void
+ff_layout_encode_netaddr(struct xdr_stream *xdr, struct nfs4_pnfs_ds_addr *da)
+{
+	struct sockaddr *sap = (struct sockaddr *)&da->da_addr;
+	char portbuf[RPCBIND_MAXUADDRPLEN];
+	char addrbuf[RPCBIND_MAXUADDRLEN];
+	char *netid;
+	unsigned short port;
+	int len, netid_len;
+	__be32 *p;
+
+	switch (sap->sa_family) {
+	case AF_INET:
+		if (ff_layout_ntop4(sap, addrbuf, sizeof(addrbuf)) == 0)
+			return;
+		port = ntohs(((struct sockaddr_in *)sap)->sin_port);
+		netid = "tcp";
+		netid_len = 3;
+		break;
+	case AF_INET6:
+		if (ff_layout_ntop6_noscopeid(sap, addrbuf, sizeof(addrbuf)) == 0)
+			return;
+		port = ntohs(((struct sockaddr_in6 *)sap)->sin6_port);
+		netid = "tcp6";
+		netid_len = 4;
+		break;
+	default:
+		/* we only support tcp and tcp6 */
+		WARN_ON_ONCE(1);
+		return;
+	}
+
+	snprintf(portbuf, sizeof(portbuf), ".%u.%u", port >> 8, port & 0xff);
+	len = strlcat(addrbuf, portbuf, sizeof(addrbuf));
+
+	p = xdr_reserve_space(xdr, 4 + netid_len);
+	xdr_encode_opaque(p, netid, netid_len);
+
+	p = xdr_reserve_space(xdr, 4 + len);
+	xdr_encode_opaque(p, addrbuf, len);
+}
+
+static void
+ff_layout_encode_nfstime(struct xdr_stream *xdr,
+			 ktime_t t)
+{
+	struct timespec64 ts;
+	__be32 *p;
+
+	p = xdr_reserve_space(xdr, 12);
+	ts = ktime_to_timespec64(t);
+	p = xdr_encode_hyper(p, ts.tv_sec);
+	*p++ = cpu_to_be32(ts.tv_nsec);
+}
+
+static void
+ff_layout_encode_io_latency(struct xdr_stream *xdr,
+			    struct nfs4_ff_io_stat *stat)
+{
+	__be32 *p;
+
+	p = xdr_reserve_space(xdr, 5 * 8);
+	p = xdr_encode_hyper(p, stat->ops_requested);
+	p = xdr_encode_hyper(p, stat->bytes_requested);
+	p = xdr_encode_hyper(p, stat->ops_completed);
+	p = xdr_encode_hyper(p, stat->bytes_completed);
+	p = xdr_encode_hyper(p, stat->bytes_not_delivered);
+	ff_layout_encode_nfstime(xdr, stat->total_busy_time);
+	ff_layout_encode_nfstime(xdr, stat->aggregate_completion_time);
+}
+
+static void
+ff_layout_encode_layoutstats(struct xdr_stream *xdr,
+			     struct nfs42_layoutstat_args *args,
+			     struct nfs42_layoutstat_devinfo *devinfo)
+{
+	struct nfs4_ff_layout_mirror *mirror = devinfo->layout_private;
+	struct nfs4_pnfs_ds_addr *da;
+	struct nfs4_pnfs_ds *ds = mirror->mirror_ds->ds;
+	struct nfs_fh *fh = &mirror->fh_versions[0];
+	__be32 *p, *start;
+
+	da = list_first_entry(&ds->ds_addrs, struct nfs4_pnfs_ds_addr, da_node);
+	dprintk("%s: DS %s: encoding address %s\n",
+		__func__, ds->ds_remotestr, da->da_remotestr);
+	/* layoutupdate length */
+	start = xdr_reserve_space(xdr, 4);
+	/* netaddr4 */
+	ff_layout_encode_netaddr(xdr, da);
+	/* nfs_fh4 */
+	p = xdr_reserve_space(xdr, 4 + fh->size);
+	xdr_encode_opaque(p, fh->data, fh->size);
+	/* ff_io_latency4 read */
+	spin_lock(&mirror->lock);
+	ff_layout_encode_io_latency(xdr, &mirror->read_stat.io_stat);
+	/* ff_io_latency4 write */
+	ff_layout_encode_io_latency(xdr, &mirror->write_stat.io_stat);
+	spin_unlock(&mirror->lock);
+	/* nfstime4 */
+	ff_layout_encode_nfstime(xdr, ktime_sub(ktime_get(), mirror->start_time));
+	/* bool */
+	p = xdr_reserve_space(xdr, 4);
+	*p = cpu_to_be32(false);
+
+	*start = cpu_to_be32((xdr->p - start - 1) * 4);
+}
+
+static int
+ff_layout_mirror_prepare_stats(struct nfs42_layoutstat_args *args,
+			       struct pnfs_layout_hdr *lo,
+			       int dev_limit)
+{
+	struct nfs4_flexfile_layout *ff_layout = FF_LAYOUT_FROM_HDR(lo);
+	struct nfs4_ff_layout_mirror *mirror;
+	struct nfs4_deviceid_node *dev;
+	struct nfs42_layoutstat_devinfo *devinfo;
+	int i = 0;
+
+	list_for_each_entry(mirror, &ff_layout->mirrors, mirrors) {
+		if (i >= dev_limit)
+			break;
+		if (!mirror->mirror_ds)
+			continue;
+		/* mirror refcount put in cleanup_layoutstats */
+		if (!atomic_inc_not_zero(&mirror->ref))
+			continue;
+		dev = &mirror->mirror_ds->id_node; 
+		devinfo = &args->devinfo[i];
+		memcpy(&devinfo->dev_id, &dev->deviceid, NFS4_DEVICEID4_SIZE);
+		devinfo->offset = 0;
+		devinfo->length = NFS4_MAX_UINT64;
+		devinfo->read_count = mirror->read_stat.io_stat.ops_completed;
+		devinfo->read_bytes = mirror->read_stat.io_stat.bytes_completed;
+		devinfo->write_count = mirror->write_stat.io_stat.ops_completed;
+		devinfo->write_bytes = mirror->write_stat.io_stat.bytes_completed;
+		devinfo->layout_type = LAYOUT_FLEX_FILES;
+		devinfo->layoutstats_encode = ff_layout_encode_layoutstats;
+		devinfo->layout_private = mirror;
+
+		i++;
+	}
+	return i;
+}
+
+static int
+ff_layout_prepare_layoutstats(struct nfs42_layoutstat_args *args)
+{
+	struct nfs4_flexfile_layout *ff_layout;
+	struct nfs4_ff_layout_mirror *mirror;
+	int dev_count = 0;
+
+	spin_lock(&args->inode->i_lock);
+	ff_layout = FF_LAYOUT_FROM_HDR(NFS_I(args->inode)->layout);
+	list_for_each_entry(mirror, &ff_layout->mirrors, mirrors) {
+		if (atomic_read(&mirror->ref) != 0)
+			dev_count ++;
+	}
+	spin_unlock(&args->inode->i_lock);
+	/* For now, send at most PNFS_LAYOUTSTATS_MAXDEV statistics */
+	if (dev_count > PNFS_LAYOUTSTATS_MAXDEV) {
+		dprintk("%s: truncating devinfo to limit (%d:%d)\n",
+			__func__, dev_count, PNFS_LAYOUTSTATS_MAXDEV);
+		dev_count = PNFS_LAYOUTSTATS_MAXDEV;
+	}
+	args->devinfo = kmalloc_array(dev_count, sizeof(*args->devinfo), GFP_NOIO);
+	if (!args->devinfo)
+		return -ENOMEM;
+
+	spin_lock(&args->inode->i_lock);
+	args->num_dev = ff_layout_mirror_prepare_stats(args,
+			&ff_layout->generic_hdr, dev_count);
+	spin_unlock(&args->inode->i_lock);
+
+	return 0;
+}
+
+static void
+ff_layout_cleanup_layoutstats(struct nfs42_layoutstat_data *data)
+{
+	struct nfs4_ff_layout_mirror *mirror;
+	int i;
+
+	for (i = 0; i < data->args.num_dev; i++) {
+		mirror = data->args.devinfo[i].layout_private;
+		data->args.devinfo[i].layout_private = NULL;
+		ff_layout_put_mirror(mirror);
+	}
+}
+
 static struct pnfs_layoutdriver_type flexfilelayout_type = {
 	.id			= LAYOUT_FLEX_FILES,
 	.name			= "LAYOUT_FLEX_FILES",
@@ -1501,6 +2111,7 @@ static struct pnfs_layoutdriver_type flexfilelayout_type = {
 	.free_layout_hdr	= ff_layout_free_layout_hdr,
 	.alloc_lseg		= ff_layout_alloc_lseg,
 	.free_lseg		= ff_layout_free_lseg,
+	.add_lseg		= ff_layout_add_lseg,
 	.pg_read_ops		= &ff_layout_pg_read_ops,
 	.pg_write_ops		= &ff_layout_pg_write_ops,
 	.get_ds_info		= ff_layout_get_ds_info,
@@ -1515,6 +2126,8 @@ static struct pnfs_layoutdriver_type flexfilelayout_type = {
 	.alloc_deviceid_node    = ff_layout_alloc_deviceid_node,
 	.encode_layoutreturn    = ff_layout_encode_layoutreturn,
 	.sync			= pnfs_nfs_generic_sync,
+	.prepare_layoutstats	= ff_layout_prepare_layoutstats,
+	.cleanup_layoutstats	= ff_layout_cleanup_layoutstats,
 };
 
 static int __init nfs4flexfilelayout_init(void)