Merge tag 'nfs-for-5.20-1' of git://git.linux-nfs.org/projects/trondmy/linux-nfs

Pull NFS client updates from Trond Myklebust:
 "Highlights include:

  Stable fixes:
   - pNFS/flexfiles: Fix infinite looping when the RDMA connection
     errors out

  Bugfixes:
   - NFS: fix port value parsing
   - SUNRPC: Reinitialise the backchannel request buffers before reuse
   - SUNRPC: fix expiry of auth creds
   - NFSv4: Fix races in the legacy idmapper upcall
   - NFS: O_DIRECT fixes from Jeff Layton
   - NFSv4.1: Fix OP_SEQUENCE error handling
   - SUNRPC: Fix an RPC/RDMA performance regression
   - NFS: Fix case insensitive renames
   - NFSv4/pnfs: Fix a use-after-free bug in open
   - NFSv4.1: RECLAIM_COMPLETE must handle EACCES

  Features:
   - NFSv4.1: session trunking enhancements
   - NFSv4.2: READ_PLUS performance optimisations
   - NFS: relax the rules for rsize/wsize mount options
   - NFS: don't unhash dentry during unlink/rename
   - SUNRPC: Fail faster on bad verifier
   - NFS/SUNRPC: Various tracing improvements"

* tag 'nfs-for-5.20-1' of git://git.linux-nfs.org/projects/trondmy/linux-nfs: (46 commits)
  NFS: Improve readpage/writepage tracing
  NFS: Improve O_DIRECT tracing
  NFS: Improve write error tracing
  NFS: don't unhash dentry during unlink/rename
  NFSv4/pnfs: Fix a use-after-free bug in open
  NFS: nfs_async_write_reschedule_io must not recurse into the writeback code
  SUNRPC: Don't reuse bvec on retransmission of the request
  SUNRPC: Reinitialise the backchannel request buffers before reuse
  NFSv4.1: RECLAIM_COMPLETE must handle EACCES
  NFSv4.1 probe offline transports for trunking on session creation
  SUNRPC create a function that probes only offline transports
  SUNRPC export xprt_iter_rewind function
  SUNRPC restructure rpc_clnt_setup_test_and_add_xprt
  NFSv4.1 remove xprt from xprt_switch if session trunking test fails
  SUNRPC create an rpc function that allows xprt removal from rpc_clnt
  SUNRPC enable back offline transports in trunking discovery
  SUNRPC create an iterator to list only OFFLINE xprts
  NFSv4.1 offline trunkable transports on DESTROY_SESSION
  SUNRPC add function to offline remove trunkable transports
  SUNRPC expose functions for offline remote xprt functionality
  ...

17 files changed, 513 insertions, 289 deletions

diff --git a/fs/nfs/blocklayout/dev.c b/fs/nfs/blocklayout/dev.c
index 5e56da748b2a..fea5f8821da5 100644
--- a/fs/nfs/blocklayout/dev.c
+++ b/fs/nfs/blocklayout/dev.c
@@ -301,18 +301,14 @@ bl_validate_designator(struct pnfs_block_volume *v)
 	}
 }
 
-/*
- * Try to open the udev path for the WWN.  At least on Debian the udev
- * by-id path will always point to the dm-multipath device if one exists.
- */
 static struct block_device *
-bl_open_udev_path(struct pnfs_block_volume *v)
+bl_open_path(struct pnfs_block_volume *v, const char *prefix)
 {
 	struct block_device *bdev;
 	const char *devname;
 
-	devname = kasprintf(GFP_KERNEL, "/dev/disk/by-id/wwn-0x%*phN",
-				v->scsi.designator_len, v->scsi.designator);
+	devname = kasprintf(GFP_KERNEL, "/dev/disk/by-id/%s%*phN",
+			prefix, v->scsi.designator_len, v->scsi.designator);
 	if (!devname)
 		return ERR_PTR(-ENOMEM);
 
@@ -326,28 +322,6 @@ bl_open_udev_path(struct pnfs_block_volume *v)
 	return bdev;
 }
 
-/*
- * Try to open the RH/Fedora specific dm-mpath udev path for this WWN, as the
- * wwn- links will only point to the first discovered SCSI device there.
- */
-static struct block_device *
-bl_open_dm_mpath_udev_path(struct pnfs_block_volume *v)
-{
-	struct block_device *bdev;
-	const char *devname;
-
-	devname = kasprintf(GFP_KERNEL,
-			"/dev/disk/by-id/dm-uuid-mpath-%d%*phN",
-			v->scsi.designator_type,
-			v->scsi.designator_len, v->scsi.designator);
-	if (!devname)
-		return ERR_PTR(-ENOMEM);
-
-	bdev = blkdev_get_by_path(devname, FMODE_READ | FMODE_WRITE, NULL);
-	kfree(devname);
-	return bdev;
-}
-
 static int
 bl_parse_scsi(struct nfs_server *server, struct pnfs_block_dev *d,
 		struct pnfs_block_volume *volumes, int idx, gfp_t gfp_mask)
@@ -360,9 +334,15 @@ bl_parse_scsi(struct nfs_server *server, struct pnfs_block_dev *d,
 	if (!bl_validate_designator(v))
 		return -EINVAL;
 
-	bdev = bl_open_dm_mpath_udev_path(v);
+	/*
+	 * Try to open the RH/Fedora specific dm-mpath udev path first, as the
+	 * wwn- links will only point to the first discovered SCSI device there.
+	 * On other distributions like Debian, the default SCSI by-id path will
+	 * point to the dm-multipath device if one exists.
+	 */
+	bdev = bl_open_path(v, "dm-uuid-mpath-0x");
 	if (IS_ERR(bdev))
-		bdev = bl_open_udev_path(v);
+		bdev = bl_open_path(v, "wwn-0x");
 	if (IS_ERR(bdev))
 		return PTR_ERR(bdev);
 	d->bdev = bdev;
diff --git a/fs/nfs/client.c b/fs/nfs/client.c
index e828504cc396..da8da5cdbbc1 100644
--- a/fs/nfs/client.c
+++ b/fs/nfs/client.c
@@ -708,9 +708,9 @@ static int nfs_init_server(struct nfs_server *server,
 	}
 
 	if (ctx->rsize)
-		server->rsize = nfs_block_size(ctx->rsize, NULL);
+		server->rsize = nfs_io_size(ctx->rsize, clp->cl_proto);
 	if (ctx->wsize)
-		server->wsize = nfs_block_size(ctx->wsize, NULL);
+		server->wsize = nfs_io_size(ctx->wsize, clp->cl_proto);
 
 	server->acregmin = ctx->acregmin * HZ;
 	server->acregmax = ctx->acregmax * HZ;
@@ -755,18 +755,19 @@ error:
 static void nfs_server_set_fsinfo(struct nfs_server *server,
 				  struct nfs_fsinfo *fsinfo)
 {
+	struct nfs_client *clp = server->nfs_client;
 	unsigned long max_rpc_payload, raw_max_rpc_payload;
 
 	/* Work out a lot of parameters */
 	if (server->rsize == 0)
-		server->rsize = nfs_block_size(fsinfo->rtpref, NULL);
+		server->rsize = nfs_io_size(fsinfo->rtpref, clp->cl_proto);
 	if (server->wsize == 0)
-		server->wsize = nfs_block_size(fsinfo->wtpref, NULL);
+		server->wsize = nfs_io_size(fsinfo->wtpref, clp->cl_proto);
 
 	if (fsinfo->rtmax >= 512 && server->rsize > fsinfo->rtmax)
-		server->rsize = nfs_block_size(fsinfo->rtmax, NULL);
+		server->rsize = nfs_io_size(fsinfo->rtmax, clp->cl_proto);
 	if (fsinfo->wtmax >= 512 && server->wsize > fsinfo->wtmax)
-		server->wsize = nfs_block_size(fsinfo->wtmax, NULL);
+		server->wsize = nfs_io_size(fsinfo->wtmax, clp->cl_proto);
 
 	raw_max_rpc_payload = rpc_max_payload(server->client);
 	max_rpc_payload = nfs_block_size(raw_max_rpc_payload, NULL);
diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c
index 0c4e8dd6aa96..dbab3caa15ed 100644
--- a/fs/nfs/dir.c
+++ b/fs/nfs/dir.c
@@ -1084,7 +1084,7 @@ static void nfs_do_filldir(struct nfs_readdir_descriptor *desc,
 	struct nfs_cache_array *array;
 	unsigned int i;
 
-	array = kmap(desc->page);
+	array = kmap_local_page(desc->page);
 	for (i = desc->cache_entry_index; i < array->size; i++) {
 		struct nfs_cache_array_entry *ent;
 
@@ -1110,7 +1110,7 @@ static void nfs_do_filldir(struct nfs_readdir_descriptor *desc,
 	if (array->page_is_eof)
 		desc->eof = !desc->eob;
 
-	kunmap(desc->page);
+	kunmap_local(array);
 	dfprintk(DIRCACHE, "NFS: nfs_do_filldir() filling ended @ cookie %llu\n",
 			(unsigned long long)desc->dir_cookie);
 }
@@ -1739,6 +1739,10 @@ nfs_do_lookup_revalidate(struct inode *dir, struct dentry *dentry,
 		goto out_bad;
 	}
 
+	if ((flags & LOOKUP_RENAME_TARGET) && d_count(dentry) < 2 &&
+	    nfs_server_capable(dir, NFS_CAP_CASE_INSENSITIVE))
+		goto out_bad;
+
 	if (nfs_verifier_is_delegated(dentry))
 		return nfs_lookup_revalidate_delegated(dir, dentry, inode);
 
@@ -1778,6 +1782,8 @@ __nfs_lookup_revalidate(struct dentry *dentry, unsigned int flags,
 	int ret;
 
 	if (flags & LOOKUP_RCU) {
+		if (dentry->d_fsdata == NFS_FSDATA_BLOCKED)
+			return -ECHILD;
 		parent = READ_ONCE(dentry->d_parent);
 		dir = d_inode_rcu(parent);
 		if (!dir)
@@ -1786,6 +1792,9 @@ __nfs_lookup_revalidate(struct dentry *dentry, unsigned int flags,
 		if (parent != READ_ONCE(dentry->d_parent))
 			return -ECHILD;
 	} else {
+		/* Wait for unlink to complete */
+		wait_var_event(&dentry->d_fsdata,
+			       dentry->d_fsdata != NFS_FSDATA_BLOCKED);
 		parent = dget_parent(dentry);
 		ret = reval(d_inode(parent), dentry, flags);
 		dput(parent);
@@ -2454,7 +2463,6 @@ out:
 int nfs_unlink(struct inode *dir, struct dentry *dentry)
 {
 	int error;
-	int need_rehash = 0;
 
 	dfprintk(VFS, "NFS: unlink(%s/%lu, %pd)\n", dir->i_sb->s_id,
 		dir->i_ino, dentry);
@@ -2469,15 +2477,25 @@ int nfs_unlink(struct inode *dir, struct dentry *dentry)
 		error = nfs_sillyrename(dir, dentry);
 		goto out;
 	}
-	if (!d_unhashed(dentry)) {
-		__d_drop(dentry);
-		need_rehash = 1;
-	}
+	/* We must prevent any concurrent open until the unlink
+	 * completes.  ->d_revalidate will wait for ->d_fsdata
+	 * to clear.  We set it here to ensure no lookup succeeds until
+	 * the unlink is complete on the server.
+	 */
+	error = -ETXTBSY;
+	if (WARN_ON(dentry->d_flags & DCACHE_NFSFS_RENAMED) ||
+	    WARN_ON(dentry->d_fsdata == NFS_FSDATA_BLOCKED))
+		goto out;
+	if (dentry->d_fsdata)
+		/* old devname */
+		kfree(dentry->d_fsdata);
+	dentry->d_fsdata = NFS_FSDATA_BLOCKED;
+
 	spin_unlock(&dentry->d_lock);
 	error = nfs_safe_remove(dentry);
 	nfs_dentry_remove_handle_error(dir, dentry, error);
-	if (need_rehash)
-		d_rehash(dentry);
+	dentry->d_fsdata = NULL;
+	wake_up_var(&dentry->d_fsdata);
 out:
 	trace_nfs_unlink_exit(dir, dentry, error);
 	return error;
@@ -2584,6 +2602,15 @@ nfs_link(struct dentry *old_dentry, struct inode *dir, struct dentry *dentry)
 }
 EXPORT_SYMBOL_GPL(nfs_link);
 
+static void
+nfs_unblock_rename(struct rpc_task *task, struct nfs_renamedata *data)
+{
+	struct dentry *new_dentry = data->new_dentry;
+
+	new_dentry->d_fsdata = NULL;
+	wake_up_var(&new_dentry->d_fsdata);
+}
+
 /*
  * RENAME
  * FIXME: Some nfsds, like the Linux user space nfsd, may generate a
@@ -2614,8 +2641,9 @@ int nfs_rename(struct user_namespace *mnt_userns, struct inode *old_dir,
 {
 	struct inode *old_inode = d_inode(old_dentry);
 	struct inode *new_inode = d_inode(new_dentry);
-	struct dentry *dentry = NULL, *rehash = NULL;
+	struct dentry *dentry = NULL;
 	struct rpc_task *task;
+	bool must_unblock = false;
 	int error = -EBUSY;
 
 	if (flags)
@@ -2633,18 +2661,27 @@ int nfs_rename(struct user_namespace *mnt_userns, struct inode *old_dir,
 	 * the new target.
 	 */
 	if (new_inode && !S_ISDIR(new_inode->i_mode)) {
-		/*
-		 * To prevent any new references to the target during the
-		 * rename, we unhash the dentry in advance.
+		/* We must prevent any concurrent open until the unlink
+		 * completes.  ->d_revalidate will wait for ->d_fsdata
+		 * to clear.  We set it here to ensure no lookup succeeds until
+		 * the unlink is complete on the server.
 		 */
-		if (!d_unhashed(new_dentry)) {
-			d_drop(new_dentry);
-			rehash = new_dentry;
+		error = -ETXTBSY;
+		if (WARN_ON(new_dentry->d_flags & DCACHE_NFSFS_RENAMED) ||
+		    WARN_ON(new_dentry->d_fsdata == NFS_FSDATA_BLOCKED))
+			goto out;
+		if (new_dentry->d_fsdata) {
+			/* old devname */
+			kfree(new_dentry->d_fsdata);
+			new_dentry->d_fsdata = NULL;
 		}
 
+		spin_lock(&new_dentry->d_lock);
 		if (d_count(new_dentry) > 2) {
 			int err;
 
+			spin_unlock(&new_dentry->d_lock);
+
 			/* copy the target dentry's name */
 			dentry = d_alloc(new_dentry->d_parent,
 					 &new_dentry->d_name);
@@ -2657,14 +2694,19 @@ int nfs_rename(struct user_namespace *mnt_userns, struct inode *old_dir,
 				goto out;
 
 			new_dentry = dentry;
-			rehash = NULL;
 			new_inode = NULL;
+		} else {
+			new_dentry->d_fsdata = NFS_FSDATA_BLOCKED;
+			must_unblock = true;
+			spin_unlock(&new_dentry->d_lock);
 		}
+
 	}
 
 	if (S_ISREG(old_inode->i_mode))
 		nfs_sync_inode(old_inode);
-	task = nfs_async_rename(old_dir, new_dir, old_dentry, new_dentry, NULL);
+	task = nfs_async_rename(old_dir, new_dir, old_dentry, new_dentry,
+				must_unblock ? nfs_unblock_rename : NULL);
 	if (IS_ERR(task)) {
 		error = PTR_ERR(task);
 		goto out;
@@ -2688,8 +2730,6 @@ int nfs_rename(struct user_namespace *mnt_userns, struct inode *old_dir,
 		spin_unlock(&old_inode->i_lock);
 	}
 out:
-	if (rehash)
-		d_rehash(rehash);
 	trace_nfs_rename_exit(old_dir, old_dentry,
 			new_dir, new_dentry, error);
 	if (!error) {
diff --git a/fs/nfs/direct.c b/fs/nfs/direct.c
index c275c83f0aef..1707f46b1335 100644
--- a/fs/nfs/direct.c
+++ b/fs/nfs/direct.c
@@ -60,44 +60,12 @@
 #include "iostat.h"
 #include "pnfs.h"
 #include "fscache.h"
+#include "nfstrace.h"
 
 #define NFSDBG_FACILITY		NFSDBG_VFS
 
 static struct kmem_cache *nfs_direct_cachep;
 
-struct nfs_direct_req {
-	struct kref		kref;		/* release manager */
-
-	/* I/O parameters */
-	struct nfs_open_context	*ctx;		/* file open context info */
-	struct nfs_lock_context *l_ctx;		/* Lock context info */
-	struct kiocb *		iocb;		/* controlling i/o request */
-	struct inode *		inode;		/* target file of i/o */
-
-	/* completion state */
-	atomic_t		io_count;	/* i/os we're waiting for */
-	spinlock_t		lock;		/* protect completion state */
-
-	loff_t			io_start;	/* Start offset for I/O */
-	ssize_t			count,		/* bytes actually processed */
-				max_count,	/* max expected count */
-				bytes_left,	/* bytes left to be sent */
-				error;		/* any reported error */
-	struct completion	completion;	/* wait for i/o completion */
-
-	/* commit state */
-	struct nfs_mds_commit_info mds_cinfo;	/* Storage for cinfo */
-	struct pnfs_ds_commit_info ds_cinfo;	/* Storage for cinfo */
-	struct work_struct	work;
-	int			flags;
-	/* for write */
-#define NFS_ODIRECT_DO_COMMIT		(1)	/* an unstable reply was received */
-#define NFS_ODIRECT_RESCHED_WRITES	(2)	/* write verification failed */
-	/* for read */
-#define NFS_ODIRECT_SHOULD_DIRTY	(3)	/* dirty user-space page after read */
-#define NFS_ODIRECT_DONE		INT_MAX	/* write verification failed */
-};
-
 static const struct nfs_pgio_completion_ops nfs_direct_write_completion_ops;
 static const struct nfs_commit_completion_ops nfs_direct_commit_completion_ops;
 static void nfs_direct_write_complete(struct nfs_direct_req *dreq);
@@ -594,14 +562,17 @@ static void nfs_direct_commit_complete(struct nfs_commit_data *data)
 	struct nfs_page *req;
 	int status = data->task.tk_status;
 
+	trace_nfs_direct_commit_complete(dreq);
+
 	if (status < 0) {
 		/* Errors in commit are fatal */
 		dreq->error = status;
 		dreq->max_count = 0;
 		dreq->count = 0;
 		dreq->flags = NFS_ODIRECT_DONE;
-	} else if (dreq->flags == NFS_ODIRECT_DONE)
+	} else {
 		status = dreq->error;
+	}
 
 	nfs_init_cinfo_from_dreq(&cinfo, dreq);
 
@@ -630,6 +601,8 @@ static void nfs_direct_resched_write(struct nfs_commit_info *cinfo,
 {
 	struct nfs_direct_req *dreq = cinfo->dreq;
 
+	trace_nfs_direct_resched_write(dreq);
+
 	spin_lock(&dreq->lock);
 	if (dreq->flags != NFS_ODIRECT_DONE)
 		dreq->flags = NFS_ODIRECT_RESCHED_WRITES;
@@ -694,6 +667,7 @@ static void nfs_direct_write_schedule_work(struct work_struct *work)
 
 static void nfs_direct_write_complete(struct nfs_direct_req *dreq)
 {
+	trace_nfs_direct_write_complete(dreq);
 	queue_work(nfsiod_workqueue, &dreq->work); /* Calls nfs_direct_write_schedule_work */
 }
 
@@ -704,6 +678,8 @@ static void nfs_direct_write_completion(struct nfs_pgio_header *hdr)
 	struct nfs_page *req = nfs_list_entry(hdr->pages.next);
 	int flags = NFS_ODIRECT_DONE;
 
+	trace_nfs_direct_write_completion(dreq);
+
 	nfs_init_cinfo_from_dreq(&cinfo, dreq);
 
 	spin_lock(&dreq->lock);
@@ -713,7 +689,7 @@ static void nfs_direct_write_completion(struct nfs_pgio_header *hdr)
 	}
 
 	nfs_direct_count_bytes(dreq, hdr);
-	if (hdr->good_bytes != 0 && nfs_write_need_commit(hdr)) {
+	if (test_bit(NFS_IOHDR_UNSTABLE_WRITES, &hdr->flags)) {
 		if (!dreq->flags)
 			dreq->flags = NFS_ODIRECT_DO_COMMIT;
 		flags = dreq->flags;
@@ -758,6 +734,8 @@ static void nfs_direct_write_reschedule_io(struct nfs_pgio_header *hdr)
 {
 	struct nfs_direct_req *dreq = hdr->dreq;
 
+	trace_nfs_direct_write_reschedule_io(dreq);
+
 	spin_lock(&dreq->lock);
 	if (dreq->error == 0) {
 		dreq->flags = NFS_ODIRECT_RESCHED_WRITES;
@@ -798,6 +776,8 @@ static ssize_t nfs_direct_write_schedule_iovec(struct nfs_direct_req *dreq,
 	size_t requested_bytes = 0;
 	size_t wsize = max_t(size_t, NFS_SERVER(inode)->wsize, PAGE_SIZE);
 
+	trace_nfs_direct_write_schedule_iovec(dreq);
+
 	nfs_pageio_init_write(&desc, inode, ioflags, false,
 			      &nfs_direct_write_completion_ops);
 	desc.pg_dreq = dreq;
diff --git a/fs/nfs/file.c b/fs/nfs/file.c
index 549baed76351..d2bcd4834c0e 100644
--- a/fs/nfs/file.c
+++ b/fs/nfs/file.c
@@ -661,8 +661,6 @@ ssize_t nfs_file_write(struct kiocb *iocb, struct iov_iter *from)
 		result = filemap_fdatawait_range(file->f_mapping,
 						 iocb->ki_pos - written,
 						 iocb->ki_pos - 1);
-		if (result < 0)
-			goto out;
 	}
 	result = generic_write_sync(iocb, written);
 	if (result < 0)
diff --git a/fs/nfs/filelayout/filelayout.c b/fs/nfs/filelayout/filelayout.c
index 2b2661582bbe..ad34a33b0737 100644
--- a/fs/nfs/filelayout/filelayout.c
+++ b/fs/nfs/filelayout/filelayout.c
@@ -181,6 +181,8 @@ static int filelayout_async_handle_error(struct rpc_task *task,
 	case -EIO:
 	case -ETIMEDOUT:
 	case -EPIPE:
+	case -EPROTO:
+	case -ENODEV:
 		dprintk("%s DS connection error %d\n", __func__,
 			task->tk_status);
 		nfs4_mark_deviceid_unavailable(devid);
diff --git a/fs/nfs/flexfilelayout/flexfilelayout.c b/fs/nfs/flexfilelayout/flexfilelayout.c
index 604be402ae13..7d285561e59f 100644
--- a/fs/nfs/flexfilelayout/flexfilelayout.c
+++ b/fs/nfs/flexfilelayout/flexfilelayout.c
@@ -1131,6 +1131,8 @@ static int ff_layout_async_handle_error_v4(struct rpc_task *task,
 	case -EIO:
 	case -ETIMEDOUT:
 	case -EPIPE:
+	case -EPROTO:
+	case -ENODEV:
 		dprintk("%s DS connection error %d\n", __func__,
 			task->tk_status);
 		nfs4_delete_deviceid(devid->ld, devid->nfs_client,
@@ -1236,6 +1238,8 @@ static void ff_layout_io_track_ds_error(struct pnfs_layout_segment *lseg,
 		case -ENOBUFS:
 		case -EPIPE:
 		case -EPERM:
+		case -EPROTO:
+		case -ENODEV:
 			*op_status = status = NFS4ERR_NXIO;
 			break;
 		case -EACCES:
diff --git a/fs/nfs/flexfilelayout/flexfilelayoutdev.c b/fs/nfs/flexfilelayout/flexfilelayoutdev.c
index bfa7202ca7be..e028f5a0ef5f 100644
--- a/fs/nfs/flexfilelayout/flexfilelayoutdev.c
+++ b/fs/nfs/flexfilelayout/flexfilelayoutdev.c
@@ -113,8 +113,10 @@ nfs4_ff_alloc_deviceid_node(struct nfs_server *server, struct pnfs_device *pdev,
 			goto out_err_drain_dsaddrs;
 		ds_versions[i].version = be32_to_cpup(p++);
 		ds_versions[i].minor_version = be32_to_cpup(p++);
-		ds_versions[i].rsize = nfs_block_size(be32_to_cpup(p++), NULL);
-		ds_versions[i].wsize = nfs_block_size(be32_to_cpup(p++), NULL);
+		ds_versions[i].rsize = nfs_io_size(be32_to_cpup(p++),
+						   server->nfs_client->cl_proto);
+		ds_versions[i].wsize = nfs_io_size(be32_to_cpup(p++),
+						   server->nfs_client->cl_proto);
 		ds_versions[i].tightly_coupled = be32_to_cpup(p);
 
 		if (ds_versions[i].rsize > NFS_MAX_FILE_IO_SIZE)
diff --git a/fs/nfs/fs_context.c b/fs/nfs/fs_context.c
index 9a16897e8dc6..4da701fd1424 100644
--- a/fs/nfs/fs_context.c
+++ b/fs/nfs/fs_context.c
@@ -21,6 +21,8 @@
 #include "nfs.h"
 #include "internal.h"
 
+#include "nfstrace.h"
+
 #define NFSDBG_FACILITY		NFSDBG_MOUNT
 
 #if IS_ENABLED(CONFIG_NFS_V3)
@@ -284,7 +286,6 @@ static int nfs_verify_server_address(struct sockaddr *addr)
 	}
 	}
 
-	dfprintk(MOUNT, "NFS: Invalid IP address specified\n");
 	return 0;
 }
 
@@ -378,7 +379,7 @@ static int nfs_parse_security_flavors(struct fs_context *fc,
 	char *string = param->string, *p;
 	int ret;
 
-	dfprintk(MOUNT, "NFS: parsing %s=%s option\n", param->key, param->string);
+	trace_nfs_mount_assign(param->key, string);
 
 	while ((p = strsep(&string, ":")) != NULL) {
 		if (!*p)
@@ -480,11 +481,11 @@ static int nfs_fs_context_parse_param(struct fs_context *fc,
 	unsigned int len;
 	int ret, opt;
 
-	dfprintk(MOUNT, "NFS:   parsing nfs mount option '%s'\n", param->key);
+	trace_nfs_mount_option(param);
 
 	opt = fs_parse(fc, nfs_fs_parameters, param, &result);
 	if (opt < 0)
-		return ctx->sloppy ? 1 : opt;
+		return (opt == -ENOPARAM && ctx->sloppy) ? 1 : opt;
 
 	if (fc->security)
 		ctx->has_sec_mnt_opts = 1;
@@ -683,6 +684,7 @@ static int nfs_fs_context_parse_param(struct fs_context *fc,
 			return ret;
 		break;
 	case Opt_vers:
+		trace_nfs_mount_assign(param->key, param->string);
 		ret = nfs_parse_version_string(fc, param->string);
 		if (ret < 0)
 			return ret;
@@ -694,6 +696,7 @@ static int nfs_fs_context_parse_param(struct fs_context *fc,
 		break;
 
 	case Opt_proto:
+		trace_nfs_mount_assign(param->key, param->string);
 		protofamily = AF_INET;
 		switch (lookup_constant(nfs_xprt_protocol_tokens, param->string, -1)) {
 		case Opt_xprt_udp6:
@@ -729,6 +732,7 @@ static int nfs_fs_context_parse_param(struct fs_context *fc,
 		break;
 
 	case Opt_mountproto:
+		trace_nfs_mount_assign(param->key, param->string);
 		mountfamily = AF_INET;
 		switch (lookup_constant(nfs_xprt_protocol_tokens, param->string, -1)) {
 		case Opt_xprt_udp6:
@@ -751,6 +755,7 @@ static int nfs_fs_context_parse_param(struct fs_context *fc,
 		break;
 
 	case Opt_addr:
+		trace_nfs_mount_assign(param->key, param->string);
 		len = rpc_pton(fc->net_ns, param->string, param->size,
 			       &ctx->nfs_server.address,
 			       sizeof(ctx->nfs_server._address));
@@ -759,16 +764,19 @@ static int nfs_fs_context_parse_param(struct fs_context *fc,
 		ctx->nfs_server.addrlen = len;
 		break;
 	case Opt_clientaddr:
+		trace_nfs_mount_assign(param->key, param->string);
 		kfree(ctx->client_address);
 		ctx->client_address = param->string;
 		param->string = NULL;
 		break;
 	case Opt_mounthost:
+		trace_nfs_mount_assign(param->key, param->string);
 		kfree(ctx->mount_server.hostname);
 		ctx->mount_server.hostname = param->string;
 		param->string = NULL;
 		break;
 	case Opt_mountaddr:
+		trace_nfs_mount_assign(param->key, param->string);
 		len = rpc_pton(fc->net_ns, param->string, param->size,
 			       &ctx->mount_server.address,
 			       sizeof(ctx->mount_server._address));
@@ -846,7 +854,6 @@ static int nfs_fs_context_parse_param(struct fs_context *fc,
 		 */
 	case Opt_sloppy:
 		ctx->sloppy = true;
-		dfprintk(MOUNT, "NFS:   relaxing parsing rules\n");
 		break;
 	}
 
@@ -879,10 +886,8 @@ static int nfs_parse_source(struct fs_context *fc,
 	size_t len;
 	const char *end;
 
-	if (unlikely(!dev_name || !*dev_name)) {
-		dfprintk(MOUNT, "NFS: device name not specified\n");
+	if (unlikely(!dev_name || !*dev_name))
 		return -EINVAL;
-	}
 
 	/* Is the host name protected with square brakcets? */
 	if (*dev_name == '[') {
@@ -922,7 +927,7 @@ static int nfs_parse_source(struct fs_context *fc,
 	if (!ctx->nfs_server.export_path)
 		goto out_nomem;
 
-	dfprintk(MOUNT, "NFS: MNTPATH: '%s'\n", ctx->nfs_server.export_path);
+	trace_nfs_mount_path(ctx->nfs_server.export_path);
 	return 0;
 
 out_bad_devname:
@@ -1116,7 +1121,6 @@ out_no_sec:
 	return nfs_invalf(fc, "NFS: nfs_mount_data version supports only AUTH_SYS");
 
 out_nomem:
-	dfprintk(MOUNT, "NFS: not enough memory to handle mount options");
 	return -ENOMEM;
 
 out_no_address:
@@ -1248,7 +1252,7 @@ static int nfs4_parse_monolithic(struct fs_context *fc,
 	if (IS_ERR(c))
 		return PTR_ERR(c);
 	ctx->nfs_server.export_path = c;
-	dfprintk(MOUNT, "NFS: MNTPATH: '%s'\n", c);
+	trace_nfs_mount_path(c);
 
 	c = strndup_user(data->client_addr.data, 16);
 	if (IS_ERR(c))
diff --git a/fs/nfs/internal.h b/fs/nfs/internal.h
index 437ebe544aaf..27c720d71b4e 100644
--- a/fs/nfs/internal.h
+++ b/fs/nfs/internal.h
@@ -707,6 +707,24 @@ unsigned long nfs_block_size(unsigned long bsize, unsigned char *nrbitsp)
 }
 
 /*
+ * Compute and set NFS server rsize / wsize
+ */
+static inline
+unsigned long nfs_io_size(unsigned long iosize, enum xprt_transports proto)
+{
+	if (iosize < NFS_MIN_FILE_IO_SIZE)
+		iosize = NFS_DEF_FILE_IO_SIZE;
+	else if (iosize >= NFS_MAX_FILE_IO_SIZE)
+		iosize = NFS_MAX_FILE_IO_SIZE;
+	else
+		iosize = iosize & PAGE_MASK;
+
+	if (proto == XPRT_TRANSPORT_UDP)
+		return nfs_block_bits(iosize, NULL);
+	return iosize;
+}
+
+/*
  * Determine the maximum file size for a superblock
  */
 static inline
@@ -861,3 +879,36 @@ static inline void nfs_set_port(struct sockaddr *sap, int *port,
 
 	rpc_set_port(sap, *port);
 }
+
+struct nfs_direct_req {
+	struct kref		kref;		/* release manager */
+
+	/* I/O parameters */
+	struct nfs_open_context	*ctx;		/* file open context info */
+	struct nfs_lock_context *l_ctx;		/* Lock context info */
+	struct kiocb *		iocb;		/* controlling i/o request */
+	struct inode *		inode;		/* target file of i/o */
+
+	/* completion state */
+	atomic_t		io_count;	/* i/os we're waiting for */
+	spinlock_t		lock;		/* protect completion state */
+
+	loff_t			io_start;	/* Start offset for I/O */
+	ssize_t			count,		/* bytes actually processed */
+				max_count,	/* max expected count */
+				bytes_left,	/* bytes left to be sent */
+				error;		/* any reported error */
+	struct completion	completion;	/* wait for i/o completion */
+
+	/* commit state */
+	struct nfs_mds_commit_info mds_cinfo;	/* Storage for cinfo */
+	struct pnfs_ds_commit_info ds_cinfo;	/* Storage for cinfo */
+	struct work_struct	work;
+	int			flags;
+	/* for write */
+#define NFS_ODIRECT_DO_COMMIT		(1)	/* an unstable reply was received */
+#define NFS_ODIRECT_RESCHED_WRITES	(2)	/* write verification failed */
+	/* for read */
+#define NFS_ODIRECT_SHOULD_DIRTY	(3)	/* dirty user-space page after read */
+#define NFS_ODIRECT_DONE		INT_MAX	/* write verification failed */
+};
diff --git a/fs/nfs/nfs3client.c b/fs/nfs/nfs3client.c
index 5601e47360c2..b49359afac88 100644
--- a/fs/nfs/nfs3client.c
+++ b/fs/nfs/nfs3client.c
@@ -108,7 +108,6 @@ struct nfs_client *nfs3_set_ds_client(struct nfs_server *mds_srv,
 	if (mds_srv->flags & NFS_MOUNT_NORESVPORT)
 		__set_bit(NFS_CS_NORESVPORT, &cl_init.init_flags);
 
-	__set_bit(NFS_CS_NOPING, &cl_init.init_flags);
 	__set_bit(NFS_CS_DS, &cl_init.init_flags);
 
 	/* Use the MDS nfs_client cl_ipaddr. */
diff --git a/fs/nfs/nfs42xdr.c b/fs/nfs/nfs42xdr.c
index 271e5f92ed01..b56f05113d36 100644
--- a/fs/nfs/nfs42xdr.c
+++ b/fs/nfs/nfs42xdr.c
@@ -1025,82 +1025,95 @@ static int decode_deallocate(struct xdr_stream *xdr, struct nfs42_falloc_res *re
 	return decode_op_hdr(xdr, OP_DEALLOCATE);
 }
 
-static int decode_read_plus_data(struct xdr_stream *xdr,
-				 struct nfs_pgio_args *args,
-				 struct nfs_pgio_res *res)
-{
-	uint32_t count, recvd;
+struct read_plus_segment {
+	enum data_content4 type;
 	uint64_t offset;
-	__be32 *p;
-
-	p = xdr_inline_decode(xdr, 8 + 4);
-	if (!p)
-		return 1;
+	union {
+		struct {
+			uint64_t length;
+		} hole;
+
+		struct {
+			uint32_t length;
+			unsigned int from;
+		} data;
+	};
+};
 
-	p = xdr_decode_hyper(p, &offset);
-	count = be32_to_cpup(p);
-	recvd = xdr_align_data(xdr, res->count, xdr_align_size(count));
-	if (recvd > count)
-		recvd = count;
-	if (res->count + recvd > args->count) {
-		if (args->count > res->count)
-			res->count += args->count - res->count;
-		return 1;
-	}
-	res->count += recvd;
-	if (count > recvd)
-		return 1;
-	return 0;
+static inline uint64_t read_plus_segment_length(struct read_plus_segment *seg)
+{
+	return seg->type == NFS4_CONTENT_DATA ? seg->data.length : seg->hole.length;
 }
 
-static int decode_read_plus_hole(struct xdr_stream *xdr,
-				 struct nfs_pgio_args *args,
-				 struct nfs_pgio_res *res, uint32_t *eof)
+static int decode_read_plus_segment(struct xdr_stream *xdr,
+				    struct read_plus_segment *seg)
 {
-	uint64_t offset, length, recvd;
 	__be32 *p;
 
-	p = xdr_inline_decode(xdr, 8 + 8);
+	p = xdr_inline_decode(xdr, 4);
 	if (!p)
-		return 1;
-
-	p = xdr_decode_hyper(p, &offset);
-	p = xdr_decode_hyper(p, &length);
-	if (offset != args->offset + res->count) {
-		/* Server returned an out-of-sequence extent */
-		if (offset > args->offset + res->count ||
-		    offset + length < args->offset + res->count) {
-			dprintk("NFS: server returned out of sequence extent: "
-				"offset/size = %llu/%llu != expected %llu\n",
-				(unsigned long long)offset,
-				(unsigned long long)length,
-				(unsigned long long)(args->offset +
-						     res->count));
-			return 1;
-		}
-		length -= args->offset + res->count - offset;
-	}
-	if (length + res->count > args->count) {
-		*eof = 0;
-		if (unlikely(res->count >= args->count))
-			return 1;
-		length = args->count - res->count;
-	}
-	recvd = xdr_expand_hole(xdr, res->count, length);
-	res->count += recvd;
+		return -EIO;
+	seg->type = be32_to_cpup(p++);
+
+	p = xdr_inline_decode(xdr, seg->type == NFS4_CONTENT_DATA ? 12 : 16);
+	if (!p)
+		return -EIO;
+	p = xdr_decode_hyper(p, &seg->offset);
+
+	if (seg->type == NFS4_CONTENT_DATA) {
+		struct xdr_buf buf;
+		uint32_t len = be32_to_cpup(p);
+
+		seg->data.length = len;
+		seg->data.from = xdr_stream_pos(xdr);
 
-	if (recvd < length)
-		return 1;
+		if (!xdr_stream_subsegment(xdr, &buf, xdr_align_size(len)))
+			return -EIO;
+	} else if (seg->type == NFS4_CONTENT_HOLE) {
+		xdr_decode_hyper(p, &seg->hole.length);
+	} else
+		return -EINVAL;
 	return 0;
 }
 
+static int process_read_plus_segment(struct xdr_stream *xdr,
+				     struct nfs_pgio_args *args,
+				     struct nfs_pgio_res *res,
+				     struct read_plus_segment *seg)
+{
+	unsigned long offset = seg->offset;
+	unsigned long length = read_plus_segment_length(seg);
+	unsigned int bufpos;
+
+	if (offset + length < args->offset)
+		return 0;
+	else if (offset > args->offset + args->count) {
+		res->eof = 0;
+		return 0;
+	} else if (offset < args->offset) {
+		length -= (args->offset - offset);
+		offset = args->offset;
+	} else if (offset + length > args->offset + args->count) {
+		length = (args->offset + args->count) - offset;
+		res->eof = 0;
+	}
+
+	bufpos = xdr->buf->head[0].iov_len + (offset - args->offset);
+	if (seg->type == NFS4_CONTENT_HOLE)<