3 files changed, 240 insertions, 149 deletions

diff --git a/fs/nfs/flexfilelayout/flexfilelayout.c b/fs/nfs/flexfilelayout/flexfilelayout.c
index 0cb1abd535e3..51b51369704c 100644
--- a/fs/nfs/flexfilelayout/flexfilelayout.c
+++ b/fs/nfs/flexfilelayout/flexfilelayout.c
@@ -26,6 +26,8 @@
 
 #define FF_LAYOUT_POLL_RETRY_MAX     (15*HZ)
 
+static struct group_info	*ff_zero_group;
+
 static struct pnfs_layout_hdr *
 ff_layout_alloc_layout_hdr(struct inode *inode, gfp_t gfp_flags)
 {
@@ -35,6 +37,7 @@ ff_layout_alloc_layout_hdr(struct inode *inode, gfp_t gfp_flags)
 	if (ffl) {
 		INIT_LIST_HEAD(&ffl->error_list);
 		INIT_LIST_HEAD(&ffl->mirrors);
+		ffl->last_report_time = ktime_get();
 		return &ffl->generic_hdr;
 	} else
 		return NULL;
@@ -53,14 +56,15 @@ ff_layout_free_layout_hdr(struct pnfs_layout_hdr *lo)
 	kfree(FF_LAYOUT_FROM_HDR(lo));
 }
 
-static int decode_stateid(struct xdr_stream *xdr, nfs4_stateid *stateid)
+static int decode_pnfs_stateid(struct xdr_stream *xdr, nfs4_stateid *stateid)
 {
 	__be32 *p;
 
 	p = xdr_inline_decode(xdr, NFS4_STATEID_SIZE);
 	if (unlikely(p == NULL))
 		return -ENOBUFS;
-	memcpy(stateid, p, NFS4_STATEID_SIZE);
+	stateid->type = NFS4_PNFS_DS_STATEID_TYPE;
+	memcpy(stateid->data, p, NFS4_STATEID_SIZE);
 	dprintk("%s: stateid id= [%x%x%x%x]\n", __func__,
 		p[0], p[1], p[2], p[3]);
 	return 0;
@@ -211,10 +215,16 @@ static struct nfs4_ff_layout_mirror *ff_layout_alloc_mirror(gfp_t gfp_flags)
 
 static void ff_layout_free_mirror(struct nfs4_ff_layout_mirror *mirror)
 {
+	struct rpc_cred	*cred;
+
 	ff_layout_remove_mirror(mirror);
 	kfree(mirror->fh_versions);
-	if (mirror->cred)
-		put_rpccred(mirror->cred);
+	cred = rcu_access_pointer(mirror->ro_cred);
+	if (cred)
+		put_rpccred(cred);
+	cred = rcu_access_pointer(mirror->rw_cred);
+	if (cred)
+		put_rpccred(cred);
 	nfs4_ff_layout_put_deviceid(mirror->mirror_ds);
 	kfree(mirror);
 }
@@ -290,6 +300,8 @@ ff_lseg_merge(struct pnfs_layout_segment *new,
 {
 	u64 new_end, old_end;
 
+	if (test_bit(NFS_LSEG_LAYOUTRETURN, &old->pls_flags))
+		return false;
 	if (new->pls_range.iomode != old->pls_range.iomode)
 		return false;
 	old_end = pnfs_calc_offset_end(old->pls_range.offset,
@@ -310,8 +322,6 @@ ff_lseg_merge(struct pnfs_layout_segment *new,
 			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;
 }
 
@@ -407,8 +417,9 @@ ff_layout_alloc_lseg(struct pnfs_layout_hdr *lh,
 		struct nfs4_ff_layout_mirror *mirror;
 		struct nfs4_deviceid devid;
 		struct nfs4_deviceid_node *idnode;
-		u32 ds_count;
-		u32 fh_count;
+		struct auth_cred acred = { .group_info = ff_zero_group };
+		struct rpc_cred	__rcu *cred;
+		u32 ds_count, fh_count, id;
 		int j;
 
 		rc = -EIO;
@@ -456,7 +467,7 @@ ff_layout_alloc_lseg(struct pnfs_layout_hdr *lh,
 		fls->mirror_array[i]->efficiency = be32_to_cpup(p);
 
 		/* stateid */
-		rc = decode_stateid(&stream, &fls->mirror_array[i]->stateid);
+		rc = decode_pnfs_stateid(&stream, &fls->mirror_array[i]->stateid);
 		if (rc)
 			goto out_err_free;
 
@@ -484,24 +495,49 @@ ff_layout_alloc_lseg(struct pnfs_layout_hdr *lh,
 		fls->mirror_array[i]->fh_versions_cnt = fh_count;
 
 		/* user */
-		rc = decode_name(&stream, &fls->mirror_array[i]->uid);
+		rc = decode_name(&stream, &id);
 		if (rc)
 			goto out_err_free;
 
+		acred.uid = make_kuid(&init_user_ns, id);
+
 		/* group */
-		rc = decode_name(&stream, &fls->mirror_array[i]->gid);
+		rc = decode_name(&stream, &id);
 		if (rc)
 			goto out_err_free;
 
+		acred.gid = make_kgid(&init_user_ns, id);
+
+		/* find the cred for it */
+		rcu_assign_pointer(cred, rpc_lookup_generic_cred(&acred, 0, gfp_flags));
+		if (IS_ERR(cred)) {
+			rc = PTR_ERR(cred);
+			goto out_err_free;
+		}
+
+		if (lgr->range.iomode == IOMODE_READ)
+			rcu_assign_pointer(fls->mirror_array[i]->ro_cred, cred);
+		else
+			rcu_assign_pointer(fls->mirror_array[i]->rw_cred, cred);
+
 		mirror = ff_layout_add_mirror(lh, fls->mirror_array[i]);
 		if (mirror != fls->mirror_array[i]) {
+			/* swap cred ptrs so free_mirror will clean up old */
+			if (lgr->range.iomode == IOMODE_READ) {
+				cred = xchg(&mirror->ro_cred, cred);
+				rcu_assign_pointer(fls->mirror_array[i]->ro_cred, cred);
+			} else {
+				cred = xchg(&mirror->rw_cred, cred);
+				rcu_assign_pointer(fls->mirror_array[i]->rw_cred, cred);
+			}
 			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);
+		dprintk("%s: iomode %s uid %u gid %u\n", __func__,
+			lgr->range.iomode == IOMODE_READ ? "READ" : "RW",
+			from_kuid(&init_user_ns, acred.uid),
+			from_kgid(&init_user_ns, acred.gid));
 	}
 
 	p = xdr_inline_decode(&stream, 4);
@@ -605,19 +641,18 @@ nfs4_ff_layoutstat_start_io(struct nfs4_ff_layout_mirror *mirror,
 {
 	static const ktime_t notime = {0};
 	s64 report_interval = FF_LAYOUTSTATS_REPORT_INTERVAL;
+	struct nfs4_flexfile_layout *ffl = FF_LAYOUT_FROM_HDR(mirror->layout);
 
 	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 (mirror->report_interval != 0)
 		report_interval = (s64)mirror->report_interval * 1000LL;
 	else if (layoutstats_timer != 0)
 		report_interval = (s64)layoutstats_timer * 1000LL;
-	if (ktime_to_ms(ktime_sub(now, mirror->last_report_time)) >=
+	if (ktime_to_ms(ktime_sub(now, ffl->last_report_time)) >=
 			report_interval) {
-		mirror->last_report_time = now;
+		ffl->last_report_time = now;
 		return true;
 	}
 
@@ -745,7 +780,7 @@ ff_layout_alloc_commit_info(struct pnfs_layout_segment *lseg,
 	else {
 		int i;
 
-		spin_lock(cinfo->lock);
+		spin_lock(&cinfo->inode->i_lock);
 		if (cinfo->ds->nbuckets != 0)
 			kfree(buckets);
 		else {
@@ -759,7 +794,7 @@ ff_layout_alloc_commit_info(struct pnfs_layout_segment *lseg,
 					NFS_INVALID_STABLE_HOW;
 			}
 		}
-		spin_unlock(cinfo->lock);
+		spin_unlock(&cinfo->inode->i_lock);
 		return 0;
 	}
 }
@@ -771,11 +806,14 @@ ff_layout_choose_best_ds_for_read(struct pnfs_layout_segment *lseg,
 {
 	struct nfs4_ff_layout_segment *fls = FF_LAYOUT_LSEG(lseg);
 	struct nfs4_pnfs_ds *ds;
+	bool fail_return = false;
 	int idx;
 
 	/* mirrors are sorted by efficiency */
 	for (idx = start_idx; idx < fls->mirror_array_cnt; idx++) {
-		ds = nfs4_ff_layout_prepare_ds(lseg, idx, false);
+		if (idx+1 == fls->mirror_array_cnt)
+			fail_return = true;
+		ds = nfs4_ff_layout_prepare_ds(lseg, idx, fail_return);
 		if (ds) {
 			*best_idx = idx;
 			return ds;
@@ -786,6 +824,36 @@ ff_layout_choose_best_ds_for_read(struct pnfs_layout_segment *lseg,
 }
 
 static void
+ff_layout_pg_get_read(struct nfs_pageio_descriptor *pgio,
+		      struct nfs_page *req,
+		      bool strict_iomode)
+{
+retry_strict:
+	pnfs_put_lseg(pgio->pg_lseg);
+	pgio->pg_lseg = pnfs_update_layout(pgio->pg_inode,
+					   req->wb_context,
+					   0,
+					   NFS4_MAX_UINT64,
+					   IOMODE_READ,
+					   strict_iomode,
+					   GFP_KERNEL);
+	if (IS_ERR(pgio->pg_lseg)) {
+		pgio->pg_error = PTR_ERR(pgio->pg_lseg);
+		pgio->pg_lseg = NULL;
+	}
+
+	/* If we don't have checking, do get a IOMODE_RW
+	 * segment, and the server wants to avoid READs
+	 * there, then retry!
+	 */
+	if (pgio->pg_lseg && !strict_iomode &&
+	    ff_layout_avoid_read_on_rw(pgio->pg_lseg)) {
+		strict_iomode = true;
+		goto retry_strict;
+	}
+}
+
+static void
 ff_layout_pg_init_read(struct nfs_pageio_descriptor *pgio,
 			struct nfs_page *req)
 {
@@ -794,27 +862,28 @@ ff_layout_pg_init_read(struct nfs_pageio_descriptor *pgio,
 	struct nfs4_pnfs_ds *ds;
 	int ds_idx;
 
+retry:
 	/* Use full layout for now */
-	if (!pgio->pg_lseg) {
-		pgio->pg_lseg = pnfs_update_layout(pgio->pg_inode,
-						   req->wb_context,
-						   0,
-						   NFS4_MAX_UINT64,
-						   IOMODE_READ,
-						   GFP_KERNEL);
-		if (IS_ERR(pgio->pg_lseg)) {
-			pgio->pg_error = PTR_ERR(pgio->pg_lseg);
-			pgio->pg_lseg = NULL;
-			return;
-		}
-	}
+	if (!pgio->pg_lseg)
+		ff_layout_pg_get_read(pgio, req, false);
+	else if (ff_layout_avoid_read_on_rw(pgio->pg_lseg))
+		ff_layout_pg_get_read(pgio, req, true);
+
 	/* If no lseg, fall back to read through mds */
 	if (pgio->pg_lseg == NULL)
 		goto out_mds;
 
 	ds = ff_layout_choose_best_ds_for_read(pgio->pg_lseg, 0, &ds_idx);
-	if (!ds)
-		goto out_mds;
+	if (!ds) {
+		if (!ff_layout_no_fallback_to_mds(pgio->pg_lseg))
+			goto out_mds;
+		pnfs_put_lseg(pgio->pg_lseg);
+		pgio->pg_lseg = NULL;
+		/* Sleep for 1 second before retrying */
+		ssleep(1);
+		goto retry;
+	}
+
 	mirror = FF_LAYOUT_COMP(pgio->pg_lseg, ds_idx);
 
 	pgio->pg_mirror_idx = ds_idx;
@@ -841,12 +910,14 @@ ff_layout_pg_init_write(struct nfs_pageio_descriptor *pgio,
 	int i;
 	int status;
 
+retry:
 	if (!pgio->pg_lseg) {
 		pgio->pg_lseg = pnfs_update_layout(pgio->pg_inode,
 						   req->wb_context,
 						   0,
 						   NFS4_MAX_UINT64,
 						   IOMODE_RW,
+						   false,
 						   GFP_NOFS);
 		if (IS_ERR(pgio->pg_lseg)) {
 			pgio->pg_error = PTR_ERR(pgio->pg_lseg);
@@ -870,8 +941,15 @@ ff_layout_pg_init_write(struct nfs_pageio_descriptor *pgio,
 
 	for (i = 0; i < pgio->pg_mirror_count; i++) {
 		ds = nfs4_ff_layout_prepare_ds(pgio->pg_lseg, i, true);
-		if (!ds)
-			goto out_mds;
+		if (!ds) {
+			if (!ff_layout_no_fallback_to_mds(pgio->pg_lseg))
+				goto out_mds;
+			pnfs_put_lseg(pgio->pg_lseg);
+			pgio->pg_lseg = NULL;
+			/* Sleep for 1 second before retrying */
+			ssleep(1);
+			goto retry;
+		}
 		pgm = &pgio->pg_mirrors[i];
 		mirror = FF_LAYOUT_COMP(pgio->pg_lseg, i);
 		pgm->pg_bsize = mirror->mirror_ds->ds_versions[0].wsize;
@@ -895,6 +973,7 @@ ff_layout_pg_get_mirror_count_write(struct nfs_pageio_descriptor *pgio,
 						   0,
 						   NFS4_MAX_UINT64,
 						   IOMODE_RW,
+						   false,
 						   GFP_NOFS);
 		if (IS_ERR(pgio->pg_lseg)) {
 			pgio->pg_error = PTR_ERR(pgio->pg_lseg);
@@ -1067,8 +1146,7 @@ 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_no_fallback_to_mds(lseg) ||
-		    ff_layout_has_available_ds(lseg))
+		if (ff_layout_avoid_mds_available_ds(lseg))
 			return -NFS4ERR_RESET_TO_PNFS;
 reset:
 		dprintk("%s Retry through MDS. Error %d\n", __func__,
@@ -1215,8 +1293,6 @@ static int ff_layout_read_done_cb(struct rpc_task *task,
 					hdr->pgio_mirror_idx + 1,
 					&hdr->pgio_mirror_idx))
 			goto out_eagain;
-		set_bit(NFS_LAYOUT_RETURN_REQUESTED,
-			&hdr->lseg->pls_layout->plh_flags);
 		pnfs_read_resend_pnfs(hdr);
 		return task->tk_status;
 	case -NFS4ERR_RESET_TO_MDS:
@@ -1248,19 +1324,20 @@ ff_layout_need_layoutcommit(struct pnfs_layout_segment *lseg)
  * we always send layoutcommit after DS writes.
  */
 static void
-ff_layout_set_layoutcommit(struct nfs_pgio_header *hdr)
+ff_layout_set_layoutcommit(struct inode *inode,
+		struct pnfs_layout_segment *lseg,
+		loff_t end_offset)
 {
-	if (!ff_layout_need_layoutcommit(hdr->lseg))
+	if (!ff_layout_need_layoutcommit(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,
-		(unsigned long) NFS_I(hdr->inode)->layout->plh_lwb);
+	pnfs_set_layoutcommit(inode, lseg, end_offset);
+	dprintk("%s inode %lu pls_end_pos %llu\n", __func__, inode->i_ino,
+		(unsigned long long) NFS_I(inode)->layout->plh_lwb);
 }
 
 static bool
-ff_layout_reset_to_mds(struct pnfs_layout_segment *lseg, int idx)
+ff_layout_device_unavailable(struct pnfs_layout_segment *lseg, int idx)
 {
 	/* No mirroring for now */
 	struct nfs4_deviceid_node *node = FF_LAYOUT_DEVID_NODE(lseg, idx);
@@ -1297,16 +1374,10 @@ static int ff_layout_read_prepare_common(struct rpc_task *task,
 		rpc_exit(task, -EIO);
 		return -EIO;
 	}
-	if (ff_layout_reset_to_mds(hdr->lseg, hdr->pgio_mirror_idx)) {
-		dprintk("%s task %u reset io to MDS\n", __func__, task->tk_pid);
-		if (ff_layout_has_available_ds(hdr->lseg))
-			pnfs_read_resend_pnfs(hdr);
-		else
-			ff_layout_reset_read(hdr);
-		rpc_exit(task, 0);
+	if (ff_layout_device_unavailable(hdr->lseg, hdr->pgio_mirror_idx)) {
+		rpc_exit(task, -EHOSTDOWN);
 		return -EAGAIN;
 	}
-	hdr->pgio_done_cb = ff_layout_read_done_cb;
 
 	ff_layout_read_record_layoutstats_start(task, hdr);
 	return 0;
@@ -1398,6 +1469,7 @@ static void ff_layout_read_release(void *data)
 static int ff_layout_write_done_cb(struct rpc_task *task,
 				struct nfs_pgio_header *hdr)
 {
+	loff_t end_offs = 0;
 	int err;
 
 	trace_nfs4_pnfs_write(hdr, task->tk_status);
@@ -1423,7 +1495,10 @@ static int ff_layout_write_done_cb(struct rpc_task *task,
 
 	if (hdr->res.verf->committed == NFS_FILE_SYNC ||
 	    hdr->res.verf->committed == NFS_DATA_SYNC)
-		ff_layout_set_layoutcommit(hdr);
+		end_offs = hdr->mds_offset + (loff_t)hdr->res.count;
+
+	/* Note: if the write is unstable, don't set end_offs until commit */
+	ff_layout_set_layoutcommit(hdr->inode, hdr->lseg, end_offs);
 
 	/* zero out fattr since we don't care DS attr at all */
 	hdr->fattr.valid = 0;
@@ -1459,9 +1534,7 @@ static int ff_layout_commit_done_cb(struct rpc_task *task,
 		return -EAGAIN;
 	}
 
-	if (data->verf.committed == NFS_UNSTABLE
-	    && ff_layout_need_layoutcommit(data->lseg))
-		pnfs_set_layoutcommit(data->inode, data->lseg, data->lwb);
+	ff_layout_set_layoutcommit(data->inode, data->lseg, data->lwb);
 
 	return 0;
 }
@@ -1496,14 +1569,8 @@ static int ff_layout_write_prepare_common(struct rpc_task *task,
 		return -EIO;
 	}
 
-	if (ff_layout_reset_to_mds(hdr->lseg, hdr->pgio_mirror_idx)) {
-		bool retry_pnfs;
-
-		retry_pnfs = ff_layout_has_available_ds(hdr->lseg);
-		dprintk("%s task %u reset io to %s\n", __func__,
-			task->tk_pid, retry_pnfs ? "pNFS" : "MDS");
-		ff_layout_reset_write(hdr, retry_pnfs);
-		rpc_exit(task, 0);
+	if (ff_layout_device_unavailable(hdr->lseg, hdr->pgio_mirror_idx)) {
+		rpc_exit(task, -EHOSTDOWN);
 		return -EAGAIN;
 	}
 
@@ -1712,7 +1779,7 @@ ff_layout_read_pagelist(struct nfs_pgio_header *hdr)
 		goto out_failed;
 
 	ds_cred = ff_layout_get_ds_cred(lseg, idx, hdr->cred);
-	if (IS_ERR(ds_cred))
+	if (!ds_cred)
 		goto out_failed;
 
 	vers = nfs4_ff_layout_ds_version(lseg, idx);
@@ -1720,6 +1787,7 @@ ff_layout_read_pagelist(struct nfs_pgio_header *hdr)
 	dprintk("%s USE DS: %s cl_count %d vers %d\n", __func__,
 		ds->ds_remotestr, atomic_read(&ds->ds_clp->cl_count), vers);
 
+	hdr->pgio_done_cb = ff_layout_read_done_cb;
 	atomic_inc(&ds->ds_clp->cl_count);
 	hdr->ds_clp = ds->ds_clp;
 	fh = nfs4_ff_layout_select_ds_fh(lseg, idx);
@@ -1737,11 +1805,11 @@ ff_layout_read_pagelist(struct nfs_pgio_header *hdr)
 			  vers == 3 ? &ff_layout_read_call_ops_v3 :
 				      &ff_layout_read_call_ops_v4,
 			  0, RPC_TASK_SOFTCONN);
-
+	put_rpccred(ds_cred);
 	return PNFS_ATTEMPTED;
 
 out_failed:
-	if (ff_layout_has_available_ds(lseg))
+	if (ff_layout_avoid_mds_available_ds(lseg))
 		return PNFS_TRY_AGAIN;
 	return PNFS_NOT_ATTEMPTED;
 }
@@ -1769,7 +1837,7 @@ ff_layout_write_pagelist(struct nfs_pgio_header *hdr, int sync)
 		return PNFS_NOT_ATTEMPTED;
 
 	ds_cred = ff_layout_get_ds_cred(lseg, idx, hdr->cred);
-	if (IS_ERR(ds_cred))
+	if (!ds_cred)
 		return PNFS_NOT_ATTEMPTED;
 
 	vers = nfs4_ff_layout_ds_version(lseg, idx);
@@ -1798,6 +1866,7 @@ ff_layout_write_pagelist(struct nfs_pgio_header *hdr, int sync)
 			  vers == 3 ? &ff_layout_write_call_ops_v3 :
 				      &ff_layout_write_call_ops_v4,
 			  sync, RPC_TASK_SOFTCONN);
+	put_rpccred(ds_cred);
 	return PNFS_ATTEMPTED;
 }
 
@@ -1824,7 +1893,7 @@ static int ff_layout_initiate_commit(struct nfs_commit_data *data, int how)
 	struct rpc_clnt *ds_clnt;
 	struct rpc_cred *ds_cred;
 	u32 idx;
-	int vers;
+	int vers, ret;
 	struct nfs_fh *fh;
 
 	idx = calc_ds_index_from_commit(lseg, data->ds_commit_index);
@@ -1838,7 +1907,7 @@ static int ff_layout_initiate_commit(struct nfs_commit_data *data, int how)
 		goto out_err;
 
 	ds_cred = ff_layout_get_ds_cred(lseg, idx, data->cred);
-	if (IS_ERR(ds_cred))
+	if (!ds_cred)
 		goto out_err;
 
 	vers = nfs4_ff_layout_ds_version(lseg, idx);
@@ -1854,10 +1923,12 @@ static int ff_layout_initiate_commit(struct nfs_commit_data *data, int how)
 	if (fh)
 		data->args.fh = fh;
 
-	return nfs_initiate_commit(ds_clnt, data, ds->ds_clp->rpc_ops,
+	ret = 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,
 				   how, RPC_TASK_SOFTCONN);
+	put_rpccred(ds_cred);
+	return ret;
 out_err:
 	pnfs_generic_prepare_to_resend_writes(data);
 	pnfs_generic_commit_release(data);
@@ -2223,6 +2294,11 @@ static int __init nfs4flexfilelayout_init(void)
 {
 	printk(KERN_INFO "%s: NFSv4 Flexfile Layout Driver Registering...\n",
 	       __func__);
+	if (!ff_zero_group) {
+		ff_zero_group = groups_alloc(0);
+		if (!ff_zero_group)
+			return -ENOMEM;
+	}
 	return pnfs_register_layoutdriver(&flexfilelayout_type);
 }
 
@@ -2231,6 +2307,10 @@ static void __exit nfs4flexfilelayout_exit(void)
 	printk(KERN_INFO "%s: NFSv4 Flexfile Layout Driver Unregistering...\n",
 	       __func__);
 	pnfs_unregister_layoutdriver(&flexfilelayout_type);
+	if (ff_zero_group) {
+		put_group_info(ff_zero_group);
+		ff_zero_group = NULL;
+	}
 }
 
 MODULE_ALIAS("nfs-layouttype4-4");
diff --git a/fs/nfs/flexfilelayout/flexfilelayout.h b/fs/nfs/flexfilelayout/flexfilelayout.h
index dd353bb7dc0a..3ee0c9fcea76 100644
--- a/fs/nfs/flexfilelayout/flexfilelayout.h
+++ b/fs/nfs/flexfilelayout/flexfilelayout.h
@@ -10,7 +10,8 @@
 #define FS_NFS_NFS4FLEXFILELAYOUT_H
 
 #define FF_FLAGS_NO_LAYOUTCOMMIT 1
-#define FF_FLAGS_NO_IO_THRU_MDS 2
+#define FF_FLAGS_NO_IO_THRU_MDS  2
+#define FF_FLAGS_NO_READ_IO      4
 
 #include "../pnfs.h"
 
@@ -76,15 +77,13 @@ struct nfs4_ff_layout_mirror {
 	u32				fh_versions_cnt;
 	struct nfs_fh			*fh_versions;
 	nfs4_stateid			stateid;
-	u32				uid;
-	u32				gid;
-	struct rpc_cred			*cred;
+	struct rpc_cred	__rcu		*ro_cred;
+	struct rpc_cred	__rcu		*rw_cred;
 	atomic_t			ref;
 	spinlock_t			lock;
 	struct nfs4_ff_layoutstat	read_stat;
 	struct nfs4_ff_layoutstat	write_stat;
 	ktime_t				start_time;
-	ktime_t				last_report_time;
 	u32				report_interval;
 };
 
@@ -101,6 +100,7 @@ struct nfs4_flexfile_layout {
 	struct pnfs_ds_commit_info commit_info;
 	struct list_head	mirrors;
 	struct list_head	error_list; /* nfs4_ff_layout_ds_err */
+	ktime_t			last_report_time; /* Layoutstat report times */
 };
 
 static inline struct nfs4_flexfile_layout *
@@ -154,6 +154,12 @@ ff_layout_no_fallback_to_mds(struct pnfs_layout_segment *lseg)
 }
 
 static inline bool
+ff_layout_no_read_on_rw(struct pnfs_layout_segment *lseg)
+{
+	return FF_LAYOUT_LSEG(lseg)->flags & FF_FLAGS_NO_READ_IO;
+}
+
+static inline bool
 ff_layout_test_devid_unavailable(struct nfs4_deviceid_node *node)
 {
 	return nfs4_test_deviceid_unavailable(node);
@@ -192,4 +198,7 @@ nfs4_ff_find_or_create_ds_client(struct pnfs_layout_segment *lseg,
 struct rpc_cred *ff_layout_get_ds_cred(struct pnfs_layout_segment *lseg,
 				       u32 ds_idx, struct rpc_cred *mdscred);
 bool ff_layout_has_available_ds(struct pnfs_layout_segment *lseg);
+bool ff_layout_avoid_mds_available_ds(struct pnfs_layout_segment *lseg);
+bool ff_layout_avoid_read_on_rw(struct pnfs_layout_segment *lseg);
+
 #endif /* FS_NFS_NFS4FLEXFILELAYOUT_H */
diff --git a/fs/nfs/flexfilelayout/flexfilelayoutdev.c b/fs/nfs/flexfilelayout/flexfilelayoutdev.c
index add0e5a70bd6..f7a3f6b05369 100644
--- a/fs/nfs/flexfilelayout/flexfilelayoutdev.c
+++ b/fs/nfs/flexfilelayout/flexfilelayoutdev.c
@@ -17,8 +17,8 @@
 
 #define NFSDBG_FACILITY		NFSDBG_PNFS_LD
 
-static unsigned int dataserver_timeo = NFS4_DEF_DS_TIMEO;
-static unsigned int dataserver_retrans = NFS4_DEF_DS_RETRANS;
+static unsigned int dataserver_timeo = NFS_DEF_TCP_RETRANS;
+static unsigned int dataserver_retrans;
 
 void nfs4_ff_layout_put_deviceid(struct nfs4_ff_layout_ds *mirror_ds)
 {
@@ -228,7 +228,8 @@ ff_ds_error_match(const struct nfs4_ff_layout_ds_err *e1,
 		return e1->opnum < e2->opnum ? -1 : 1;
 	if (e1->status != e2->status)
 		return e1->status < e2->status ? -1 : 1;
-	ret = memcmp(&e1->stateid, &e2->stateid, sizeof(e1->stateid));
+	ret = memcmp(e1->stateid.data, e2->stateid.data,
+			sizeof(e1->stateid.data));
 	if (ret != 0)
 		return ret;
 	ret = memcmp(&e1->deviceid, &e2->deviceid, sizeof(e1->deviceid));
@@ -302,40 +303,26 @@ int ff_layout_track_ds_error(struct nfs4_flexfile_layout *flo,
 	return 0;
 }
 
-/* currently we only support AUTH_NONE and AUTH_SYS */
-static rpc_authflavor_t
-nfs4_ff_layout_choose_authflavor(struct nfs4_ff_layout_mirror *mirror)
+static struct rpc_cred *
+ff_layout_get_mirror_cred(struct nfs4_ff_layout_mirror *mirror, u32 iomode)
 {
-	if (mirror->uid == (u32)-1)
-		return RPC_AUTH_NULL;
-	return RPC_AUTH_UNIX;
-}
+	struct rpc_cred *cred, __rcu **pcred;
 
-/* fetch cred for NFSv3 DS */
-static int ff_layout_update_mirror_cred(struct nfs4_ff_layout_mirror *mirror,
-				      struct nfs4_pnfs_ds *ds)
-{
-	if (ds->ds_clp && !mirror->cred &&
-	    mirror->mirror_ds->ds_versions[0].version == 3) {
-		struct rpc_auth *auth = ds->ds_clp->cl_rpcclient->cl_auth;
-		struct rpc_cred *cred;
-		struct auth_cred acred = {
-			.uid = make_kuid(&init_user_ns, mirror->uid),
-			.gid = make_kgid(&init_user_ns, mirror->gid),
-		};
-
-		/* AUTH_NULL ignores acred */
-		cred = auth->au_ops->lookup_cred(auth, &acred, 0);
-		if (IS_ERR(cred)) {
-			dprintk("%s: lookup_cred failed with %ld\n",
-				__func__, PTR_ERR(cred));
-			return PTR_ERR(cred);
-		} else {
-			if (cmpxchg(&mirror->cred, NULL, cred))
-				put_rpccred(cred);
-		}
-	}
-	return 0;
+	if (iomode == IOMODE_READ)
+		pcred = &mirror->ro_cred;
+	else
+		pcred = &mirror->rw_cred;
+
+	rcu_read_lock();
+	do {
+		cred = rcu_dereference(*pcred);
+		if (!cred)
+			break;
+
+		cred = get_rpccred_rcu(cred);
+	} while(!cred);
+	rcu_read_unlock();
+	return cred;
 }
 
 struct nfs_fh *
@@ -356,7 +343,23 @@ out:
 	return fh;
 }
 
-/* Upon return, either ds is connected, or ds is NULL */
+/**
+ * nfs4_ff_layout_prepare_ds - prepare a DS connection for an RPC call
+ * @lseg: the layout segment we're operating on
+ * @ds_idx: index of the DS to use
+ * @fail_return: return layout on connect failure?
+ *
+ * Try to prepare a DS connection to accept an RPC call. This involves
+ * selecting a mirror to use and connecting the client to it if it's not
+ * already connected.
+ *
+ * Since we only need a single functioning mirror to satisfy a read, we don't
+ * want to return the layout if there is one. For writes though, any down
+ * mirror should result in a LAYOUTRETURN. @fail_return is how we distinguish
+ * between the two cases.
+ *
+ * Returns a pointer to a connected DS object on success or NULL on failure.
+ */
 struct nfs4_pnfs_ds *
 nfs4_ff_layout_prepare_ds(struct pnfs_layout_segment *lseg, u32 ds_idx,
 			  bool fail_return)
@@ -367,7 +370,6 @@ nfs4_ff_layout_prepare_ds(struct pnfs_layout_segment *lseg, u32 ds_idx,
 	struct inode *ino = lseg->pls_layout->plh_inode;
 	struct nfs_server *s = NFS_SERVER(ino);
 	unsigned int max_payload;
-	rpc_authflavor_t flavor;
 
 	if (!ff_layout_mirror_valid(lseg, mirror)) {
 		pr_err_ratelimited("NFS: %s: No data server for offset index %d\n",
@@ -377,15 +379,13 @@ nfs4_ff_layout_prepare_ds(struct pnfs_layout_segment *lseg, u32 ds_idx,
 
 	devid = &mirror->mirror_ds->id_node;
 	if (ff_layout_test_devid_unavailable(devid))
-		goto out;
+		goto out_fail;
 
 	ds = mirror->mirror_ds->ds;
 	/* matching smp_wmb() in _nfs4_pnfs_v3/4_ds_connect */
 	smp_rmb();
 	if (ds->ds_clp)
-		goto out_update_creds;
-
-	flavor = nfs4_ff_layout_choose_authflavor(mirror);
+		goto out;
 
 	/* FIXME: For now we assume the server sent only one version of NFS
 	 * to use for the DS.
@@ -394,7 +394,7 @@ nfs4_ff_layout_prepare_ds(struct pnfs_layout_segment *lseg, u32 ds_idx,
 			     dataserver_retrans,
 			     mirror->mirror_ds->ds_versions[0].version,
 			     mirror->mirror_ds->ds_versions[0].minor_version,
-			     flavor);
+			     RPC_AUTH_UNIX);
 
 	/* connect success, check rsize/wsize limit */
 	if (ds->ds_clp) {
@@ -405,25 +405,16 @@ nfs4_ff_layout_prepare_ds(struct pnfs_layout_segment *lseg, u32 ds_idx,
 			mirror->mirror_ds->ds_versions[0].rsize = max_payload;
 		if (mirror->mirror_ds->ds_versions[0].wsize > max_payload)
 			mirror->mirror_ds->ds_versions[0].wsize = max_payload;
-	} else {
-		ff_layout_track_ds_error(FF_LAYOUT_FROM_HDR(lseg->pls_layout),
-					 mirror, lseg->pls_range.offset,
-					 lseg->pls_range.length, NFS4ERR_NXIO,
-					 OP_ILLEGAL, GFP_NOIO);
-		if (!fail_return) {
-			if (ff_layout_has_available_ds(lseg))
-				set_bit(NFS_LAYOUT_RETURN_REQUESTED,
-					&lseg->pls_layout->plh_flags);
-			else
-				pnfs_error_mark_layout_for_return(ino, lseg);
-		} else
-			pnfs_error_mark_layout_for_return(ino, lseg);
-		ds = NULL;
 		goto out;
 	}
-out_update_creds:
-	if (ff_layout_update_mirror_cred(mirror, ds))
-		ds = NULL;
+	ff_layout_track_ds_error(FF_LAYOUT_FROM_HDR(lseg->pls_layout),
+				 mirror, lseg->pls_range.offset,
+				 lseg->pls_range.length, NFS4ERR_NXIO,
+				 OP_ILLEGAL, GFP_NOIO);
+out_fail:
+	if (fail_return || !ff_layout_has_available_ds(lseg))
+		pnfs_error_mark_layout_for_return(ino, lseg);
+	ds = NULL;
 out:
 	return ds;
 }
@@ -433,16 +424,15 @@ ff_layout_get_ds_cred(struct pnfs_layout_segment *lseg, u32 ds_idx,
 		      struct rpc_cred *mdscred)
 {
 	struct nfs4_ff_layout_mirror *mirror = FF_LAYOUT_COMP(lseg, ds_idx);
-	struct rpc_cred *cred = ERR_PTR(-EINVAL);
-
-	if (!nfs4_ff_layout_prepare_ds(lseg, ds_idx, true))
-		goto out;
+	struct rpc_cred *cred;
 
-	if (mirror && mirror->cred)
-		cred = mirror->cred;
-	else
-		cred = mdscred;
-out:
+	if (mirror) {
+		cred = ff_layout_get_mirror_cred(mirror, lseg->pls_range.iomode);
+		if (!cred)
+			cred = get_rpccred(mdscred);
+	} else {
+		cred = get_rpccred(mdscred);
+	}
 	return cred;
 }
 
@@ -562,6 +552,18 @@ bool ff_layout_has_available_ds(struct pnfs_layout_segment *lseg)
 	return ff_rw_layout_has_available_ds(lseg);
 }
 
+bool ff_layout_avoid_mds_available_ds(struct pnfs_layout_segment *lseg)
+{
+	return ff_layout_no_fallback_to_mds(lseg) ||
+	       ff_layout_has_available_ds(lseg);
+}
+
+bool ff_layout_avoid_read_on_rw(struct pnfs_layout_segment *lseg)
+{
+	return lseg->pls_range.iomode == IOMODE_RW &&
+	       ff_layout_no_read_on_rw(lseg);
+}
+
 module_param(dataserver_retrans, uint, 0644);
 MODULE_PARM_DESC(dataserver_retrans, "The  number of times the NFSv4.1 client "
 			"retries a request before it attempts further "