Merge tag 'nfs-for-5.14-1' of git://git.linux-nfs.org/projects/trondmy/linux-nfs

Pull NFS client updates from Trond Myklebust: "Highlights include: Features: - Multiple patches to add support for fcntl() leases over NFSv4. - A sysfs interface to display more information about the various transport connections used by the RPC client - A sysfs interface to allow a suitably privileged user to offline a transport that may no longer point to a valid server - A sysfs interface to allow a suitably privileged user to change the server IP address used by the RPC client Stable fixes: - Two sunrpc fixes for deadlocks involving privileged rpc_wait_queues Bugfixes: - SUNRPC: Avoid a KASAN slab-out-of-bounds bug in xdr_set_page_base() - SUNRPC: prevent port reuse on transports which don't request it. - NFSv3: Fix memory leak in posix_acl_create() - NFS: Various fixes to attribute revalidation timeouts - NFSv4: Fix handling of non-atomic change attribute updates - NFSv4: If a server is down, don't cause mounts to other servers to hang as well - pNFS: Fix an Oops in pnfs_mark_request_commit() when doing O_DIRECT - NFS: Fix mount failures due to incorrect setting of the has_sec_mnt_opts filesystem flag - NFS: Ensure nfs_readpage returns promptly when an internal error occurs - NFS: Fix fscache read from NFS after cache error - pNFS: Various bugfixes around the LAYOUTGET operation" * tag 'nfs-for-5.14-1' of git://git.linux-nfs.org/projects/trondmy/linux-nfs: (46 commits) NFSv4/pNFS: Return an error if _nfs4_pnfs_v3_ds_connect can't load NFSv3 NFSv4/pNFS: Don't call _nfs4_pnfs_v3_ds_connect multiple times NFSv4/pnfs: Clean up layout get on open NFSv4/pnfs: Fix layoutget behaviour after invalidation NFSv4/pnfs: Fix the layout barrier update NFS: Fix fscache read from NFS after cache error NFS: Ensure nfs_readpage returns promptly when internal error occurs sunrpc: remove an offlined xprt using sysfs sunrpc: provide showing transport's state info in the sysfs directory sunrpc: display xprt's queuelen of assigned tasks via sysfs sunrpc: provide multipath info in the sysfs directory NFSv4.1 identify and mark RPC tasks that can move between transports sunrpc: provide transport info in the sysfs directory SUNRPC: take a xprt offline using sysfs sunrpc: add dst_attr attributes to the sysfs xprt directory SUNRPC for TCP display xprt's source port in sysfs xprt_info SUNRPC query transport's source port SUNRPC display xprt's main value in sysfs's xprt_info SUNRPC mark the first transport sunrpc: add add sysfs directory per xprt under each xprt_switch ...
author: Linus Torvalds 2021-07-09 09:43:57 -0700
committer: Linus Torvalds 2021-07-09 09:43:57 -0700
commit: 96890bc2eaa1f6bfc1b194e0f0815a10824352a4 (patch)
tree: c26a6f72adc8aa5ce62cc7549d95e275109e5617
parent: 227c4d507c71acb7bece298a98d83e5b44433f62 (diff)
parent: 878b3dfc42c4ddbf9e38cd9061e3ddd99a69747a (diff)
34 files changed, 1140 insertions, 234 deletions
diff --git a/fs/nfs/delegation.c b/fs/nfs/delegation.c
index e6ec6f09ac6e..11118398f495 100644
--- a/fs/nfs/delegation.c
+++ b/fs/nfs/delegation.c
@@ -75,6 +75,13 @@ void nfs_mark_delegation_referenced(struct nfs_delegation *delegation)
 	set_bit(NFS_DELEGATION_REFERENCED, &delegation->flags);
 }
 
+static void nfs_mark_return_delegation(struct nfs_server *server,
+				       struct nfs_delegation *delegation)
+{
+	set_bit(NFS_DELEGATION_RETURN, &delegation->flags);
+	set_bit(NFS4CLNT_DELEGRETURN, &server->nfs_client->cl_state);
+}
+
 static bool
 nfs4_is_valid_delegation(const struct nfs_delegation *delegation,
 		fmode_t flags)
@@ -293,6 +300,7 @@ nfs_start_delegation_return_locked(struct nfs_inode *nfsi)
 		goto out;
 	spin_lock(&delegation->lock);
 	if (!test_and_set_bit(NFS_DELEGATION_RETURNING, &delegation->flags)) {
+		clear_bit(NFS_DELEGATION_RETURN_DELAYED, &delegation->flags);
 		/* Refcount matched in nfs_end_delegation_return() */
 		ret = nfs_get_delegation(delegation);
 	}
@@ -314,16 +322,17 @@ nfs_start_delegation_return(struct nfs_inode *nfsi)
 	return delegation;
 }
 
-static void
-nfs_abort_delegation_return(struct nfs_delegation *delegation,
-		struct nfs_client *clp)
+static void nfs_abort_delegation_return(struct nfs_delegation *delegation,
+					struct nfs_client *clp, int err)
 {
 
 	spin_lock(&delegation->lock);
 	clear_bit(NFS_DELEGATION_RETURNING, &delegation->flags);
-	set_bit(NFS_DELEGATION_RETURN, &delegation->flags);
+	if (err == -EAGAIN) {
+		set_bit(NFS_DELEGATION_RETURN_DELAYED, &delegation->flags);
+		set_bit(NFS4CLNT_DELEGRETURN_DELAYED, &clp->cl_state);
+	}
 	spin_unlock(&delegation->lock);
-	set_bit(NFS4CLNT_DELEGRETURN, &clp->cl_state);
 }
 
 static struct nfs_delegation *
@@ -521,11 +530,18 @@ out:
 static int nfs_end_delegation_return(struct inode *inode, struct nfs_delegation *delegation, int issync)
 {
 	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
+	unsigned int mode = O_WRONLY | O_RDWR;
 	int err = 0;
 
 	if (delegation == NULL)
 		return 0;
-	do {
+
+	if (!issync)
+		mode |= O_NONBLOCK;
+	/* Recall of any remaining application leases */
+	err = break_lease(inode, mode);
+
+	while (err == 0) {
 		if (test_bit(NFS_DELEGATION_REVOKED, &delegation->flags))
 			break;
 		err = nfs_delegation_claim_opens(inode, &delegation->stateid,
@@ -536,10 +552,10 @@ static int nfs_end_delegation_return(struct inode *inode, struct nfs_delegation
 		 * Guard against state recovery
 		 */
 		err = nfs4_wait_clnt_recover(clp);
-	} while (err == 0);
+	}
 
 	if (err) {
-		nfs_abort_delegation_return(delegation, clp);
+		nfs_abort_delegation_return(delegation, clp, err);
 		goto out;
 	}
 
@@ -568,6 +584,7 @@ static bool nfs_delegation_need_return(struct nfs_delegation *delegation)
 	if (ret)
 		clear_bit(NFS_DELEGATION_RETURN_IF_CLOSED, &delegation->flags);
 	if (test_bit(NFS_DELEGATION_RETURNING, &delegation->flags) ||
+	    test_bit(NFS_DELEGATION_RETURN_DELAYED, &delegation->flags) ||
 	    test_bit(NFS_DELEGATION_REVOKED, &delegation->flags))
 		ret = false;
 
@@ -647,6 +664,38 @@ out:
 	return err;
 }
 
+static bool nfs_server_clear_delayed_delegations(struct nfs_server *server)
+{
+	struct nfs_delegation *d;
+	bool ret = false;
+
+	list_for_each_entry_rcu (d, &server->delegations, super_list) {
+		if (!test_bit(NFS_DELEGATION_RETURN_DELAYED, &d->flags))
+			continue;
+		nfs_mark_return_delegation(server, d);
+		clear_bit(NFS_DELEGATION_RETURN_DELAYED, &d->flags);
+		ret = true;
+	}
+	return ret;
+}
+
+static bool nfs_client_clear_delayed_delegations(struct nfs_client *clp)
+{
+	struct nfs_server *server;
+	bool ret = false;
+
+	if (!test_and_clear_bit(NFS4CLNT_DELEGRETURN_DELAYED, &clp->cl_state))
+		goto out;
+	rcu_read_lock();
+	list_for_each_entry_rcu (server, &clp->cl_superblocks, client_link) {
+		if (nfs_server_clear_delayed_delegations(server))
+			ret = true;
+	}
+	rcu_read_unlock();
+out:
+	return ret;
+}
+
 /**
  * nfs_client_return_marked_delegations - return previously marked delegations
  * @clp: nfs_client to process
@@ -659,8 +708,14 @@ out:
  */
 int nfs_client_return_marked_delegations(struct nfs_client *clp)
 {
-	return nfs_client_for_each_server(clp,
-			nfs_server_return_marked_delegations, NULL);
+	int err = nfs_client_for_each_server(
+		clp, nfs_server_return_marked_delegations, NULL);
+	if (err)
+		return err;
+	/* If a return was delayed, sleep to prevent hard looping */
+	if (nfs_client_clear_delayed_delegations(clp))
+		ssleep(1);
+	return 0;
 }
 
 /**
@@ -698,13 +753,14 @@ int nfs4_inode_return_delegation(struct inode *inode)
 {
 	struct nfs_inode *nfsi = NFS_I(inode);
 	struct nfs_delegation *delegation;
-	int err = 0;
 
-	nfs_wb_all(inode);
 	delegation = nfs_start_delegation_return(nfsi);
+	/* Synchronous recall of any application leases */
+	break_lease(inode, O_WRONLY | O_RDWR);
+	nfs_wb_all(inode);
 	if (delegation != NULL)
-		err = nfs_end_delegation_return(inode, delegation, 1);
-	return err;
+		return nfs_end_delegation_return(inode, delegation, 1);
+	return 0;
 }
 
 /**
@@ -775,13 +831,6 @@ static void nfs_mark_return_if_closed_delegation(struct nfs_server *server,
 	set_bit(NFS4CLNT_DELEGRETURN, &server->nfs_client->cl_state);
 }
 
-static void nfs_mark_return_delegation(struct nfs_server *server,
-		struct nfs_delegation *delegation)
-{
-	set_bit(NFS_DELEGATION_RETURN, &delegation->flags);
-	set_bit(NFS4CLNT_DELEGRETURN, &server->nfs_client->cl_state);
-}
-
 static bool nfs_server_mark_return_all_delegations(struct nfs_server *server)
 {
 	struct nfs_delegation *delegation;
@@ -1010,6 +1059,9 @@ int nfs_async_inode_return_delegation(struct inode *inode,
 	nfs_mark_return_delegation(server, delegation);
 	rcu_read_unlock();
 
+	/* If there are any application leases or delegations, recall them */
+	break_lease(inode, O_WRONLY | O_RDWR | O_NONBLOCK);
+
 	nfs_delegation_run_state_manager(clp);
 	return 0;
 out_enoent:
diff --git a/fs/nfs/delegation.h b/fs/nfs/delegation.h
index c19b4fd20781..1c378992b7c0 100644
--- a/fs/nfs/delegation.h
+++ b/fs/nfs/delegation.h
@@ -36,6 +36,7 @@ enum {
 	NFS_DELEGATION_REVOKED,
 	NFS_DELEGATION_TEST_EXPIRED,
 	NFS_DELEGATION_INODE_FREEING,
+	NFS_DELEGATION_RETURN_DELAYED,
 };
 
 int nfs_inode_set_delegation(struct inode *inode, const struct cred *cred,
diff --git a/fs/nfs/direct.c b/fs/nfs/direct.c
index 2d30a4da49fa..2e894fec036b 100644
--- a/fs/nfs/direct.c
+++ b/fs/nfs/direct.c
@@ -700,8 +700,8 @@ static void nfs_direct_write_completion(struct nfs_pgio_header *hdr)
 {
 	struct nfs_direct_req *dreq = hdr->dreq;
 	struct nfs_commit_info cinfo;
-	bool request_commit = false;
 	struct nfs_page *req = nfs_list_entry(hdr->pages.next);
+	int flags = NFS_ODIRECT_DONE;
 
 	nfs_init_cinfo_from_dreq(&cinfo, dreq);
 
@@ -713,15 +713,9 @@ 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)) {
-		switch (dreq->flags) {
-		case 0:
+		if (!dreq->flags)
 			dreq->flags = NFS_ODIRECT_DO_COMMIT;
-			request_commit = true;
-			break;
-		case NFS_ODIRECT_RESCHED_WRITES:
-		case NFS_ODIRECT_DO_COMMIT:
-			request_commit = true;
-		}
+		flags = dreq->flags;
 	}
 	spin_unlock(&dreq->lock);
 
@@ -729,12 +723,15 @@ static void nfs_direct_write_completion(struct nfs_pgio_header *hdr)
 
 		req = nfs_list_entry(hdr->pages.next);
 		nfs_list_remove_request(req);
-		if (request_commit) {
+		if (flags == NFS_ODIRECT_DO_COMMIT) {
 			kref_get(&req->wb_kref);
 			memcpy(&req->wb_verf, &hdr->verf.verifier,
 			       sizeof(req->wb_verf));
 			nfs_mark_request_commit(req, hdr->lseg, &cinfo,
 				hdr->ds_commit_idx);
+		} else if (flags == NFS_ODIRECT_RESCHED_WRITES) {
+			kref_get(&req->wb_kref);
+			nfs_mark_request_commit(req, NULL, &cinfo, 0);
 		}
 		nfs_unlock_and_release_request(req);
 	}
diff --git a/fs/nfs/fscache.c b/fs/nfs/fscache.c
index c4c021c6ebbd..d743629e05e1 100644
--- a/fs/nfs/fscache.c
+++ b/fs/nfs/fscache.c
@@ -385,12 +385,15 @@ static void nfs_readpage_from_fscache_complete(struct page *page,
 		 "NFS: readpage_from_fscache_complete (0x%p/0x%p/%d)\n",
 		 page, context, error);
 
-	/* if the read completes with an error, we just unlock the page and let
-	 * the VM reissue the readpage */
-	if (!error) {
+	/*
+	 * If the read completes with an error, mark the page with PG_checked,
+	 * unlock the page, and let the VM reissue the readpage.
+	 */
+	if (!error)
 		SetPageUptodate(page);
-		unlock_page(page);
-	}
+	else
+		SetPageChecked(page);
+	unlock_page(page);
 }
 
 /*
@@ -405,6 +408,11 @@ int __nfs_readpage_from_fscache(struct nfs_open_context *ctx,
 		 "NFS: readpage_from_fscache(fsc:%p/p:%p(i:%lx f:%lx)/0x%p)\n",
 		 nfs_i_fscache(inode), page, page->index, page->flags, inode);
 
+	if (PageChecked(page)) {
+		ClearPageChecked(page);
+		return 1;
+	}
+
 	ret = fscache_read_or_alloc_page(nfs_i_fscache(inode),
 					 page,
 					 nfs_readpage_from_fscache_complete,
diff --git a/fs/nfs/getroot.c b/fs/nfs/getroot.c
index aaeeb4659bff..59355c106ece 100644
--- a/fs/nfs/getroot.c
+++ b/fs/nfs/getroot.c
@@ -67,7 +67,7 @@ static int nfs_superblock_set_dummy_root(struct super_block *sb, struct inode *i
 int nfs_get_root(struct super_block *s, struct fs_context *fc)
 {
 	struct nfs_fs_context *ctx = nfs_fc2context(fc);
-	struct nfs_server *server = NFS_SB(s);
+	struct nfs_server *server = NFS_SB(s), *clone_server;
 	struct nfs_fsinfo fsinfo;
 	struct dentry *root;
 	struct inode *inode;
@@ -127,7 +127,7 @@ int nfs_get_root(struct super_block *s, struct fs_context *fc)
 	}
 	spin_unlock(&root->d_lock);
 	fc->root = root;
-	if (NFS_SB(s)->caps & NFS_CAP_SECURITY_LABEL)
+	if (server->caps & NFS_CAP_SECURITY_LABEL)
 		kflags |= SECURITY_LSM_NATIVE_LABELS;
 	if (ctx->clone_data.sb) {
 		if (d_inode(fc->root)->i_fop != &nfs_dir_operations) {
@@ -137,15 +137,19 @@ int nfs_get_root(struct super_block *s, struct fs_context *fc)
 		/* clone lsm security options from the parent to the new sb */
 		error = security_sb_clone_mnt_opts(ctx->clone_data.sb,
 						   s, kflags, &kflags_out);
+		if (error)
+			goto error_splat_root;
+		clone_server = NFS_SB(ctx->clone_data.sb);
+		server->has_sec_mnt_opts = clone_server->has_sec_mnt_opts;
 	} else {
 		error = security_sb_set_mnt_opts(s, fc->security,
 							kflags, &kflags_out);
 	}
 	if (error)
 		goto error_splat_root;
-	if (NFS_SB(s)->caps & NFS_CAP_SECURITY_LABEL &&
+	if (server->caps & NFS_CAP_SECURITY_LABEL &&
 		!(kflags_out & SECURITY_LSM_NATIVE_LABELS))
-		NFS_SB(s)->caps &= ~NFS_CAP_SECURITY_LABEL;
+		server->caps &= ~NFS_CAP_SECURITY_LABEL;
 
 	nfs_setsecurity(inode, fsinfo.fattr, fsinfo.fattr->label);
 	error = 0;
diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index 529c4099f482..853213b3a209 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -1101,6 +1101,7 @@ EXPORT_SYMBOL_GPL(nfs_inode_attach_open_context);
 void nfs_file_set_open_context(struct file *filp, struct nfs_open_context *ctx)
 {
 	filp->private_data = get_nfs_open_context(ctx);
+	set_bit(NFS_CONTEXT_FILE_OPEN, &ctx->flags);
 	if (list_empty(&ctx->list))
 		nfs_inode_attach_open_context(ctx);
 }
@@ -1120,6 +1121,8 @@ struct nfs_open_context *nfs_find_open_context(struct inode *inode, const struct
 			continue;
 		if ((pos->mode & (FMODE_READ|FMODE_WRITE)) != mode)
 			continue;
+		if (!test_bit(NFS_CONTEXT_FILE_OPEN, &pos->flags))
+			continue;
 		ctx = get_nfs_open_context(pos);
 		if (ctx)
 			break;
@@ -1135,6 +1138,7 @@ void nfs_file_clear_open_context(struct file *filp)
 	if (ctx) {
 		struct inode *inode = d_inode(ctx->dentry);
 
+		clear_bit(NFS_CONTEXT_FILE_OPEN, &ctx->flags);
 		/*
 		 * We fatal error on write before. Try to writeback
 		 * every page again.
@@ -2055,35 +2059,33 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
 					| NFS_INO_INVALID_OTHER;
 				if (S_ISDIR(inode->i_mode))
 					nfs_force_lookup_revalidate(inode);
+				attr_changed = true;
 				dprintk("NFS: change_attr change on server for file %s/%ld\n",
 						inode->i_sb->s_id,
 						inode->i_ino);
 			} else if (!have_delegation)
 				nfsi->cache_validity |= NFS_INO_DATA_INVAL_DEFER;
 			inode_set_iversion_raw(inode, fattr->change_attr);
-			attr_changed = true;
 		}
 	} else {
 		nfsi->cache_validity |=
 			save_cache_validity & NFS_INO_INVALID_CHANGE;
-		cache_revalidated = false;
+		if (!have_delegation ||
+		    (nfsi->cache_validity & NFS_INO_INVALID_CHANGE) != 0)
+			cache_revalidated = false;
 	}
 
-	if (fattr->valid & NFS_ATTR_FATTR_MTIME) {
+	if (fattr->valid & NFS_ATTR_FATTR_MTIME)
 		inode->i_mtime = fattr->mtime;
-	} else if (fattr_supported & NFS_ATTR_FATTR_MTIME) {
+	else if (fattr_supported & NFS_ATTR_FATTR_MTIME)
 		nfsi->cache_validity |=
 			save_cache_validity & NFS_INO_INVALID_MTIME;
-		cache_revalidated = false;
-	}
 
-	if (fattr->valid & NFS_ATTR_FATTR_CTIME) {
+	if (fattr->valid & NFS_ATTR_FATTR_CTIME)
 		inode->i_ctime = fattr->ctime;
-	} else if (fattr_supported & NFS_ATTR_FATTR_CTIME) {
+	else if (fattr_supported & NFS_ATTR_FATTR_CTIME)
 		nfsi->cache_validity |=
 			save_cache_validity & NFS_INO_INVALID_CTIME;
-		cache_revalidated = false;
-	}
 
 	/* Check if our cached file size is stale */
 	if (fattr->valid & NFS_ATTR_FATTR_SIZE) {
@@ -2096,7 +2098,6 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
 				i_size_write(inode, new_isize);
 				if (!have_writers)
 					invalid |= NFS_INO_INVALID_DATA;
-				attr_changed = true;
 			}
 			dprintk("NFS: isize change on server for file %s/%ld "
 					"(%Ld to %Ld)\n",
@@ -2111,19 +2112,15 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
 			fattr->du.nfs3.used = 0;
 			fattr->valid |= NFS_ATTR_FATTR_SPACE_USED;
 		}
-	} else {
+	} else
 		nfsi->cache_validity |=
 			save_cache_validity & NFS_INO_INVALID_SIZE;
-		cache_revalidated = false;
-	}
 
 	if (fattr->valid & NFS_ATTR_FATTR_ATIME)
 		inode->i_atime = fattr->atime;
-	else if (fattr_supported & NFS_ATTR_FATTR_ATIME) {
+	else if (fattr_supported & NFS_ATTR_FATTR_ATIME)
 		nfsi->cache_validity |=
 			save_cache_validity & NFS_INO_INVALID_ATIME;
-		cache_revalidated = false;
-	}
 
 	if (fattr->valid & NFS_ATTR_FATTR_MODE) {
 		if ((inode->i_mode & S_IALLUGO) != (fattr->mode & S_IALLUGO)) {
@@ -2132,71 +2129,55 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
 			inode->i_mode = newmode;
 			invalid |= NFS_INO_INVALID_ACCESS
 				| NFS_INO_INVALID_ACL;
-			attr_changed = true;
 		}
-	} else if (fattr_supported & NFS_ATTR_FATTR_MODE) {
+	} else if (fattr_supported & NFS_ATTR_FATTR_MODE)
 		nfsi->cache_validity |=
 			save_cache_validity & NFS_INO_INVALID_MODE;
-		cache_revalidated = false;
-	}
 
 	if (fattr->valid & NFS_ATTR_FATTR_OWNER) {
 		if (!uid_eq(inode->i_uid, fattr->uid)) {
 			invalid |= NFS_INO_INVALID_ACCESS
 				| NFS_INO_INVALID_ACL;
 			inode->i_uid = fattr->uid;
-			attr_changed = true;
 		}
-	} else if (fattr_supported & NFS_ATTR_FATTR_OWNER) {
+	} else if (fattr_supported & NFS_ATTR_FATTR_OWNER)
 		nfsi->cache_validity |=
 			save_cache_validity & NFS_INO_INVALID_OTHER;
-		cache_revalidated = false;
-	}
 
 	if (fattr->valid & NFS_ATTR_FATTR_GROUP) {
 		if (!gid_eq(inode->i_gid, fattr->gid)) {
 			invalid |= NFS_INO_INVALID_ACCESS
 				| NFS_INO_INVALID_ACL;
 			inode->i_gid = fattr->gid;
-			attr_changed = true;
 		}
-	} else if (fattr_supported & NFS_ATTR_FATTR_GROUP) {
+	} else if (fattr_supported & NFS_ATTR_FATTR_GROUP)
 		nfsi->cache_validity |=
 			save_cache_validity & NFS_INO_INVALID_OTHER;
-		cache_revalidated = false;
-	}
 
 	if (fattr->valid & NFS_ATTR_FATTR_NLINK) {
 		if (inode->i_nlink != fattr->nlink) {
 			if (S_ISDIR(inode->i_mode))
 				invalid |= NFS_INO_INVALID_DATA;
 			set_nlink(inode, fattr->nlink);
-			attr_changed = true;
 		}
-	} else if (fattr_supported & NFS_ATTR_FATTR_NLINK) {
+	} else if (fattr_supported & NFS_ATTR_FATTR_NLINK)
 		nfsi->cache_validity |=
 			save_cache_validity & NFS_INO_INVALID_NLINK;
-		cache_revalidated = false;
-	}
 
 	if (fattr->valid & NFS_ATTR_FATTR_SPACE_USED) {
 		/*
 		 * report the blocks in 512byte units
 		 */
 		inode->i_blocks = nfs_calc_block_size(fattr->du.nfs3.used);
-	} else if (fattr_supported & NFS_ATTR_FATTR_SPACE_USED) {
+	} else if (fattr_supported & NFS_ATTR_FATTR_SPACE_USED)
 		nfsi->cache_validity |=
 			save_cache_validity & NFS_INO_INVALID_BLOCKS;
-		cache_revalidated = false;
-	}
 
-	if (fattr->valid & NFS_ATTR_FATTR_BLOCKS_USED) {
+	if (fattr->valid & NFS_ATTR_FATTR_BLOCKS_USED)
 		inode->i_blocks = fattr->du.nfs2.blocks;
-	} else if (fattr_supported & NFS_ATTR_FATTR_BLOCKS_USED) {
+	else if (fattr_supported & NFS_ATTR_FATTR_BLOCKS_USED)
 		nfsi->cache_validity |=
 			save_cache_validity & NFS_INO_INVALID_BLOCKS;
-		cache_revalidated = false;
-	}
 
 	/* Update attrtimeo value if we're out of the unstable period */
 	if (attr_changed) {
diff --git a/fs/nfs/nfs3proc.c b/fs/nfs/nfs3proc.c
index 5c4e23abc345..2299446b3b89 100644
--- a/fs/nfs/nfs3proc.c
+++ b/fs/nfs/nfs3proc.c
@@ -385,7 +385,7 @@ nfs3_proc_create(struct inode *dir, struct dentry *dentry, struct iattr *sattr,
 				break;
 
 			case NFS3_CREATE_UNCHECKED:
-				goto out;
+				goto out_release_acls;
 		}
 		nfs_fattr_init(data->res.dir_attr);
 		nfs_fattr_init(data->res.fattr);
@@ -751,7 +751,7 @@ nfs3_proc_mknod(struct inode *dir, struct dentry *dentry, struct iattr *sattr,
 		break;
 	default:
 		status = -EINVAL;
-		goto out;
+		goto out_release_acls;
 	}
 
 	d_alias = nfs3_do_create(dir, dentry, data);
diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h
index 543d916f79ab..ba78df4b13d9 100644
--- a/fs/nfs/nfs4_fs.h
+++ b/fs/nfs/nfs4_fs.h
@@ -45,6 +45,7 @@ enum nfs4_client_state {
 	NFS4CLNT_RECALL_RUNNING,
 	NFS4CLNT_RECALL_ANY_LAYOUT_READ,
 	NFS4CLNT_RECALL_ANY_LAYOUT_RW,
+	NFS4CLNT_DELEGRETURN_DELAYED,
 };
 
 #define NFS4_RENEW_TIMEOUT		0x01
@@ -322,7 +323,8 @@ extern int update_open_stateid(struct nfs4_state *state,
 				const nfs4_stateid *open_stateid,
 				const nfs4_stateid *deleg_stateid,
 				fmode_t fmode);
-
+extern int nfs4_proc_setlease(struct file *file, long arg,
+			      struct file_lock **lease, void **priv);
 extern int nfs4_proc_get_lease_time(struct nfs_client *clp,
 		struct nfs_fsinfo *fsinfo);
 extern void nfs4_update_changeattr(struct inode *dir,
diff --git a/fs/nfs/nfs4client.c b/fs/nfs/nfs4client.c
index 42719384e25f..28431acd1230 100644
--- a/fs/nfs/nfs4client.c
+++ b/fs/nfs/nfs4client.c
@@ -197,8 +197,11 @@ void nfs40_shutdown_client(struct nfs_client *clp)
 
 struct nfs_client *nfs4_alloc_client(const struct nfs_client_initdata *cl_init)
 {
-	int err;
+	char buf[INET6_ADDRSTRLEN + 1];
+	const char *ip_addr = cl_init->ip_addr;
 	struct nfs_client *clp = nfs_alloc_client(cl_init);
+	int err;
+
 	if (IS_ERR(clp))
 		return clp;
 
@@ -222,6 +225,44 @@ struct nfs_client *nfs4_alloc_client(const struct nfs_client_initdata *cl_init)
 	init_waitqueue_head(&clp->cl_lock_waitq);
 #endif
 	INIT_LIST_HEAD(&clp->pending_cb_stateids);
+
+	if (cl_init->minorversion != 0)
+		__set_bit(NFS_CS_INFINITE_SLOTS, &clp->cl_flags);
+	__set_bit(NFS_CS_DISCRTRY, &clp->cl_flags);
+	__set_bit(NFS_CS_NO_RETRANS_TIMEOUT, &clp->cl_flags);
+
+	/*
+	 * Set up the connection to the server before we add add to the
+	 * global list.
+	 */
+	err = nfs_create_rpc_client(clp, cl_init, RPC_AUTH_GSS_KRB5I);
+	if (err == -EINVAL)
+		err = nfs_create_rpc_client(clp, cl_init, RPC_AUTH_UNIX);
+	if (err < 0)
+		goto error;
+
+	/* If no clientaddr= option was specified, find a usable cb address */
+	if (ip_addr == NULL) {
+		struct sockaddr_storage cb_addr;
+		struct sockaddr *sap = (struct sockaddr *)&cb_addr;
+
+		err = rpc_localaddr(clp->cl_rpcclient, sap, sizeof(cb_addr));
+		if (err < 0)
+			goto error;
+		err = rpc_ntop(sap, buf, sizeof(buf));
+		if (err < 0)
+			goto error;
+		ip_addr = (const char *)buf;
+	}
+	strlcpy(clp->cl_ipaddr, ip_addr, sizeof(clp->cl_ipaddr));
+
+	err = nfs_idmap_new(clp);
+	if (err < 0) {
+		dprintk("%s: failed to create idmapper. Error = %d\n",
+			__func__, err);
+		goto error;
+	}
+	__set_bit(NFS_CS_IDMAP, &clp->cl_res_state);
 	return clp;
 
 error:
@@ -372,8 +413,6 @@ static int nfs4_init_client_minor_version(struct nfs_client *clp)
 struct nfs_client *nfs4_init_client(struct nfs_client *clp,
 				    const struct nfs_client_initdata *cl_init)
 {
-	char buf[INET6_ADDRSTRLEN + 1];
-	const char *ip_addr = cl_init->ip_addr;
 	struct nfs_client *old;
 	int error;
 
@@ -381,43 +420,6 @@ struct nfs_client *nfs4_init_client(struct nfs_client *clp,
 		/* the client is initialised already */
 		return clp;
 
-	/* Check NFS protocol revision and initialize RPC op vector */
-	clp->rpc_ops = &nfs_v4_clientops;
-
-	if (clp->cl_minorversion != 0)
-		__set_bit(NFS_CS_INFINITE_SLOTS, &clp->cl_flags);
-	__set_bit(NFS_CS_DISCRTRY, &clp->cl_flags);
-	__set_bit(NFS_CS_NO_RETRANS_TIMEOUT, &clp->cl_flags);
-
-	error = nfs_create_rpc_client(clp, cl_init, RPC_AUTH_GSS_KRB5I);
-	if (error == -EINVAL)
-		error = nfs_create_rpc_client(clp, cl_init, RPC_AUTH_UNIX);
-	if (error < 0)
-		goto error;
-
-	/* If no clientaddr= option was specified, find a usable cb address */
-	if (ip_addr == NULL) {
-		struct sockaddr_storage cb_addr;
-		struct sockaddr *sap = (struct sockaddr *)&cb_addr;
-
-		error = rpc_localaddr(clp->cl_rpcclient, sap, sizeof(cb_addr));
-		if (error < 0)
-			goto error;
-		error = rpc_ntop(sap, buf, sizeof(buf));
-		if (error < 0)
-			goto error;
-		ip_addr = (const char *)buf;
-	}
-	strlcpy(clp->cl_ipaddr, ip_addr, sizeof(clp->cl_ipaddr));
-
-	error = nfs_idmap_new(clp);
-	if (error < 0) {
-		dprintk("%s: failed to create idmapper. Error = %d\n",
-			__func__, error);
-		goto error;
-	}
-	__set_bit(NFS_CS_IDMAP, &clp->cl_res_state);
-
 	error = nfs4_init_client_minor_version(clp);
 	if (error < 0)
 		goto error;
diff --git a/fs/nfs/nfs4file.c b/fs/nfs/nfs4file.c
index a1e5c6b85ded..c820de58a661 100644
--- a/fs/nfs/nfs4file.c
+++ b/fs/nfs/nfs4file.c
@@ -435,6 +435,12 @@ void nfs42_ssc_unregister_ops(void)
 }
 #endif /* CONFIG_NFS_V4_2 */
 
+static int nfs4_setlease(struct file *file, long arg, struct file_lock **lease,
+			 void **priv)
+{
+	return nfs4_proc_setlease(file, arg, lease, priv);
+}
+
 const struct file_operations nfs4_file_operations = {
 	.read_iter	= nfs_file_read,
 	.write_iter	= nfs_file_write,
@@ -448,7 +454,7 @@ const struct file_operations nfs4_file_operations = {
 	.splice_read	= generic_file_splice_read,
 	.splice_write	= iter_file_splice_write,
 	.check_flags	= nfs_check_flags,
-	.setlease	= simple_nosetlease,
+	.setlease	= nfs4_setlease,
 #ifdef CONFIG_NFS_V4_2
 	.copy_file_range = nfs4_copy_file_range,
 	.llseek		= nfs4_file_llseek,
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index e653654c10bc..e1214bb6b7ee 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -1155,7 +1155,11 @@ static int nfs4_call_sync_sequence(struct rpc_clnt *clnt,
 				   struct nfs4_sequence_args *args,
 				   struct nfs4_sequence_res *res)
 {
-	return nfs4_do_call_sync(clnt, server, msg, args, res, 0);
+	unsigned short task_flags = 0;
+
+	if (server->nfs_client->cl_minorversion)
+		task_flags = RPC_TASK_MOVEABLE;
+	return nfs4_do_call_sync(clnt, server, msg, args, res, task_flags);
 }
 
 
@@ -1205,12 +1209,12 @@ nfs4_update_changeattr_locked(struct inode *inode,
 	u64 change_attr = inode_peek_iversion_raw(inode);
 
 	cache_validity |= NFS_INO_INVALID_CTIME | NFS_INO_INVALID_MTIME;
+	if (S_ISDIR(inode->i_mode))
+		cache_validity |= NFS_INO_INVALID_DATA;
 
 	switch (NFS_SERVER(inode)->change_attr_type) {
 	case NFS4_CHANGE_TYPE_IS_UNDEFINED:
-		break;
-	case NFS4_CHANGE_TYPE_IS_TIME_METADATA:
-		if ((s64)(change_attr - cinfo->after) > 0)
+		if (cinfo->after == change_attr)
 			goto out;
 		break;
 	default:
@@ -1218,24 +1222,21 @@ nfs4_update_changeattr_locked(struct inode *inode,
 			goto out;
 	}
 
-	if (cinfo->atomic && cinfo->before == change_attr) {
-		nfsi->attrtimeo_timestamp = jiffies;
-	} else {
-		if (S_ISDIR(inode->i_mode)) {
-			cache_validity |= NFS_INO_INVALID_DATA;
+	inode_set_iversion_raw(inode, cinfo->after);
+	if (!cinfo->atomic || cinfo->before != change_attr) {
+		if (S_ISDIR(inode->i_mode))
 			nfs_force_lookup_revalidate(inode);
-		} else {
-			if (!NFS_PROTO(inode)->have_delegation(inode,
-							       FMODE_READ))
-				cache_validity |= NFS_INO_REVAL_PAGECACHE;
-		}
 
-		if (cinfo->before != change_attr)
-			cache_validity |= NFS_INO_INVALID_ACCESS |
-					  NFS_INO_INVALID_ACL |
-					  NFS_INO_INVALID_XATTR;
+		if (!NFS_PROTO(inode)->have_delegation(inode, FMODE_READ))
+			cache_validity |=
+				NFS_INO_INVALID_ACCESS | NFS_INO_INVALID_ACL |
+				NFS_INO_INVALID_SIZE | NFS_INO_INVALID_OTHER |
+				NFS_INO_INVALID_BLOCKS | NFS_INO_INVALID_NLINK |
+				NFS_INO_INVALID_MODE | NFS_INO_INVALID_XATTR |
+				NFS_INO_REVAL_PAGECACHE;
+		nfsi->attrtimeo = NFS_MINATTRTIMEO(inode);
 	}
-	inode_set_iversion_raw(inode, cinfo->after);
+	nfsi->attrtimeo_timestamp = jiffies;
 	nfsi->read_cache_jiffies = timestamp;
 	nfsi->attr_gencount = nfs_inc_attr_generation_counter();
 	nfsi->cache_validity &= ~NFS_INO_INVALID_CHANGE;
@@ -2569,6 +2570,9 @@ static int nfs4_run_open_task(struct nfs4_opendata *data,
 	};
 	int status;
 
+	if (server->nfs_client->cl_minorversion)
+		task_setup_data.flags |= RPC_TASK_MOVEABLE;
+
 	kref_get(&data->kref);
 	data->rpc_done = false;
 	data->rpc_status = 0;
@@ -3749,6 +3753,9 @@ int nfs4_do_close(struct nfs4_state *state, gfp_t gfp_mask, int wait)
 	};
 	int status = -ENOMEM;
 
+	if (server->nfs_client->cl_minorversion)
+		task_setup_data.flags |= RPC_TASK_MOVEABLE;
+
 	nfs4_state_protect(server->nfs_client, NFS_SP4_MACH_CRED_CLEANUP,
 		&task_setup_data.rpc_client, &msg);
 
@@ -4188,6 +4195,9 @@ static int _nfs4_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle,
 	};
 	unsigned short task_flags = 0;
 
+	if (nfs4_has_session(server->nfs_client))
+		task_flags = RPC_TASK_MOVEABLE;
+
 	/* Is this is an attribute revalidation, subject to softreval? */
 	if (inode && (server->flags & NFS_MOUNT_SOFTREVAL))
 		task_flags |= RPC_TASK_TIMEOUT;
@@ -4307,6 +4317,9 @@ static int _nfs4_proc_lookup(struct rpc_clnt *clnt, struct inode *dir,
 	};
 	unsigned short task_flags = 0;
 
+	if (server->nfs_client->cl_minorversion)
+		task_flags = RPC_TASK_MOVEABLE;
+
 	/* Is this is an attribute revalidation, subject to softreval? */
 	if (nfs_lookup_is_soft_revalidate(dentry))
 		task_flags |= RPC_TASK_TIMEOUT;
@@ -6538,7 +6551,7 @@ static int _nfs4_proc_delegreturn(struct inode *inode, const struct cred *cred,
 		.rpc_client = server->client,
 		.rpc_message = &msg,
 		.callback_ops = &nfs4_delegreturn_ops,
-		.flags = RPC_TASK_ASYNC | RPC_TASK_TIMEOUT,
+		.flags = RPC_TASK_ASYNC | RPC_TASK_TIMEOUT | RPC_TASK_MOVEABLE,
 	};
 	int status = 0;
 
@@ -6856,6 +6869,11 @@ static struct rpc_task *nfs4_do_unlck(struct file_lock *fl,
 		.workqueue = nfsiod_workqueue,
 		.flags = RPC_TASK_ASYNC,
 	};
+	struct nfs_client *client =
+		NFS_SERVER(lsp->ls_state->inode)->nfs_client;
+
+	if (client->cl_minorversion)
+		task_setup_data.flags |= RPC_TASK_MOVEABLE;
 
 	nfs4_state_protect(NFS_SERVER(lsp->ls_state->inode)->nfs_client,
 		NFS_SP4_MACH_CRED_CLEANUP, &task_setup_data.rpc_client, &msg);
@@ -7130,6 +7148,10 @@ static int _nfs4_do_setlk(struct nfs4_state *state, int cmd, struct file_lock *f
 		.flags = RPC_TASK_ASYNC | RPC_TASK_CRED_NOREF,
 	};
 	int ret;
+	struct nfs_client *client = NFS_SERVER(state->inode)->nfs_client;
+
+	if (client->cl_minorversion)
+		task_setup_data.flags |= RPC_TASK_MOVEABLE;
 
 	dprintk("%s: begin!\n", __func__);
 	data = nfs4_alloc_lockdata(fl, nfs_file_open_context(fl->fl_file),
@@ -7438,6 +7460,43 @@ nfs4_proc_lock(struct file *filp, int cmd, struct file_lock *request)
 	return nfs4_retry_setlk(state, cmd, request);
 }
 
+static int nfs4_delete_lease(struct file *file, void **priv)
+{
+	return generic_setlease(file, F_UNLCK, NULL, priv);
+}
+
+static int nfs4_add_lease(struct file *file, long arg, struct file_lock **lease,
+			  void **priv)
+{
+	struct inode *inode = file_inode(file);
+	fmode_t type = arg == F_RDLCK ? FMODE_READ : FMODE_WRITE;
+	int ret;
+
+	/* No delegation, no lease */
+	if (!nfs4_have_delegation(inode, type))
+		return -EAGAIN;
+	ret = generic_setlease(file, arg, lease, priv);
+	if (ret || nfs4_have_delegation(inode, type))
+		return ret;
+	/* We raced with a delegation return */
+	nfs4_delete_lease(file, priv);
+	return -EAGAIN;
+}
+
+int nfs4_proc_setlease(struct file *file, long arg, struct file_lock **lease,
+		       void **priv)
+{
+	switch (arg) {
+	case F_RDLCK:
+	case F_WRLCK:
+		return nfs4_add_lease(file, arg, lease, priv);
+	case F_UNLCK:
+		return nfs4_delete_lease(file, priv);
+	default:
+		return -EINVAL;
+	}
+}
+
 int nfs4_lock_delegation_recall(struct file_lock *fl, struct nfs4_state *state, const nfs4_stateid *stateid)
 {
 	struct nfs_server *server = NFS_SERVER(state->inode);
@@ -9186,7 +9245,7 @@ static struct rpc_task *_nfs41_proc_sequence(struct nfs_client *clp,
 		.rpc_client = clp->cl_rpcclient,
 		.rpc_message = &msg,
 		.callback_ops = &nfs41_sequence_ops,
-		.flags = RPC_TASK_ASYNC | RPC_TASK_TIMEOUT,
+		.flags = RPC_TASK_ASYNC | RPC_TASK_TIMEOUT | RPC_TASK_MOVEABLE,
 	};
 	struct rpc_task *ret;
 
@@ -9385,7 +9444,7 @@ nfs4_layoutget_handle_exception(struct rpc_task *task,
 {
 	struct inode *inode = lgp->args.inode;
 	struct nfs_server *server = NFS_SERVER(inode);
-	struct pnfs_layout_hdr *lo;
+	struct pnfs_layout_hdr *lo = lgp->lo;
 	int nfs4err = task->tk_status;
 	int err, status = 0;
 	LIST_HEAD(head);
@@ -9437,7 +9496,6 @@ nfs4_layoutget_handle_exception(struct rpc_task *task,
 	case -NFS4ERR_BAD_STATEID:
 		exception->timeout = 0;
 		spin_lock(&inode->i_lock);
-		lo = NFS_I(inode)->layout;
 		/* If the open stateid was bad, then recover it. */
 		if (!lo || test_bit(NFS_LAYOUT_INVALID_STID, &lo->plh_flags) ||
 		    !nfs4_stateid_match_other(&lgp->args.stateid, &lo->plh_stateid)) {
@@ -9509,7 +9567,8 @@ nfs4_proc_layoutget(struct nfs4_layoutget *lgp, long *timeout)
 		.rpc_message = &msg,
 		.callback_ops = &nfs4_layoutget_call_ops,
 		.callback_data = lgp,
-		.flags = RPC_TASK_ASYNC | RPC_TASK_CRED_NOREF,
+		.flags = RPC_TASK_ASYNC | RPC_TASK_CRED_NOREF |
+			 RPC_TASK_MOVEABLE,
 	};
 	struct pnfs_layout_segment *lseg = NULL;
 	struct nfs4_exception exception = {
@@ -9520,9 +9579,6 @@ nfs4_proc_layoutget(struct nfs4_layoutget *lgp, long *timeout)
 
 	dprintk("--> %s\n", __func__);
 
-	/* nfs4_layoutget_release calls pnfs_put_layout_hdr */
-	pnfs_get_layout_hdr(NFS_I(inode)->layout);
-
 	nfs4_init_sequence(&lgp->args.seq_args, &lgp->res.seq_res, 0, 0);
 
 	task = rpc_run_task(&task_setup_data);
@@ -9650,6 +9706,7 @@ int nfs4_proc_layoutreturn(struct nfs4_layoutreturn *lrp, bool sync)
 		.rpc_message = &msg,
 		.callback_ops = &nfs4_layoutreturn_call_ops,
 		.callback_data = lrp,
+		.flags = RPC_TASK_MOVEABLE,
 	};
 	int status = 0;
 
@@ -9804,6 +9861,7 @@ nfs4_proc_layoutcommit(struct nfs4_layoutcommit_data *data, bool sync)
 		.rpc_message = &msg,
 		.callback_ops = &nfs4_layoutcommit_ops,
 		.callback_data = data,
+		.flags = RPC_TASK_MOVEABLE,
 	};
 	struct rpc_task *task;
 	int status = 0;
@@ -10131,7 +10189,7 @@ static int nfs41_free_stateid(struct nfs_server *server,
 		.rpc_client = server->client,
 		.rpc_message = &msg,
 		.callback_ops = &nfs41_free_stateid_ops,
-		.flags = RPC_TASK_ASYNC,
+		.flags = RPC_TASK_ASYNC | RPC_TASK_MOVEABLE,
 	};
 	struct nfs_free_stateid_data *data;
 	struct rpc_task *task;
diff --git a/fs/nfs/pagelist.c b/fs/nfs/pagelist.c
index cf9cc62ec48e..cc232d1f16f2 100644
--- a/fs/nfs/pagelist.c
+++ b/fs/nfs/pagelist.c
@@ -954,6 +954,7 @@ static int nfs_generic_pg_pgios(struct nfs_pageio_descriptor *desc)
 {
 	struct nfs_pgio_header *hdr;
 	int ret;
+	unsigned short task_flags = 0;
 
 	hdr = nfs_pgio_header_alloc(desc->pg_rw_ops);
 	if (!hdr) {
@@ -962,14 +963,17 @@ static int nfs_generic_pg_pgios(struct nfs_pageio_descriptor *desc)
 	}
 	nfs_pgheader_init(desc, hdr, nfs_pgio_header_free);
 	ret = nfs_generic_pgio(desc, hdr);
-	if (ret == 0)
+	if (ret == 0) {
+		if (NFS_SERVER(hdr->inode)->nfs_client->cl_minorversion)
+			task_flags = RPC_TASK_MOVEABLE;
 		ret = nfs_initiate_pgio(NFS_CLIENT(hdr->inode),
 					hdr,
 					hdr->cred,
 					NFS_PROTO(hdr->inode),
 					desc->pg_rpc_callops,
 					desc->pg_ioflags,
-					RPC_TASK_CRED_NOREF);
+					RPC_TASK_CRED_NOREF | task_flags);
+	}
 	return ret;
 }
 
diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
index 2c01ee805306..ef14ea0b6ab8 100644
--- a/fs/nfs/pnfs.c
+++ b/fs/nfs/pnfs.c
@@ -966,10 +966,8 @@ void
 pnfs_set_layout_stateid(struct pnfs_layout_hdr *lo, const nfs4_stateid *new,
 			const struct cred *cred, bool update_barrier)
 {
-	u32 oldseq, newseq, new_barrier = 0;
-
-	oldseq = be32_to_cpu(lo->plh_stateid.seqid);
-	newseq = be32_to_cpu(new->seqid);
+	u32 oldseq = be32_to_cpu(lo->plh_stateid.seqid);
+	u32 newseq = be32_to_cpu(new->seqid);
 
 	if (!pnfs_layout_is_valid(lo)) {
 		pnfs_set_layout_cred(lo, cred);
@@ -979,19 +977,21 @@ pnfs_set_layout_stateid(struct pnfs_layout_hdr *lo, const nfs4_stateid *new,
 		clear_bit(NFS_LAYOUT_INVALID_STID, &lo->plh_flags);
 		return;
 	}
-	if (pnfs_seqid_is_newer(newseq, oldseq)) {
+
+	if (pnfs_seqid_is_newer(newseq, oldseq))
 		nfs4_stateid_copy(&lo->plh_stateid, new);
-		/*
-		 * Because of wraparound, we want to keep the barrier
-		 * "close" to the current seqids.
-		 */
-		new_barrier = newseq - atomic_read(&lo->plh_outstanding);
-	}
-	if (update_barrier)
-		new_barrier = be32_to_cpu(new->seqid);
-	else if (new_barrier == 0)
+
+	if (update_barrier) {
+		pnfs_barrier_update(lo, newseq);
 		return;
-	pnfs_barrier_update(lo, new_barrier);
+	}
+	/*
+	 * Because of wraparound, we want to keep the barrier
+	 * "close" to the current seqids. We really only want to
+	 * get here from a layoutget call.
+	 */
+	if (atomic_read(&lo->plh_outstanding) == 1)
+		 pnfs_barrier_update(lo, be32_to_cpu(lo->plh_stateid.seqid));
 }
 
 static bool
@@ -1128,8 +1128,7 @@ void pnfs_layoutget_free(struct nfs4_layoutget *lgp)
 	size_t max_pages = lgp->args.layout.pglen / PAGE_SIZE;
 
 	nfs4_free_pages(lgp->args.layout.pages, max_pages);
-	if (lgp->args.inode)
-		pnfs_put_layout_hdr(NFS_I(lgp->args.inode)->layout);
+	pnfs_put_layout_hdr(lgp->lo);
 	put_nfs_open_context(lgp->args.ctx);
 	kfree(lgp);
 }
@@ -2014,7 +2013,7 @@ lookup_again:
 	 * If the layout segment list is empty, but there are outstanding
 	 * layoutget calls, then they might be subject to a layoutrecall.
 	 */
-	if (list_empty(&lo->plh_segs) &&
+	if ((list_empty(&lo->plh_segs) || !pnfs_layout_is_valid(lo)) &&
 	    atomic_read(&lo->plh_outstanding) != 0) {
 		spin_unlock(&ino->i_lock);
 		lseg = ERR_PTR(wait_var_event_killable(&lo->plh_outstanding,
@@ -2124,6 +2123,9 @@ lookup_again:
 		goto out_put_layout_hdr;
 	}
 
+	lgp->lo = lo;
+	pnfs_get_layout_hdr(lo);
+
 	lseg = nfs4_proc_layoutget(lgp, &timeout);
 	trace_pnfs_update_layout(ino, pos, count, iomode, lo, lseg,
 				 PNFS_UPDATE_LAYOUT_SEND_LAYOUTGET);
@@ -2255,6 +2257,7 @@ static void _lgopen_prepare_attached(struct nfs4_opendata *data,
 		pnfs_put_layout_hdr(lo);
 		return;
 	}
+	lgp->lo = lo;
 	data->lgp = lgp;
 	data->o_arg.lg_args = &lgp->args;
 	data->o_res.lg_res = &lgp->res;
@@ -2263,6 +2266,7 @@ static void _lgopen_prepare_attached(struct nfs4_opendata *data,
 static void _lgopen_prepare_floating(struct nfs4_opendata *data,
 				     struct nfs_open_context *ctx)
 {
+	struct inode *ino = data->dentry->d_inode;
 	struct pnfs_layout_range rng = {
 		.iomode = (data->o_arg.fmode & FMODE_WRITE) ?
 			  IOMODE_RW: IOMODE_READ,
@@ -2271,7 +2275,7 @@ static void _lgopen_prepare_floating(struct nfs4_opendata *data,
 	};
 	struct nfs4_layoutget *lgp;
 
-	lgp = pnfs_alloc_init_layoutget_args(NULL, ctx, &current_stateid,
+	lgp = pnfs_alloc_init_layoutget_args(ino, ctx, &current_stateid,
 					     &rng, GFP_KERNEL);
 	if (!lgp)
 		return;
@@ -2291,6 +2295,8 @@ void pnfs_lgopen_prepare(struct nfs4_opendata *data,
 	/* Could check on max_ops, but currently hardcoded high enough */
 	if (!nfs_server_capable(data->dir->d_inode, NFS_CAP_LGOPEN))
 		return;
+	if (data->lgp)
+		return;
 	if (data->state)
 		_lgopen_prepare_attached(data, ctx);
 	else
@@ -2330,13 +2336,13 @@ void pnfs_parse_lgopen(struct inode *ino, struct nfs4_layoutget *lgp,
 		}
 		return;
 	}
-	if (!lgp->args.inode) {
+	if (!lgp->lo) {
 		lo = _pnfs_grab_empty_layout(ino, ctx);
 		if (!lo)
 			return;
-		lgp->args.inode = ino;
+		lgp->lo = lo;
 	} else
-		lo = NFS_I(lgp->args.inode)->layout;
+		lo = lgp->lo;
 
 	lseg = pnfs_layout_process(lgp);
 	if (!IS_ERR(lseg)) {
@@ -2349,11 +2355,9 @@ void pnfs_parse_lgopen(struct inode *ino, struct nfs4_layoutget *lgp,
 void nfs4_lgopen_release(struct nfs4_layoutget *lgp)
 {
 	if (lgp != NULL) {
-		struct inode *inode = lgp->args.inode;
-		if (inode) {
-			struct pnfs_layout_hdr *lo = NFS_I(inode)->layout;
-			pnfs_clear_first_layoutget(lo);
-			nfs_layoutget_end(lo);
+		if (lgp->lo) {
+			pnfs_clear_first_layoutget(lgp->lo);
+			nfs_layoutget_end(lgp->lo);
 		}
 		pnfs_layoutget_free(lgp);
 	}
@@ -2362,7 +2366,7 @@ void nfs4_lgopen_release(struct nfs4_layoutget *lgp)
 struct pnfs_layout_segment *
 pnfs_layout_process(struct nfs4_layoutget *lgp)
 {
-	struct pnfs_layout_hdr *lo = NFS_I(lgp->args.inode)->layout;
+	struct pnfs_layout_hdr *lo = lgp->lo;
 	struct nfs4_layoutget_res *res = &lgp->res;
 	struct pnfs_layout_segment *lseg;
 	struct inode *ino = lo->plh_inode;
@@ -2390,11 +2394,13 @@ pnfs_layout_process(struct nfs4_layoutget *lgp)
 		goto out_forget;
 	}
 
+	if (!pnfs_layout_is_valid(lo) && !pnfs_is_first_layoutget(lo))
+		goto out_forget;
+
 	if (nfs4_stateid_match_other(&lo->plh_stateid, &res->stateid)) {
 		/* existing state ID, make sure the sequence number matches. */
 		if (pnfs_layout_stateid_blocked(lo, &res->stateid)) {
-			if (!pnfs_layout_is_valid(lo) &&
-			    pnfs_is_first_layoutget(lo))
+			if (!pnfs_layout_is_valid(lo))
 				lo->plh_barrier = 0;
 			dprintk("%s forget reply due to sequence\n", __func__);
 			goto out_forget;
@@ -2413,8 +2419,6 @@ pnfs_layout_process(struct nfs4_layoutget *lgp)
 		goto out_forget;
 	} else {
 		/* We have a completely new layout */
-		if (!pnfs_is_first_layoutget(lo))
-			goto out_forget;
 		pnfs_set_layout_stateid(lo, &res->stateid, lgp->cred, true);
 	}
 
diff --git a/fs/nfs/pnfs_nfs.c b/fs/nfs/pnfs_nfs.c
index 49d3389bd813..cf19914fec81 100644
--- a/fs/nfs/pnfs_nfs.c
+++ b/fs/nfs/pnfs_nfs.c
@@ -805,19 +805,16 @@ out:
 }
 EXPORT_SYMBOL_GPL(nfs4_pnfs_ds_add);
 
-static void nfs4_wait_ds_connect(struct nfs4_pnfs_ds *ds)
+static int nfs4_wait_ds_connect(struct nfs4_pnfs_ds *ds)
 {
 	might_sleep();
-	wait_on_bit(&ds->ds_state, NFS4DS_CONNECTING,
-			TASK_KILLABLE);
+	return wait_on_bit(&ds->ds_state, NFS4DS_CONNECTING, TASK_KILLABLE);
 }
 
 static void nfs4_clear_ds_conn_bit(struct nfs4_pnfs_ds *ds)
 {
 	smp_mb__before_atomic();
-	clear_bit(NFS4DS_CONNECTING, &ds->ds_state);
-	smp_mb__after_atomic();
-	wake_up_bit(&ds->ds_state, NFS4DS_CONNECTING);
+	clear_and_wake_up_bit(NFS4DS_CONNECTING, &ds->ds_state);
 }
 
 static struct nfs_client *(*get_v3_ds_connect)(
@@ -858,7 +855,7 @@ static int _nfs4_pnfs_v3_ds_connect(struct nfs_server *mds_srv,
 	dprintk("--> %s DS %s\n", __func__, ds->ds_remotestr);
 
 	if (!load_v3_ds_connect())
-		goto out;
+		return -EPROTONOSUPPORT;
 
 	list_for_each_entry(da, &ds->ds_addrs, da_node) {
 		dprintk("%s: DS %s: trying address %s\n",
@@ -993,30 +990,33 @@ int nfs4_pnfs_ds_connect(struct nfs_server *mds_srv, struct nfs4_pnfs_ds *ds,
 {
 	int err;
 
-again:
-	err = 0;
-	if (test_and_set_bit(NFS4DS_CONNECTING, &ds->ds_state) == 0) {
-		if (version == 3) {
-			err = _nfs4_pnfs_v3_ds_connect(mds_srv, ds, timeo,
-						       retrans);
-		} else if (version == 4) {
-			err = _nfs4_pnfs_v4_ds_connect(mds_srv, ds, timeo,
-						       retrans, minor_version);
-		} else {
-			dprintk("%s: unsupported DS version %d\n", __func__,
-				version);
-			err = -EPROTONOSUPPORT;
-		}
+	do {
+		err = nfs4_wait_ds_connect(ds);
+		if (err || ds->ds_clp)
+			goto out;
+		if (nfs4_test_deviceid_unavailable(devid))
+			return -ENODEV;
+	} while (test_and_set_bit(NFS4DS_CONNECTING, &ds->ds_state) != 0);
 
-		nfs4_clear_ds_conn_bit(ds);
-	} else {
-		nfs4_wait_ds_connect(ds);
+	if (ds->ds_clp)
+		goto connect_done;
 
-		/* what was waited on didn't connect AND didn't mark unavail */
-		if (!ds->ds_clp && !nfs4_test_deviceid_unavailable(devid))
-			goto again;
+	switch (version) {
+	case 3:
+		err = _nfs4_pnfs_v3_ds_connect(mds_srv, ds, timeo, retrans);
+		break;
+	case 4:
+		err = _nfs4_pnfs_v4_ds_connect(mds_srv, ds, timeo, retrans,
+					       minor_version);
+		break;
+	default:
+		dprintk("%s: unsupported DS version %d\n", __func__, version);
+		err = -EPROTONOSUPPORT;
 	}
 
+connect_done:
+	nfs4_clear_ds_conn_bit(ds);
+out:
 	/*
 	 * At this point the ds->ds_clp should be ready, but it might have
 	 * hit an error.
diff --git a/fs/nfs/read.c b/fs/nfs/read.c
index d2b6dce1f99f..9f39e0a1a38b 100644
--- a/fs/nfs/read.c
+++ b/fs/nfs/read.c
@@ -74,8 +74,7 @@ void nfs_pageio_init_read(struct nfs_pageio_descriptor *pgio,
 }
 EXPORT_SYMBOL_GPL(nfs_pageio_init_read);
 
-static void nfs_pageio_complete_read(struct nfs_pageio_descriptor *pgio,
-				     struct inode *inode)
+static void nfs_pageio_complete_read(struct nfs_pageio_descriptor *pgio)
 {
 	struct nfs_pgio_mirror *pgm;
 	unsigned long npages;
@@ -86,9 +85,9 @@ static void nfs_pageio_complete_read(struct nfs_pageio_descriptor *pgio,
 	WARN_ON_ONCE(pgio->pg_mirror_count != 1);
 
 	pgm = &pgio->pg_mirrors[0];
-	NFS_I(inode)->read_io += pgm->pg_bytes_written;
+	NFS_I(pgio->pg_inode)->read_io += pgm->pg_bytes_written;
 	npages = (pgm->pg_bytes_written + PAGE_SIZE - 1) >> PAGE_SHIFT;
-	nfs_add_stats(inode, NFSIOS_READPAGES, npages);
+	nfs_add_stats(pgio->pg_inode, NFSIOS_READPAGES, npages);
 }
 
 
@@ -363,22 +362,23 @@ int nfs_readpage(struct file *file, struct page *page)
 	} else
 		desc.ctx = get_nfs_open_context(nfs_file_open_context(file));
 
+	xchg(&desc.ctx->error, 0);
 	if (!IS_SYNC(inode)) {
 		ret = nfs_readpage_from_fscache(desc.ctx, inode, page);
 		if (ret == 0)
-			goto out;
+			goto out_wait;
 	}
 
-	xchg(&desc.ctx->error, 0);
 	nfs_pageio_init_read(&desc.pgio, inode, false,
 			     &nfs_async_read_completion_ops);
 
 	ret = readpage_async_filler(&desc, page);
+	if (ret)
+		goto out;
 
-	if (!ret)
-		nfs_pageio_complete_read(&desc.pgio, inode);
-
+	nfs_pageio_complete_read(&desc.pgio);
 	ret = desc.pgio.pg_error < 0 ? desc.pgio.pg_error : 0;
+out_wait:
 	if (!ret) {
 		ret = wait_on_page_locked_killable(page);
 		if (!PageUptodate(page) && !ret)
@@ -430,7 +430,7 @@ int nfs_readpages(struct file *file, struct address_space *mapping,
 
 	ret = read_cache_pages(mapping, pages, readpage_async_filler, &desc);
 
-	nfs_pageio_complete_read(&desc.pgio, inode);
+	nfs_pageio_complete_read(&desc.pgio);
 
 read_complete:
 	put_nfs_open_context(desc.ctx);
diff --git a/fs/nfs/write.c b/fs/nfs/write.c
index 3bf82178166a..eae9bf114041 100644
--- a/fs/nfs/write.c
+++ b/fs/nfs/write.c
@@ -1810,6 +1810,7 @@ nfs_commit_list(struct inode *inode, struct list_head *head, int how,
 		struct nfs_commit_info *cinfo)
 {
 	struct nfs_commit_data	*data;
+	unsigned short task_flags = 0;
 
 	/* another commit raced with us */
 	if (list_empty(head))
@@ -1820,8 +1821,11 @@ nfs_commit_list(struct inode *inode, struct list_head *head, int how,
 	/* Set up the argument struct */
 	nfs_init_commit(data, head, NULL, cinfo);
 	atomic_inc(&cinfo->mds->rpcs_out);
+	if (NFS_SERVER(inode)->nfs_client->cl_minorversion)
+		task_flags = RPC_TASK_MOVEABLE;
 	return nfs_initiate_commit(NFS_CLIENT(inode), data, NFS_PROTO(inode),
-				   data->mds_ops, how, RPC_TASK_CRED_NOREF);
+				   data->mds_ops, how,
+				   RPC_TASK_CRED_NOREF | task_flags);
 }
 
 /*
diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h
index ffba254d2098..ce6474594872 100644
--- a/include/linux/nfs_fs.h
+++ b/include/linux/nfs_fs.h
@@ -84,6 +84,7 @@ struct nfs_open_context {
 #define NFS_CONTEXT_RESEND_WRITES	(1)
 #define NFS_CONTEXT_BAD			(2)
 #define NFS_CONTEXT_UNLOCK	(3)
+#define NFS_CONTEXT_FILE_OPEN		(4)
 	int error;
 
 	struct list_head list;
diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h
index 717ecc87c9e7..e9698b6278a5 100644
--- a/include/linux/nfs_xdr.h
+++ b/include/linux/nfs_xdr.h
@@ -277,6 +277,7 @@ struct nfs4_layoutget {
 	struct nfs4_layoutget_args args;
 	struct nfs4_layoutget_res res;
 	const struct cred *cred;
+	struct pnfs_layout_hdr *lo;
 	gfp_t gfp_flags;
 };
 
diff --git a/include/linux/sunrpc/clnt.h b/include/linux/sunrpc/clnt.h
index 02e7a5863d28..8b5d5c97553e 100644
--- a/include/linux/sunrpc/clnt.h
+++ b/include/linux/sunrpc/clnt.h
@@ -29,6 +29,7 @@
 #include <linux/sunrpc/xprtmultipath.h>
 
 struct rpc_inode;
+struct rpc_sysfs_client;
 
 /*
  * The high-level client handle
@@ -71,6 +72,7 @@ struct rpc_clnt {
 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
 	struct dentry		*cl_debugfs;	/* debugfs directory */
 #endif
+	struct rpc_sysfs_client *cl_sysfs;	/* sysfs directory */
 	/* cl_work is only needed after cl_xpi is no longer used,
 	 * and that are of similar size
 	 */
diff --git a/include/linux/sunrpc/sched.h b/include/linux/sunrpc/sched.h
index df696efdd675..a237b8dbf608 100644
--- a/include/linux/sunrpc/sched.h
+++ b/include/linux/sunrpc/sched.h
@@ -121,6 +121,7 @@ struct rpc_task_setup {
  */
 #define RPC_TASK_ASYNC		0x0001		/* is an async task */
 #define RPC_TASK_SWAPPER	0x0002		/* is swapping in/out */
+#define RPC_TASK_MOVEABLE	0x0004		/* nfs4.1+ rpc tasks */
 #define RPC_TASK_NULLCREDS	0x0010		/* Use AUTH_NULL credential */
 #define RPC_CALL_MAJORSEEN	0x0020		/* major timeout seen */
 #define RPC_TASK_ROOTCREDS	0x0040		/* force root creds */
@@ -139,6 +140,7 @@ struct rpc_task_setup {
 #define RPC_IS_SOFT(t)		((t)->tk_flags & (RPC_TASK_SOFT|RPC_TASK_TIMEOUT))
 #define RPC_IS_SOFTCONN(t)	((t)->tk_flags & RPC_TASK_SOFTCONN)
 #define RPC_WAS_SENT(t)		((t)->tk_flags & RPC_TASK_SENT)
+#define RPC_IS_MOVEABLE(t)	((t)->tk_flags & RPC_TASK_MOVEABLE)
 
 #define RPC_TASK_RUNNING	0
 #define RPC_TASK_QUEUED		1
diff --git a/include/linux/sunrpc/xprt.h b/include/linux/sunrpc/xprt.h
index 61b622e334ee..c8c39f22d3b1 100644
--- a/include/linux/sunrpc/xprt.h
+++ b/include/linux/sunrpc/xprt.h
@@ -53,6 +53,7 @@ enum rpc_display_format_t {
 
 struct rpc_task;
 struct rpc_xprt;
+struct xprt_class;
 struct seq_file;
 struct svc_serv;
 struct net;
@@ -182,9 +183,11 @@ enum xprt_transports {
 	XPRT_TRANSPORT_LOCAL	= 257,
 };
 
+struct rpc_sysfs_xprt;
 struct rpc_xprt {
 	struct kref		kref;		/* Reference count */
 	const struct rpc_xprt_ops *ops;		/* transport methods */
+	unsigned int		id;		/* transport id */
 
 	const struct rpc_timeout *timeout;	/* timeout parms */
 	struct sockaddr_storage	addr;		/* server address */
@@ -288,6 +291,9 @@ struct rpc_xprt {
 	atomic_t		inject_disconnect;
 #endif
 	struct rcu_head		rcu;
+	const struct xprt_class	*xprt_class;
+	struct rpc_sysfs_xprt	*xprt_sysfs;
+	bool			main; /*mark if this is the 1st transport */
 };
 
 #if defined(CONFIG_SUNRPC_BACKCHANNEL)
@@ -370,6 +376,7 @@ struct rpc_xprt *	xprt_alloc(struct net *net, size_t size,
 void			xprt_free(struct rpc_xprt *);
 void			xprt_add_backlog(struct rpc_xprt *xprt, struct rpc_task *task);
 bool			xprt_wake_up_backlog(struct rpc_xprt *xprt, struct rpc_rqst *req);
+void			xprt_cleanup_ids(void);
 
 static inline int
 xprt_enable_swap(struct rpc_xprt *xprt)
@@ -408,6 +415,7 @@ void			xprt_conditional_disconnect(struct rpc_xprt *xprt, unsigned int cookie);
 
 bool			xprt_lock_connect(struct rpc_xprt *, struct rpc_task *, void *);
 void			xprt_unlock_connect(struct rpc_xprt *, void *);
+void			xprt_release_write(struct rpc_xprt *, struct rpc_task *);
 
 /*
  * Reserved bit positions in xprt->state
@@ -419,6 +427,8 @@ void			xprt_unlock_connect(struct rpc_xprt *, void *);
 #define XPRT_BOUND		(4)
 #define XPRT_BINDING		(5)
 #define XPRT_CLOSING		(6)
+#define XPRT_OFFLINE		(7)
+#define XPRT_REMOVE		(8)
 #define XPRT_CONGESTED		(9)
 #define XPRT_CWND_WAIT		(10)
 #define XPRT_WRITE_SPACE	(11)
diff --git a/include/linux/sunrpc/xprtmultipath.h b/include/linux/sunrpc/xprtmultipath.h
index c6cce3fbf29d..b19addc8b715 100644
--- a/include/linux/sunrpc/xprtmultipath.h
+++ b/include/linux/sunrpc/xprtmultipath.h
@@ -10,10 +10,12 @@
 #define _NET_SUNRPC_XPRTMULTIPATH_H
 
 struct rpc_xprt_iter_ops;
+struct rpc_sysfs_xprt_switch;
 struct rpc_xprt_switch {
 	spinlock_t		xps_lock;
 	struct kref		xps_kref;
 
+	unsigned int		xps_id;
 	unsigned int		xps_nxprts;
 	unsigned int		xps_nactive;
 	atomic_long_t		xps_queuelen;
@@ -23,6 +25,7 @@ struct rpc_xprt_switch {
 
 	const struct rpc_xprt_iter_ops *xps_iter_ops;
 
+	struct rpc_sysfs_xprt_switch *xps_sysfs;
 	struct rcu_head		xps_rcu;
 };
 
@@ -71,4 +74,7 @@ extern struct rpc_xprt *xprt_iter_get_next(struct rpc_xprt_iter *xpi);
 
 extern bool rpc_xprt_switch_has_addr(struct rpc_xprt_switch *xps,
 		const struct sockaddr *sap);
+
+extern void xprt_multipath_cleanup_ids(void);
+
 #endif
diff --git a/include/linux/sunrpc/xprtsock.h b/include/linux/sunrpc/xprtsock.h
index 3c1423ee74b4..8c2a712cb242 100644
--- a/include/linux/sunrpc/xprtsock.h
+++ b/include/linux/sunrpc/xprtsock.h
@@ -10,6 +10,7 @@
 
 int		init_socket_xprt(void);
 void		cleanup_socket_xprt(void);
+unsigned short	get_srcport(struct rpc_xprt *);
 
 #define RPC_MIN_RESVPORT	(1U)
 #define RPC_MAX_RESVPORT	(65535U)
diff --git a/net/sunrpc/Makefile b/net/sunrpc/Makefile
index 9488600451e8..1c8de397d6ad 100644
--- a/net/sunrpc/Makefile
+++ b/net/sunrpc/Makefile
@@ -12,7 +12,7 @@ sunrpc-y := clnt.o xprt.o socklib.o xprtsock.o sched.o \
 	    auth.o auth_null.o auth_unix.o \
 	    svc.o svcsock.o svcauth.o svcauth_unix.o \
 	    addr.o rpcb_clnt.o timer.o xdr.o \
-	    sunrpc_syms.o cache.o rpc_pipe.o \
+	    sunrpc_syms.o cache.o rpc_pipe.o sysfs.o \
 	    svc_xprt.o \
 	    xprtmultipath.o
 sunrpc-$(CONFIG_SUNRPC_DEBUG) += debugfs.o
diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c
index 42623d6b8f0e..8b4de70e8ead 100644
--- a/net/sunrpc/clnt.c
+++ b/net/sunrpc/clnt.c
@@ -41,6 +41,7 @@
 #include <trace/events/sunrpc.h>
 
 #include "sunrpc.h"
+#include "sysfs.h"
 #include "netns.h"
 
 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
@@ -327,6 +328,7 @@ err_auth:
 out:
 	if (pipefs_sb)
 		rpc_put_sb_net(net);
+	rpc_sysfs_client_destroy(clnt);
 	rpc_clnt_debugfs_unregister(clnt);
 	return err;
 }
@@ -410,6 +412,7 @@ static struct rpc_clnt * rpc_new_client(const struct rpc_create_args *args,
 	}
 
 	rpc_clnt_set_transport(clnt, xprt, timeout);
+	xprt->main = true;
 	xprt_iter_init(&clnt->cl_xpi, xps);
 	xprt_switch_put(xps);
 
@@ -423,6 +426,7 @@ static struct rpc_clnt * rpc_new_client(const struct rpc_create_args *args,
 	/* save the nodename */
 	rpc_clnt_set_nodename(clnt, nodename);
 
+	rpc_sysfs_client_setup(clnt, xps, rpc_net_ns(clnt));
 	err = rpc_client_register(clnt, args->authflavor, args->client_name);
 	if (err)
 		goto out_no_path;
@@ -733,6 +737,7 @@ int rpc_switch_client_transport(struct rpc_clnt *clnt,
 
 	rpc_unregister_client(clnt);
 	__rpc_clnt_remove_pipedir(clnt);
+	rpc_sysfs_client_destroy(clnt);
 	rpc_clnt_debugfs_unregister(clnt);
 
 	/*
@@ -879,6 +884,7 @@ static void rpc_free_client_work(struct work_struct *work)
 	 * so they cannot be called in rpciod, so they are handled separately
 	 * here.
 	 */
+	rpc_sysfs_client_destroy(clnt);
 	rpc_clnt_debugfs_unregister(clnt);
 	rpc_free_clid(clnt);
 	rpc_clnt_remove_pipedir(clnt);
@@ -2100,6 +2106,30 @@ call_connect_status(struct rpc_task *task)
 	case -ENOTCONN:
 	case -EAGAIN:
 	case -ETIMEDOUT:
+		if (!(task->tk_flags & RPC_TASK_NO_ROUND_ROBIN) &&
+		    (task->tk_flags & RPC_TASK_MOVEABLE) &&
+		    test_bit(XPRT_REMOVE, &xprt->state)) {
+			struct rpc_xprt *saved = task->tk_xprt;
+			struct rpc_xprt_switch *xps;
+
+			rcu_read_lock();
+			xps = xprt_switch_get(rcu_dereference(clnt->cl_xpi.xpi_xpswitch));
+			rcu_read_unlock();
+			if (xps->xps_nxprts > 1) {
+				long value;
+
+				xprt_release(task);
+				value = atomic_long_dec_return(&xprt->queuelen);
+				if (value == 0)
+					rpc_xprt_switch_remove_xprt(xps, saved);
+				xprt_put(saved);
+				task->tk_xprt = NULL;
+				task->tk_action = call_start;
+			}
+			xprt_switch_put(xps);
+			if (!task->tk_xprt)
+				return;
+		}
 		goto out_retry;
 	case -ENOBUFS:
 		rpc_delay(task, HZ >> 2);
diff --git a/net/sunrpc/sched.c b/net/sunrpc/sched.c
index 39ed0e0afe6d..c045f63d11fa 100644
--- a/net/sunrpc/sched.c
+++ b/net/sunrpc/sched.c
@@ -592,10 +592,20 @@ static struct rpc_task *__rpc_find_next_queued_priority(struct rpc_wait_queue *q
 	struct rpc_task *task;
 
 	/*
+	 * Service the privileged queue.
+	 */
+	q = &queue->tasks[RPC_NR_PRIORITY - 1];
+	if (queue->maxpriority > RPC_PRIORITY_PRIVILEGED && !list_empty(q)) {
+		task = list_first_entry(q, struct rpc_task, u.tk_wait.list);
+		goto out;
+	}
+
+	/*
 	 * Service a batch of tasks from a single owner.
 	 */
 	q = &queue->tasks[queue->priority];
-	if (!list_empty(q) && --queue->nr) {
+	if (!list_empty(q) && queue->nr) {
+		queue->nr--;
 		task = list_first_entry(q, struct rpc_task, u.tk_wait.list);
 		goto out;
 	}
diff --git a/net/sunrpc/sunrpc_syms.c b/net/sunrpc/sunrpc_syms.c
index 236fadc4a439..691c0000e9ea 100644
--- a/net/sunrpc/sunrpc_syms.c
+++ b/net/sunrpc/sunrpc_syms.c
@@ -24,6 +24,7 @@
 #include <linux/sunrpc/xprtsock.h>
 
 #include "sunrpc.h"
+#include "sysfs.h"
 #include "netns.h"
 
 unsigned int sunrpc_net_id;
@@ -103,6 +104,10 @@ init_sunrpc(void)
 	if (err)
 		goto out4;
 
+	err = rpc_sysfs_init();
+	if (err)
+		goto out5;
+
 	sunrpc_debugfs_init();
 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
 	rpc_register_sysctl();
@@ -111,6 +116,8 @@ init_sunrpc(void)
 	init_socket_xprt();	/* clnt sock transport */
 	return 0;
 
+out5:
+	unregister_rpc_pipefs();
 out4:
 	unregister_pernet_subsys(&sunrpc_net_ops);
 out3:
@@ -124,7 +131,10 @@ out:
 static void __exit
 cleanup_sunrpc(void)
 {
+	rpc_sysfs_exit();
 	rpc_cleanup_clids();
+	xprt_cleanup_ids();
+	xprt_multipath_cleanup_ids();
 	rpcauth_remove_module();
 	cleanup_socket_xprt();
 	svc_cleanup_xprt_sock();
diff --git a/net/sunrpc/sysfs.c b/net/sunrpc/sysfs.c
new file mode 100644
index 000000000000..64da3bfd28e6
--- /dev/null
+++ b/net/sunrpc/sysfs.c
@@ -0,0 +1,588 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2020 Anna Schumaker <Anna.Schumaker@Netapp.com>
+ */
+#include <linux/sunrpc/clnt.h>
+#include <linux/kobject.h>
+#include <linux/sunrpc/addr.h>
+#include <linux/sunrpc/xprtsock.h>
+
+#include "sysfs.h"
+
+struct xprt_addr {
+	const char *addr;
+	struct rcu_head rcu;
+};
+
+static void free_xprt_addr(struct rcu_head *head)
+{
+	struct xprt_addr *addr = container_of(head, struct xprt_addr, rcu);
+
+	kfree(addr->addr);
+	kfree(addr);
+}
+
+static struct kset *rpc_sunrpc_kset;
+static struct kobject *rpc_sunrpc_client_kobj, *rpc_sunrpc_xprt_switch_kobj;
+
+static void rpc_sysfs_object_release(struct kobject *kobj)
+{
+	kfree(kobj);
+}
+
+static const struct kobj_ns_type_operations *
+rpc_sysfs_object_child_ns_type(struct kobject *kobj)
+{
+	return &net_ns_type_operations;
+}
+
+static struct kobj_type rpc_sysfs_object_type = {
+	.release = rpc_sysfs_object_release,
+	.sysfs_ops = &kobj_sysfs_ops,
+	.child_ns_type = rpc_sysfs_object_child_ns_type,
+};
+
+static struct kobject *rpc_sysfs_object_alloc(const char *name,
+					      struct kset *kset,
+					      struct kobject *parent)
+{
+	struct kobject *kobj;
+
+	kobj = kzalloc(sizeof(*kobj), GFP_KERNEL);
+	if (kobj) {
+		kobj->kset = kset;
+		if (kobject_init_and_add(kobj, &rpc_sysfs_object_type,
+					 parent, "%s", name) == 0)
+			return kobj;
+		kobject_put(kobj);
+	}
+	return NULL;
+}
+
+static inline struct rpc_xprt *
+rpc_sysfs_xprt_kobj_get_xprt(struct kobject *kobj)
+{
+	struct rpc_sysfs_xprt *x = container_of(kobj,
+		struct rpc_sysfs_xprt, kobject);
+
+	return xprt_get(x->xprt);
+}
+
+static inline struct rpc_xprt_switch *
+rpc_sysfs_xprt_kobj_get_xprt_switch(struct kobject *kobj)
+{
+	struct rpc_sysfs_xprt *x = container_of(kobj,
+		struct rpc_sysfs_xprt, kobject);
+
+	return xprt_switch_get(x->xprt_switch);
+}
+
+static inline struct rpc_xprt_switch *
+rpc_sysfs_xprt_switch_kobj_get_xprt(struct kobject *kobj)
+{
+	struct rpc_sysfs_xprt_switch *x = container_of(kobj,
+		struct rpc_sysfs_xprt_switch, kobject);
+
+	return xprt_switch_get(x->xprt_switch);
+}
+
+static ssize_t rpc_sysfs_xprt_dstaddr_show(struct kobject *kobj,
+					   struct kobj_attribute *attr,
+					   char *buf)
+{
+	struct rpc_xprt *xprt = rpc_sysfs_xprt_kobj_get_xprt(kobj);
+	ssize_t ret;
+
+	if (!xprt)
+		return 0;
+	ret = sprintf(buf, "%s\n", xprt->address_strings[RPC_DISPLAY_ADDR]);
+	xprt_put(xprt);
+	return ret + 1;
+}
+
+static ssize_t rpc_sysfs_xprt_info_show(struct kobject *kobj,
+					struct kobj_attribute *attr,
+					char *buf)
+{
+	struct rpc_xprt *xprt = rpc_sysfs_xprt_kobj_get_xprt(kobj);
+	ssize_t ret;
+
+	if (!xprt)
+		return 0;
+
+	ret = sprintf(buf, "last_used=%lu\ncur_cong=%lu\ncong_win=%lu\n"
+		       "max_num_slots=%u\nmin_num_slots=%u\nnum_reqs=%u\n"
+		       "binding_q_len=%u\nsending_q_len=%u\npending_q_len=%u\n"
+		       "backlog_q_len=%u\nmain_xprt=%d\nsrc_port=%u\n"
+		       "tasks_queuelen=%ld\n",
+		       xprt->last_used, xprt->cong, xprt->cwnd, xprt->max_reqs,
+		       xprt->min_reqs, xprt->num_reqs, xprt->binding.qlen,
+		       xprt->sending.qlen, xprt->pending.qlen,
+		       xprt->backlog.qlen, xprt->main,
+		       (xprt->xprt_class->ident == XPRT_TRANSPORT_TCP) ?
+		       get_srcport(xprt) : 0,
+		       atomic_long_read(&xprt->queuelen));
+	xprt_put(xprt);
+	return ret + 1;
+}
+
+static ssize_t rpc_sysfs_xprt_state_show(struct kobject *kobj,
+					 struct kobj_attribute *attr,
+					 char *buf)
+{
+	struct rpc_xprt *xprt = rpc_sysfs_xprt_kobj_get_xprt(kobj);
+	ssize_t ret;
+	int locked, connected, connecting, close_wait, bound, binding,
+	    closing, congested, cwnd_wait, write_space, offline, remove;
+
+	if (!xprt)
+		return 0;
+
+	if (!xprt->state) {
+		ret = sprintf(buf, "state=CLOSED\n");
+	} else {
+		locked = test_bit(XPRT_LOCKED, &xprt->state);
+		connected = test_bit(XPRT_CONNECTED, &xprt->state);
+		connecting = test_bit(XPRT_CONNECTING, &xprt->state);
+		close_wait = test_bit(XPRT_CLOSE_WAIT, &xprt->state);
+		bound = test_bit(XPRT_BOUND, &xprt->state);
+		binding = test_bit(XPRT_BINDING, &xprt->state);
+		closing = test_bit(XPRT_CLOSING, &xprt->state);
+		congested = test_bit(XPRT_CONGESTED, &xprt->state);
+		cwnd_wait = test_bit(XPRT_CWND_WAIT, &xprt->state);
+		write_space = test_bit(XPRT_WRITE_SPACE, &xprt->state);
+		offline = test_bit(XPRT_OFFLINE, &xprt->state);
+		remove = test_bit(XPRT_REMOVE, &xprt->state);
+
+		ret = sprintf(buf, "state=%s %s %s %s %s %s %s %s %s %s %s %s\n",
+			      locked ? "LOCKED" : "",
+			      connected ? "CONNECTED" : "",
+			      connecting ? "CONNECTING" : "",
+			      close_wait ? "CLOSE_WAIT" : "",
+			      bound ? "BOUND" : "",
+			      binding ? "BOUNDING" : "",
+			      closing ? "CLOSING" : "",
+			      congested ? "CONGESTED" : "",
+			      cwnd_wait ? "CWND_WAIT" : "",
+			      write_space ? "WRITE_SPACE" : "",
+			      offline ? "OFFLINE" : "",
+			      remove ? "REMOVE" : "");
+	}
+
+	xprt_put(xprt);
+	return ret + 1;
+}
+
+static ssize_t rpc_sysfs_xprt_switch_info_show(struct kobject *kobj,
+					       struct kobj_attribute *attr,
+					       char *buf)
+{
+	struct rpc_xprt_switch *xprt_switch =
+		rpc_sysfs_xprt_switch_kobj_get_xprt(kobj);
+	ssize_t ret;
+
+	if (!xprt_switch)
+		return 0;
+	ret = sprintf(buf, "num_xprts=%u\nnum_active=%u\nqueue_len=%ld\n",
+		      xprt_switch->xps_nxprts, xprt_switch->xps_nactive,
+		      atomic_long_read(&xprt_switch->xps_queuelen));
+	xprt_switch_put(xprt_switch);
+	return ret + 1;
+}
+
+static ssize_t rpc_sysfs_xprt_dstaddr_store(struct kobject *kobj,
+					    struct kobj_attribute *attr,
+					    const char *buf, size_t count)
+{
+	struct rpc_xprt *xprt = rpc_sysfs_xprt_kobj_get_xprt(kobj);
+	struct sockaddr *saddr;
+	char *dst_addr;
+	int port;
+	struct xprt_addr *saved_addr;
+	size_t buf_len;
+
+	if (!xprt)
+		return 0;
+	if (!(xprt->xprt_class->ident == XPRT_TRANSPORT_TCP ||
+	      xprt->xprt_class->ident == XPRT_TRANSPORT_RDMA)) {
+		xprt_put(xprt);
+		return -EOPNOTSUPP;
+	}
+
+	if (wait_on_bit_lock(&xprt->state, XPRT_LOCKED, TASK_KILLABLE)) {
+		count = -EINTR;
+		goto out_put;
+	}
+	saddr = (struct sockaddr *)&xprt->addr;
+	port = rpc_get_port(saddr);
+
+	/* buf_len is the len until the first occurence of either
+	 * '\n' or '\0'
+	 */
+	buf_len = strcspn(buf, "\n");
+
+	dst_addr = kstrndup(buf, buf_len, GFP_KERNEL);
+	if (!dst_addr)
+		goto out_err;
+	saved_addr = kzalloc(sizeof(*saved_addr), GFP_KERNEL);
+	if (!saved_addr)
+		goto out_err_free;
+	saved_addr->addr =
+		rcu_dereference_raw(xprt->address_strings[RPC_DISPLAY_ADDR]);
+	rcu_assign_pointer(xprt->address_strings[RPC_DISPLAY_ADDR], dst_addr);
+	call_rcu(&saved_addr->rcu, free_xprt_addr);
+	xprt->addrlen = rpc_pton(xprt->xprt_net, buf, buf_len, saddr,
+				 sizeof(*saddr));
+	rpc_set_port(saddr, port);
+
+	xprt_force_disconnect(xprt);
+out:
+	xprt_release_write(xprt, NULL);
+out_put:
+	xprt_put(xprt);
+	return count;
+out_err_free:
+	kfree(dst_addr);
+out_err:
+	count = -ENOMEM;
+	goto out;
+}
+
+static ssize_t rpc_sysfs_xprt_state_change(struct kobject *kobj,
+					   struct kobj_attribute *attr,
+					   const char *buf, size_t count)
+{
+	struct rpc_xprt *xprt = rpc_sysfs_xprt_kobj_get_xprt(kobj);
+	int offline = 0, online = 0, remove = 0;
+	struct rpc_xprt_switch *xps = rpc_sysfs_xprt_kobj_get_xprt_switch(kobj);
+
+	if (!xprt)
+		return 0;
+
+	if (!strncmp(buf, "offline", 7))
+		offline = 1;
+	else if (!strncmp(buf, "online", 6))
+		online = 1;
+	else if (!strncmp(buf, "remove", 6))
+		remove = 1;
+	else
+		return -EINVAL;
+
+	if (wait_on_bit_lock(&xprt->state, XPRT_LOCKED, TASK_KILLABLE)) {
+		count = -EINTR;
+		goto out_put;
+	}
+	if (xprt->main) {
+		count = -EINVAL;
+		goto release_tasks;
+	}
+	if (offline) {
+		set_bit(XPRT_OFFLINE, &xprt->state);
+		spin_lock(&xps->xps_lock);
+		xps->xps_nactive--;
+		spin_unlock(&xps->xps_lock);
+	} else if (online) {
+		clear_bit(XPRT_OFFLINE, &xprt->state);
+		spin_lock(&xps->xps_lock);
+		xps->xps_nactive++;
+		spin_unlock(&xps->xps_lock);
+	} else if (remove) {
+		if (test_bit(XPRT_OFFLINE, &xprt->state)) {
+			set_bit(XPRT_REMOVE, &xprt->state);
+			xprt_force_disconnect(xprt);
+			if (test_bit(XPRT_CONNECTED, &xprt->state)) {
+				if (!xprt->sending.qlen &&
+				    !xprt->pending.qlen &&
+				    !xprt->backlog.qlen &&
+				    !atomic_long_read(&xprt->queuelen))
+					rpc_xprt_switch_remove_xprt(xps, xprt);
+			}
+		} else {
+			count = -EINVAL;
+		}
+	}
+
+release_tasks:
+	xprt_release_write(xprt, NULL);
+out_put:
+	xprt_put(xprt);
+	xprt_switch_put(xps);
+	return count;
+}
+
+int rpc_sysfs_init(void)
+{
+	rpc_sunrpc_kset = kset_create_and_add("sunrpc", NULL, kernel_kobj);
+	if (!rpc_sunrpc_kset)
+		return -ENOMEM;
+	rpc_sunrpc_client_kobj =
+		rpc_sysfs_object_alloc("rpc-clients", rpc_sunrpc_kset, NULL);
+	if (!rpc_sunrpc_client_kobj)
+		goto err_client;
+	rpc_sunrpc_xprt_switch_kobj =
+		rpc_sysfs_object_alloc("xprt-switches", rpc_sunrpc_kset, NULL);
+	if (!rpc_sunrpc_xprt_switch_kobj)
+		goto err_switch;
+	return 0;
+err_switch:
+	kobject_put(rpc_sunrpc_client_kobj);
+	rpc_sunrpc_client_kobj = NULL;
+err_client:
+	kset_unregister(rpc_sunrpc_kset);
+	rpc_sunrpc_kset = NULL;
+	return -ENOMEM;
+}
+
+static void rpc_sysfs_client_release(struct kobject *kobj)
+{
+	struct rpc_sysfs_client *c;
+
+	c = container_of(kobj, struct rpc_sysfs_client, kobject);
+	kfree(c);
+}
+
+static void rpc_sysfs_xprt_switch_release(struct kobject *kobj)
+{
+	struct rpc_sysfs_xprt_switch *xprt_switch;
+
+	xprt_switch = container_of(kobj, struct rpc_sysfs_xprt_switch, kobject);
+	kfree(xprt_switch);
+}
+
+static void rpc_sysfs_xprt_release(struct kobject *kobj)
+{
+	struct rpc_sysfs_xprt *xprt;
+
+	xprt = container_of(kobj, struct rpc_sysfs_xprt, kobject);
+	kfree(xprt);
+}
+
+static const void *rpc_sysfs_client_namespace(struct kobject *kobj)
+{
+	return container_of(kobj, struct rpc_sysfs_client, kobject)->net;
+}
+
+static const void *rpc_sysfs_xprt_switch_namespace(struct kobject *kobj)
+{
+	return container_of(kobj, struct rpc_sysfs_xprt_switch, kobject)->net;
+}
+
+static const void *rpc_sysfs_xprt_namespace(struct kobject *kobj)
+{
+	return container_of(kobj, struct rpc_sysfs_xprt,
+			    kobject)->xprt->xprt_net;
+}
+
+static struct kobj_attribute rpc_sysfs_xprt_dstaddr = __ATTR(dstaddr,
+	0644, rpc_sysfs_xprt_dstaddr_show, rpc_sysfs_xprt_dstaddr_store);
+
+static struct kobj_attribute rpc_sysfs_xprt_info = __ATTR(xprt_info,
+	0444, rpc_sysfs_xprt_info_show, NULL);
+
+static struct kobj_attribute rpc_sysfs_xprt_change_state = __ATTR(xprt_state,
+	0644, rpc_sysfs_xprt_state_show, rpc_sysfs_xprt_state_change);
+
+static struct attribute *rpc_sysfs_xprt_attrs[] = {
+	&rpc_sysfs_xprt_dstaddr.attr,
+	&rpc_sysfs_xprt_info.attr,
+	&rpc_sysfs_xprt_change_state.attr,
+	NULL,
+};
+
+static struct kobj_attribute rpc_sysfs_xprt_switch_info =
+	__ATTR(xprt_switch_info, 0444, rpc_sysfs_xprt_switch_info_show, NULL);
+
+static struct attribute *rpc_sysfs_xprt_switch_attrs[] = {
+	&rpc_sysfs_xprt_switch_info.attr,
+	NULL,
+};
+
+static struct kobj_type rpc_sysfs_client_type = {
+	.release = rpc_sysfs_client_release,
+	.sysfs_ops = &kobj_sysfs_ops,
+	.namespace = rpc_sysfs_client_namespace,
+};
+
+static struct kobj_type rpc_sysfs_xprt_switch_type = {
+	.release = rpc_sysfs_xprt_switch_release,
+	.default_attrs = rpc_sysfs_xprt_switch_attrs,
+	.sysfs_ops = &kobj_sysfs_ops,
+	.namespace = rpc_sysfs_xprt_switch_namespace,
+};
+
+static struct kobj_type rpc_sysfs_xprt_type = {
+	.release = rpc_sysfs_xprt_release,
+	.default_attrs = rpc_sysfs_xprt_attrs,
+	.sysfs_ops = &kobj_sysfs_ops,
+	.namespace = rpc_sysfs_xprt_namespace,
+};
+
+void rpc_sysfs_exit(void)
+{
+	kobject_put(rpc_sunrpc_client_kobj);
+	kobject_put(rpc_sunrpc_xprt_switch_kobj);
+	kset_unregister(rpc_sunrpc_kset);
+}
+
+static struct rpc_sysfs_client *rpc_sysfs_client_alloc(struct kobject *parent,
+						       struct net *net,
+						       int clid)
+{
+	struct rpc_sysfs_client *p;
+
+	p = kzalloc(sizeof(*p), GFP_KERNEL);
+	if (p) {
+		p->net = net;
+		p->kobject.kset = rpc_sunrpc_kset;
+		if (kobject_init_and_add(&p->kobject, &rpc_sysfs_client_type,
+					 parent, "clnt-%d", clid) == 0)
+			return p;
+		kobject_put(&p->kobject);
+	}
+	return NULL;
+}
+
+static struct rpc_sysfs_xprt_switch *
+rpc_sysfs_xprt_switch_alloc(struct kobject *parent,
+			    struct rpc_xprt_switch *xprt_switch,
+			    struct net *net,
+			    gfp_t gfp_flags)
+{
+	struct rpc_sysfs_xprt_switch *p;
+
+	p = kzalloc(sizeof(*p), gfp_flags);
+	if (p) {
+		p->net = net;
+		p->kobject.kset = rpc_sunrpc_kset;
+		if (kobject_init_and_add(&p->kobject,
+					 &rpc_sysfs_xprt_switch_type,
+					 parent, "switch-%d",
+					 xprt_switch->xps_id) == 0)
+			return p;
+		kobject_put(&p->kobject);
+	}
+	return NULL;
+}
+
+static struct rpc_sysfs_xprt *rpc_sysfs_xprt_alloc(struct kobject *parent,
+						   struct rpc_xprt *xprt,
+						   gfp_t gfp_flags)
+{
+	struct rpc_sysfs_xprt *p;
+
+	p = kzalloc(sizeof(*p), gfp_flags);
+	if (!p)
+		goto out;
+	p->kobject.kset = rpc_sunrpc_kset;
+	if (kobject_init_and_add(&p->kobject, &rpc_sysfs_xprt_type,
+				 parent, "xprt-%d-%s", xprt->id,
+				 xprt->address_strings[RPC_DISPLAY_PROTO]) == 0)
+		return p;
+	kobject_put(&p->kobject);
+out:
+	return NULL;
+}
+
+void rpc_sysfs_client_setup(struct rpc_clnt *clnt,
+			    struct rpc_xprt_switch *xprt_switch,
+			    struct net *net)
+{
+	struct rpc_sysfs_client *rpc_client;
+
+	rpc_client = rpc_sysfs_client_alloc(rpc_sunrpc_client_kobj,
+					    net, clnt->cl_clid);
+	if (rpc_client) {
+		char name[] = "switch";
+		struct rpc_sysfs_xprt_switch *xswitch =
+			(struct rpc_sysfs_xprt_switch *)xprt_switch->xps_sysfs;
+		int ret;
+
+		clnt->cl_sysfs = rpc_client;
+		rpc_client->clnt = clnt;
+		rpc_client->xprt_switch = xprt_switch;
+		kobject_uevent(&rpc_client->kobject, KOBJ_ADD);
+		ret = sysfs_create_link_nowarn(&rpc_client->kobject,
+					       &xswitch->kobject, name);
+		if (ret)
+			pr_warn("can't create link to %s in sysfs (%d)\n",
+				name, ret);
+	}
+}
+
+void rpc_sysfs_xprt_switch_setup(struct rpc_xprt_switch *xprt_switch,
+				 struct rpc_xprt *xprt,
+				 gfp_t gfp_flags)
+{
+	struct rpc_sysfs_xprt_switch *rpc_xprt_switch;
+	struct net *net;
+
+	if (xprt_switch->xps_net)
+		net = xprt_switch->xps_net;
+	else
+		net = xprt->xprt_net;
+	rpc_xprt_switch =
+		rpc_sysfs_xprt_switch_alloc(rpc_sunrpc_xprt_switch_kobj,
+					    xprt_switch, net, gfp_flags);
+	if (rpc_xprt_switch) {
+		xprt_switch->xps_sysfs = rpc_xprt_switch;
+		rpc_xprt_switch->xprt_switch = xprt_switch;
+		rpc_xprt_switch->xprt = xprt;
+		kobject_uevent(&rpc_xprt_switch->kobject, KOBJ_ADD);
+	}
+}
+
+void rpc_sysfs_xprt_setup(struct rpc_xprt_switch *xprt_switch,
+			  struct rpc_xprt *xprt,
+			  gfp_t gfp_flags)
+{
+	struct rpc_sysfs_xprt *rpc_xprt;
+	struct rpc_sysfs_xprt_switch *switch_obj =
+		(struct rpc_sysfs_xprt_switch *)xprt_switch->xps_sysfs;
+
+	rpc_xprt = rpc_sysfs_xprt_alloc(&switch_obj->kobject, xprt, gfp_flags);
+	if (rpc_xprt) {
+		xprt->xprt_sysfs = rpc_xprt;
+		rpc_xprt->xprt = xprt;
+		rpc_xprt->xprt_switch = xprt_switch;
+		kobject_uevent(&rpc_xprt->kobject, KOBJ_ADD);
+	}
+}
+
+void rpc_sysfs_client_destroy(struct rpc_clnt *clnt)
+{
+	struct rpc_sysfs_client *rpc_client = clnt->cl_sysfs;
+
+	if (rpc_client) {
+		char name[] = "switch";
+
+		sysfs_remove_link(&rpc_client->kobject, name);
+		kobject_uevent(&rpc_client->kobject, KOBJ_REMOVE);
+		kobject_del(&rpc_client->kobject);
+		kobject_put(&rpc_client->kobject);
+		clnt->cl_sysfs = NULL;
+	}
+}
+
+void rpc_sysfs_xprt_switch_destroy(struct rpc_xprt_switch *xprt_switch)
+{
+	struct rpc_sysfs_xprt_switch *rpc_xprt_switch = xprt_switch->xps_sysfs;
+
+	if (rpc_xprt_switch) {
+		kobject_uevent(&rpc_xprt_switch->kobject, KOBJ_REMOVE);
+		kobject_del(&rpc_xprt_switch->kobject);
+		kobject_put(&rpc_xprt_switch->kobject);
+		xprt_switch->xps_sysfs = NULL;
+	}
+}
+
+void rpc_sysfs_xprt_destroy(struct rpc_xprt *xprt)
+{
+	struct rpc_sysfs_xprt *rpc_xprt = xprt->xprt_sysfs;
+
+	if (rpc_xprt) {
+		kobject_uevent(&rpc_xprt->kobject, KOBJ_REMOVE);
+		kobject_del(&rpc_xprt->kobject);
+		kobject_put(&rpc_xprt->kobject);
+		xprt->xprt_sysfs = NULL;
+	}
+}
diff --git a/net/sunrpc/sysfs.h b/net/sunrpc/sysfs.h
new file mode 100644
index 000000000000..6620cebd1037
--- /dev/null
+++ b/net/sunrpc/sysfs.h
@@ -0,0 +1,42 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2020 Anna Schumaker <Anna.Schumaker@Netapp.com>
+ */
+#ifndef __SUNRPC_SYSFS_H
+#define __SUNRPC_SYSFS_H
+
+struct rpc_sysfs_client {
+	struct kobject kobject;
+	struct net *net;
+	struct rpc_clnt *clnt;
+	struct rpc_xprt_switch *xprt_switch;
+};
+
+struct rpc_sysfs_xprt_switch {
+	struct kobject kobject;
+	struct net *net;
+	struct rpc_xprt_switch *xprt_switch;
+	struct rpc_xprt *xprt;
+};
+
+struct rpc_sysfs_xprt {
+	struct kobject kobject;
+	struct rpc_xprt *xprt;
+	struct rpc_xprt_switch *xprt_switch;
+};
+
+int rpc_sysfs_init(void);
+void rpc_sysfs_exit(void);
+
+void rpc_sysfs_client_setup(struct rpc_clnt *clnt,
+			    struct rpc_xprt_switch *xprt_switch,
+			    struct net *net);
+void rpc_sysfs_client_destroy(struct rpc_clnt *clnt);
+void rpc_sysfs_xprt_switch_setup(struct rpc_xprt_switch *xprt_switch,
+				 struct rpc_xprt *xprt, gfp_t gfp_flags);
+void rpc_sysfs_xprt_switch_destroy(struct rpc_xprt_switch *xprt);
+void rpc_sysfs_xprt_setup(struct rpc_xprt_switch *xprt_switch,
+			  struct rpc_xprt *xprt, gfp_t gfp_flags);
+void rpc_sysfs_xprt_destroy(struct rpc_xprt *xprt);
+
+#endif
diff --git a/net/sunrpc/xdr.c b/net/sunrpc/xdr.c
index 3964ff74ee51..ca10ba2626f2 100644
--- a/net/sunrpc/xdr.c
+++ b/net/sunrpc/xdr.c
@@ -1230,10 +1230,9 @@ static unsigned int xdr_set_page_base(struct xdr_stream *xdr,
 	void *kaddr;
 
 	maxlen = xdr->buf->page_len;
-	if (base >= maxlen) {
-		base = maxlen;
-		maxlen = 0;
-	} else
+	if (base >= maxlen)
+		return 0;
+	else
 		maxlen -= base;
 	if (len > maxlen)
 		len = maxlen;
diff --git a/net/sunrpc/xprt.c b/net/sunrpc/xprt.c
index 3509a7f139b9..fb6db09725c7 100644
--- a/net/sunrpc/xprt.c
+++ b/net/sunrpc/xprt.c
@@ -55,6 +55,7 @@
 #include <trace/events/sunrpc.h>
 
 #include "sunrpc.h"
+#include "sysfs.h"
 
 /*
  * Local variables
@@ -443,7 +444,7 @@ void xprt_release_xprt_cong(struct rpc_xprt *xprt, struct rpc_task *task)
 }
 EXPORT_SYMBOL_GPL(xprt_release_xprt_cong);
 
-static inline void xprt_release_write(struct rpc_xprt *xprt, struct rpc_task *task)
+void xprt_release_write(struct rpc_xprt *xprt, struct rpc_task *task)
 {
 	if (xprt->snd_task != task)
 		return;
@@ -1746,6 +1747,30 @@ static void xprt_free_all_slots(struct rpc_xprt *xprt)
 	}
 }
 
+static DEFINE_IDA(rpc_xprt_ids);
+
+void xprt_cleanup_ids(void)
+{
+	ida_destroy(&rpc_xprt_ids);
+}
+
+static int xprt_alloc_id(struct rpc_xprt *xprt)
+{
+	int id;
+
+	id = ida_simple_get(&rpc_xprt_ids, 0, 0, GFP_KERNEL);
+	if (id < 0)
+		return id;
+
+	xprt->id = id;
+	return 0;
+}
+
+static void xprt_free_id(struct rpc_xprt *xprt)
+{
+	ida_simple_remove(&rpc_xprt_ids, xprt->id);
+}
+
 struct rpc_xprt *xprt_alloc(struct net *net, size_t size,
 		unsigned int num_prealloc,
 		unsigned int max_alloc)
@@ -1758,6 +1783,7 @@ struct rpc_xprt *xprt_alloc(struct net *net, size_t size,
 	if (xprt == NULL)
 		goto out;
 
+	xprt_alloc_id(xprt);
 	xprt_init(xprt, net);
 
 	for (i = 0; i < num_prealloc; i++) {
@@ -1786,6 +1812,8 @@ void xprt_free(struct rpc_xprt *xprt)
 {
 	put_net(xprt->xprt_net);
 	xprt_free_all_slots(xprt);
+	xprt_free_id(xprt);
+	rpc_sysfs_xprt_destroy(xprt);
 	kfree_rcu(xprt, rcu);
 }
 EXPORT_SYMBOL_GPL(xprt_free);
diff --git a/net/sunrpc/xprtmultipath.c b/net/sunrpc/xprtmultipath.c
index 1b4073131c6f..c60820e45082 100644
--- a/net/sunrpc/xprtmultipath.c
+++ b/net/sunrpc/xprtmultipath.c
@@ -19,6 +19,8 @@
 #include <linux/sunrpc/addr.h>
 #include <linux/sunrpc/xprtmultipath.h>
 
+#include "sysfs.h"
+
 typedef struct rpc_xprt *(*xprt_switch_find_xprt_t)(struct rpc_xprt_switch *xps,
 		const struct rpc_xprt *cur);
 
@@ -55,6 +57,7 @@ void rpc_xprt_switch_add_xprt(struct rpc_xprt_switch *xps,
 	if (xps->xps_net == xprt->xprt_net || xps->xps_net == NULL)
 		xprt_switch_add_xprt_locked(xps, xprt);
 	spin_unlock(&xps->xps_lock);
+	rpc_sysfs_xprt_setup(xps, xprt, GFP_KERNEL);
 }
 
 static void xprt_switch_remove_xprt_locked(struct rpc_xprt_switch *xps,
@@ -62,7 +65,8 @@ static void xprt_switch_remove_xprt_locked(struct rpc_xprt_switch *xps,
 {
 	if (unlikely(xprt == NULL))
 		return;
-	xps->xps_nactive--;
+	if (!test_bit(XPRT_OFFLINE, &xprt->state))
+		xps->xps_nactive--;
 	xps->xps_nxprts--;
 	if (xps->xps_nxprts == 0)
 		xps->xps_net = NULL;
@@ -86,6 +90,30 @@ void rpc_xprt_switch_remove_xprt(struct rpc_xprt_switch *xps,
 	xprt_put(xprt);
 }
 
+static DEFINE_IDA(rpc_xprtswitch_ids);
+
+void xprt_multipath_cleanup_ids(void)
+{
+	ida_destroy(&rpc_xprtswitch_ids);
+}
+
+static int xprt_switch_alloc_id(struct rpc_xprt_switch *xps, gfp_t gfp_flags)
+{
+	int id;
+
+	id = ida_simple_get(&rpc_xprtswitch_ids, 0, 0, gfp_flags);
+	if (id < 0)
+		return id;
+
+	xps->xps_id = id;
+	return 0;
+}
+
+static void xprt_switch_free_id(struct rpc_xprt_switch *xps)
+{
+	ida_simple_remove(&rpc_xprtswitch_ids, xps->xps_id);
+}
+
 /**
  * xprt_switch_alloc - Allocate a new struct rpc_xprt_switch
  * @xprt: pointer to struct rpc_xprt
@@ -103,12 +131,15 @@ struct rpc_xprt_switch *xprt_switch_alloc(struct rpc_xprt *xprt,
 	if (xps != NULL) {
 		spin_lock_init(&xps->xps_lock);
 		kref_init(&xps->xps_kref);
+		xprt_switch_alloc_id(xps, gfp_flags);
 		xps->xps_nxprts = xps->xps_nactive = 0;
 		atomic_long_set(&xps->xps_queuelen, 0);
 		xps->xps_net = NULL;
 		INIT_LIST_HEAD(&xps->xps_xprt_list);
 		xps->xps_iter_ops = &rpc_xprt_iter_singular;
+		rpc_sysfs_xprt_switch_setup(xps, xprt, gfp_flags);
 		xprt_switch_add_xprt_locked(xps, xprt);
+		rpc_sysfs_xprt_setup(xps, xprt, gfp_flags);
 	}
 
 	return xps;
@@ -136,6 +167,8 @@ static void xprt_switch_free(struct kref *kref)
 			struct rpc_xprt_switch, xps_kref);
 
 	xprt_switch_free_entries(xps);
+	rpc_sysfs_xprt_switch_destroy(xps);
+	xprt_switch_free_id(xps);
 	kfree_rcu(xps, xps_rcu);
 }
 
@@ -198,7 +231,8 @@ void xprt_iter_default_rewind(struct rpc_xprt_iter *xpi)
 static
 bool xprt_is_active(const struct rpc_xprt *xprt)
 {
-	return kref_read(&xprt->kref) != 0;
+	return (kref_read(&xprt->kref) != 0 &&
+		!test_bit(XPRT_OFFLINE, &xprt->state));
 }
 
 static
diff --git a/net/sunrpc/xprtrdma/transport.c b/net/sunrpc/xprtrdma/transport.c
index 19a49d26b1e4..9c2ffc67c0fd 100644
--- a/net/sunrpc/xprtrdma/transport.c
+++ b/net/sunrpc/xprtrdma/transport.c
@@ -73,6 +73,7 @@ unsigned int xprt_rdma_max_inline_read = RPCRDMA_DEF_INLINE;
 unsigned int xprt_rdma_max_inline_write = RPCRDMA_DEF_INLINE;
 unsigned int xprt_rdma_memreg_strategy		= RPCRDMA_FRWR;
 int xprt_rdma_pad_optimize;
+static struct xprt_class xprt_rdma;
 
 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
 
@@ -349,6 +350,7 @@ xprt_setup_rdma(struct xprt_create *args)
 	/* Ensure xprt->addr holds valid server TCP (not RDMA)
 	 * address, for any side protocols which peek at it */
 	xprt->prot = IPPROTO_TCP;
+	xprt->xprt_class = &xprt_rdma;
 	xprt->addrlen = args->addrlen;
 	memcpy(&xprt->addr, sap, xprt->addrlen);
 
diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c
index 316d04945587..e573dcecdd66 100644
--- a/net/sunrpc/xprtsock.c
+++ b/net/sunrpc/xprtsock.c
@@ -91,6 +91,11 @@ static unsigned int xprt_max_resvport_limit = RPC_MAX_RESVPORT;
 
 static struct ctl_table_header *sunrpc_table_header;
 
+static struct xprt_class xs_local_transport;
+static struct xprt_class xs_udp_transport;
+static struct xprt_class xs_tcp_transport;
+static struct xprt_class xs_bc_tcp_transport;
+
 /*
  * FIXME: changing the UDP slot table size should also resize the UDP
  *        socket buffers for existing UDP transports
@@ -1648,6 +1653,13 @@ static int xs_get_srcport(struct sock_xprt *transport)
 	return port;
 }
 
+unsigned short get_srcport(struct rpc_xprt *xprt)
+{
+	struct sock_xprt *sock = container_of(xprt, struct sock_xprt, xprt);
+	return sock->srcport;
+}
+EXPORT_SYMBOL(get_srcport);
+
 static unsigned short xs_next_srcport(struct sock_xprt *transport, unsigned short port)
 {
 	if (transport->srcport != 0)
@@ -1689,7 +1701,8 @@ static int xs_bind(struct sock_xprt *transport, struct socket *sock)
 		err = kernel_bind(sock, (struct sockaddr *)&myaddr,
 				transport->xprt.addrlen);
 		if (err == 0) {
-			transport->srcport = port;
+			if (transport->xprt.reuseport)
+				transport->srcport = port;
 			break;
 		}
 		last = port;
@@ -2779,6 +2792,7 @@ static struct rpc_xprt *xs_setup_local(struct xprt_create *args)
 	transport = container_of(xprt, struct sock_xprt, xprt);
 
 	xprt->prot = 0;
+	xprt->xprt_class = &xs_local_transport;
 	xprt->max_payload = RPC_MAX_FRAGMENT_SIZE;
 
 	xprt->bind_timeout = XS_BIND_TO;
@@ -2848,6 +2862,7 @@ static struct rpc_xprt *xs_setup_udp(struct xprt_create *args)
 	transport = container_of(xprt, struct sock_xprt, xprt);
 
 	xprt->prot = IPPROTO_UDP;
+	xprt->xprt_class = &xs_udp_transport;
 	/* XXX: header size can vary due to auth type, IPv6, etc. */
 	xprt->max_payload = (1U << 16) - (MAX_HEADER << 3);
 
@@ -2928,6 +2943,7 @@ static struct rpc_xprt *xs_setup_tcp(struct xprt_create *args)
 	transport = container_of(xprt, struct sock_xprt, xprt);
 
 	xprt->prot = IPPROTO_TCP;
+	xprt->xprt_class = &xs_tcp_transport;
 	xprt->max_payload = RPC_MAX_FRAGMENT_SIZE;
 
 	xprt->bind_timeout = XS_BIND_TO;
@@ -3001,6 +3017,7 @@ static struct rpc_xprt *xs_setup_bc_tcp(struct xprt_create *args)
 	transport = container_of(xprt, struct sock_xprt, xprt);
 
 	xprt->prot = IPPROTO_TCP;
+	xprt->xprt_class = &xs_bc_tcp_transport;
 	xprt->max_payload = RPC_MAX_FRAGMENT_SIZE;
 	xprt->timeout = &xs_tcp_default_timeout;
author	Linus Torvalds	2021-07-09 09:43:57 -0700
committer	Linus Torvalds	2021-07-09 09:43:57 -0700
commit	96890bc2eaa1f6bfc1b194e0f0815a10824352a4 (patch)
tree	c26a6f72adc8aa5ce62cc7549d95e275109e5617
parent	227c4d507c71acb7bece298a98d83e5b44433f62 (diff)
parent	878b3dfc42c4ddbf9e38cd9061e3ddd99a69747a (diff)