Merge tag 'nfs-for-5.9-1' of git://git.linux-nfs.org/projects/trondmy/linux-nfs

Pull NFS client updates from Trond Myklebust:
 "Stable fixes:
   - pNFS: Don't return layout segments that are being used for I/O
   - pNFS: Don't move layout segments off the active list when being used for I/O

  Features:
   - NFS: Add support for user xattrs through the NFSv4.2 protocol
   - NFS: Allow applications to speed up readdir+statx() using AT_STATX_DONT_SYNC
   - NFSv4.0 allow nconnect for v4.0

  Bugfixes and cleanups:
   - nfs: ensure correct writeback errors are returned on close()
   - nfs: nfs_file_write() should check for writeback errors
   - nfs: Fix getxattr kernel panic and memory overflow
   - NFS: Fix the pNFS/flexfiles mirrored read failover code
   - SUNRPC: dont update timeout value on connection reset
   - freezer: Add unsafe versions of freezable_schedule_timeout_interruptible for NFS
   - sunrpc: destroy rpc_inode_cachep after unregister_filesystem"

* tag 'nfs-for-5.9-1' of git://git.linux-nfs.org/projects/trondmy/linux-nfs: (32 commits)
  NFS: Fix flexfiles read failover
  fs: nfs: delete repeated words in comments
  rpc_pipefs: convert comma to semicolon
  nfs: Fix getxattr kernel panic and memory overflow
  NFS: Don't return layout segments that are in use
  NFS: Don't move layouts to plh_return_segs list while in use
  NFS: Add layout segment info to pnfs read/write/commit tracepoints
  NFS: Add tracepoints for layouterror and layoutstats.
  NFS: Report the stateid + status in trace_nfs4_layoutreturn_on_close()
  SUNRPC dont update timeout value on connection reset
  nfs: nfs_file_write() should check for writeback errors
  nfs: ensure correct writeback errors are returned on close()
  NFSv4.2: xattr cache: get rid of cache discard work queue
  NFS: remove redundant initialization of variable result
  NFSv4.0 allow nconnect for v4.0
  freezer: Add unsafe versions of freezable_schedule_timeout_interruptible for NFS
  sunrpc: destroy rpc_inode_cachep after unregister_filesystem
  NFSv4.2: add client side xattr caching.
  NFSv4.2: hook in the user extended attribute handlers
  NFSv4.2: add the extended attribute proc functions.
  ...

23 files changed, 2274 insertions, 121 deletions

diff --git a/fs/nfs/Makefile b/fs/nfs/Makefile
index 2433c3e03cfa..22d11fdc6deb 100644
--- a/fs/nfs/Makefile
+++ b/fs/nfs/Makefile
@@ -30,7 +30,7 @@ nfsv4-y := nfs4proc.o nfs4xdr.o nfs4state.o nfs4renewd.o nfs4super.o nfs4file.o
 nfsv4-$(CONFIG_NFS_USE_LEGACY_DNS) += cache_lib.o
 nfsv4-$(CONFIG_SYSCTL)	+= nfs4sysctl.o
 nfsv4-$(CONFIG_NFS_V4_1)	+= pnfs.o pnfs_dev.o pnfs_nfs.o
-nfsv4-$(CONFIG_NFS_V4_2)	+= nfs42proc.o
+nfsv4-$(CONFIG_NFS_V4_2)	+= nfs42proc.o nfs42xattr.o
 
 obj-$(CONFIG_PNFS_FILE_LAYOUT) += filelayout/
 obj-$(CONFIG_PNFS_BLOCK) += blocklayout/
diff --git a/fs/nfs/blocklayout/rpc_pipefs.c b/fs/nfs/blocklayout/rpc_pipefs.c
index 9fb067a6f7e0..ef9db135c649 100644
--- a/fs/nfs/blocklayout/rpc_pipefs.c
+++ b/fs/nfs/blocklayout/rpc_pipefs.c
@@ -79,7 +79,7 @@ bl_resolve_deviceid(struct nfs_server *server, struct pnfs_block_volume *b,
 		goto out_free_data;
 
 	bl_msg = msg->data;
-	bl_msg->type = BL_DEVICE_MOUNT,
+	bl_msg->type = BL_DEVICE_MOUNT;
 	bl_msg->totallen = b->simple.len;
 	nfs4_encode_simple(msg->data + sizeof(*bl_msg), b);
 
diff --git a/fs/nfs/client.c b/fs/nfs/client.c
index f1ff3076e4a4..4b8cc93913f7 100644
--- a/fs/nfs/client.c
+++ b/fs/nfs/client.c
@@ -50,6 +50,7 @@
 #include "nfs.h"
 #include "netns.h"
 #include "sysfs.h"
+#include "nfs42.h"
 
 #define NFSDBG_FACILITY		NFSDBG_CLIENT
 
@@ -749,7 +750,7 @@ error:
 static void nfs_server_set_fsinfo(struct nfs_server *server,
 				  struct nfs_fsinfo *fsinfo)
 {
-	unsigned long max_rpc_payload;
+	unsigned long max_rpc_payload, raw_max_rpc_payload;
 
 	/* Work out a lot of parameters */
 	if (server->rsize == 0)
@@ -762,7 +763,9 @@ static void nfs_server_set_fsinfo(struct nfs_server *server,
 	if (fsinfo->wtmax >= 512 && server->wsize > fsinfo->wtmax)
 		server->wsize = nfs_block_size(fsinfo->wtmax, NULL);
 
-	max_rpc_payload = nfs_block_size(rpc_max_payload(server->client), NULL);
+	raw_max_rpc_payload = rpc_max_payload(server->client);
+	max_rpc_payload = nfs_block_size(raw_max_rpc_payload, NULL);
+
 	if (server->rsize > max_rpc_payload)
 		server->rsize = max_rpc_payload;
 	if (server->rsize > NFS_MAX_FILE_IO_SIZE)
@@ -795,6 +798,21 @@ static void nfs_server_set_fsinfo(struct nfs_server *server,
 	server->clone_blksize = fsinfo->clone_blksize;
 	/* We're airborne Set socket buffersize */
 	rpc_setbufsize(server->client, server->wsize + 100, server->rsize + 100);
+
+#ifdef CONFIG_NFS_V4_2
+	/*
+	 * Defaults until limited by the session parameters.
+	 */
+	server->gxasize = min_t(unsigned int, raw_max_rpc_payload,
+				XATTR_SIZE_MAX);
+	server->sxasize = min_t(unsigned int, raw_max_rpc_payload,
+				XATTR_SIZE_MAX);
+	server->lxasize = min_t(unsigned int, raw_max_rpc_payload,
+				nfs42_listxattr_xdrsize(XATTR_LIST_MAX));
+
+	if (fsinfo->xattr_support)
+		server->caps |= NFS_CAP_XATTR;
+#endif
 }
 
 /*
diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c
index 5a331da5f55a..a12f42e7d8c7 100644
--- a/fs/nfs/dir.c
+++ b/fs/nfs/dir.c
@@ -2460,7 +2460,7 @@ static struct nfs_access_entry *nfs_access_search_rbtree(struct inode *inode, co
 	return NULL;
 }
 
-static int nfs_access_get_cached(struct inode *inode, const struct cred *cred, struct nfs_access_entry *res, bool may_block)
+static int nfs_access_get_cached_locked(struct inode *inode, const struct cred *cred, struct nfs_access_entry *res, bool may_block)
 {
 	struct nfs_inode *nfsi = NFS_I(inode);
 	struct nfs_access_entry *cache;
@@ -2533,6 +2533,20 @@ out:
 	return err;
 }
 
+int nfs_access_get_cached(struct inode *inode, const struct cred *cred, struct
+nfs_access_entry *res, bool may_block)
+{
+	int status;
+
+	status = nfs_access_get_cached_rcu(inode, cred, res);
+	if (status != 0)
+		status = nfs_access_get_cached_locked(inode, cred, res,
+		    may_block);
+
+	return status;
+}
+EXPORT_SYMBOL_GPL(nfs_access_get_cached);
+
 static void nfs_access_add_rbtree(struct inode *inode, struct nfs_access_entry *set)
 {
 	struct nfs_inode *nfsi = NFS_I(inode);
@@ -2647,9 +2661,7 @@ static int nfs_do_access(struct inode *inode, const struct cred *cred, int mask)
 
 	trace_nfs_access_enter(inode);
 
-	status = nfs_access_get_cached_rcu(inode, cred, &cache);
-	if (status != 0)
-		status = nfs_access_get_cached(inode, cred, &cache, may_block);
+	status = nfs_access_get_cached(inode, cred, &cache, may_block);
 	if (status == 0)
 		goto out_cached;
 
@@ -2661,6 +2673,10 @@ static int nfs_do_access(struct inode *inode, const struct cred *cred, int mask)
 	 * Determine which access bits we want to ask for...
 	 */
 	cache.mask = NFS_ACCESS_READ | NFS_ACCESS_MODIFY | NFS_ACCESS_EXTEND;
+	if (nfs_server_capable(inode, NFS_CAP_XATTR)) {
+		cache.mask |= NFS_ACCESS_XAREAD | NFS_ACCESS_XAWRITE |
+		    NFS_ACCESS_XALIST;
+	}
 	if (S_ISDIR(inode->i_mode))
 		cache.mask |= NFS_ACCESS_DELETE | NFS_ACCESS_LOOKUP;
 	else
diff --git a/fs/nfs/direct.c b/fs/nfs/direct.c
index 1b79dd5cf661..2d30a4da49fa 100644
--- a/fs/nfs/direct.c
+++ b/fs/nfs/direct.c
@@ -896,7 +896,7 @@ static ssize_t nfs_direct_write_schedule_iovec(struct nfs_direct_req *dreq,
  */
 ssize_t nfs_file_direct_write(struct kiocb *iocb, struct iov_iter *iter)
 {
-	ssize_t result = -EINVAL, requested;
+	ssize_t result, requested;
 	size_t count;
 	struct file *file = iocb->ki_filp;
 	struct address_space *mapping = file->f_mapping;
diff --git a/fs/nfs/file.c b/fs/nfs/file.c
index f96367a2463e..63940a7a70be 100644
--- a/fs/nfs/file.c
+++ b/fs/nfs/file.c
@@ -140,6 +140,7 @@ static int
 nfs_file_flush(struct file *file, fl_owner_t id)
 {
 	struct inode	*inode = file_inode(file);
+	errseq_t since;
 
 	dprintk("NFS: flush(%pD2)\n", file);
 
@@ -148,7 +149,9 @@ nfs_file_flush(struct file *file, fl_owner_t id)
 		return 0;
 
 	/* Flush writes to the server and return any errors */
-	return nfs_wb_all(inode);
+	since = filemap_sample_wb_err(file->f_mapping);
+	nfs_wb_all(inode);
+	return filemap_check_wb_err(file->f_mapping, since);
 }
 
 ssize_t
@@ -587,12 +590,14 @@ static const struct vm_operations_struct nfs_file_vm_ops = {
 	.page_mkwrite = nfs_vm_page_mkwrite,
 };
 
-static int nfs_need_check_write(struct file *filp, struct inode *inode)
+static int nfs_need_check_write(struct file *filp, struct inode *inode,
+				int error)
 {
 	struct nfs_open_context *ctx;
 
 	ctx = nfs_file_open_context(filp);
-	if (nfs_ctx_key_to_expire(ctx, inode))
+	if (nfs_error_is_fatal_on_server(error) ||
+	    nfs_ctx_key_to_expire(ctx, inode))
 		return 1;
 	return 0;
 }
@@ -603,6 +608,8 @@ ssize_t nfs_file_write(struct kiocb *iocb, struct iov_iter *from)
 	struct inode *inode = file_inode(file);
 	unsigned long written = 0;
 	ssize_t result;
+	errseq_t since;
+	int error;
 
 	result = nfs_key_timeout_notify(file, inode);
 	if (result)
@@ -627,6 +634,7 @@ ssize_t nfs_file_write(struct kiocb *iocb, struct iov_iter *from)
 	if (iocb->ki_pos > i_size_read(inode))
 		nfs_revalidate_mapping(inode, file->f_mapping);
 
+	since = filemap_sample_wb_err(file->f_mapping);
 	nfs_start_io_write(inode);
 	result = generic_write_checks(iocb, from);
 	if (result > 0) {
@@ -645,7 +653,8 @@ ssize_t nfs_file_write(struct kiocb *iocb, struct iov_iter *from)
 		goto out;
 
 	/* Return error values */
-	if (nfs_need_check_write(file, inode)) {
+	error = filemap_check_wb_err(file->f_mapping, since);
+	if (nfs_need_check_write(file, inode, error)) {
 		int err = nfs_wb_all(inode);
 		if (err < 0)
 			result = err;
diff --git a/fs/nfs/flexfilelayout/flexfilelayout.c b/fs/nfs/flexfilelayout/flexfilelayout.c
index de03e440b7ee..965145592750 100644
--- a/fs/nfs/flexfilelayout/flexfilelayout.c
+++ b/fs/nfs/flexfilelayout/flexfilelayout.c
@@ -790,6 +790,19 @@ ff_layout_choose_best_ds_for_read(struct pnfs_layout_segment *lseg,
 	return ff_layout_choose_any_ds_for_read(lseg, start_idx, best_idx);
 }
 
+static struct nfs4_pnfs_ds *
+ff_layout_get_ds_for_read(struct nfs_pageio_descriptor *pgio, int *best_idx)
+{
+	struct pnfs_layout_segment *lseg = pgio->pg_lseg;
+	struct nfs4_pnfs_ds *ds;
+
+	ds = ff_layout_choose_best_ds_for_read(lseg, pgio->pg_mirror_idx,
+					       best_idx);
+	if (ds || !pgio->pg_mirror_idx)
+		return ds;
+	return ff_layout_choose_best_ds_for_read(lseg, 0, best_idx);
+}
+
 static void
 ff_layout_pg_get_read(struct nfs_pageio_descriptor *pgio,
 		      struct nfs_page *req,
@@ -840,12 +853,11 @@ retry:
 			goto out_nolseg;
 	}
 
-	ds = ff_layout_choose_best_ds_for_read(pgio->pg_lseg, 0, &ds_idx);
+	ds = ff_layout_get_ds_for_read(pgio, &ds_idx);
 	if (!ds) {
 		if (!ff_layout_no_fallback_to_mds(pgio->pg_lseg))
 			goto out_mds;
-		pnfs_put_lseg(pgio->pg_lseg);
-		pgio->pg_lseg = NULL;
+		pnfs_generic_pg_cleanup(pgio);
 		/* Sleep for 1 second before retrying */
 		ssleep(1);
 		goto retry;
@@ -871,8 +883,6 @@ out_mds:
 			0, NFS4_MAX_UINT64, IOMODE_READ,
 			NFS_I(pgio->pg_inode)->layout,
 			pgio->pg_lseg);
-	pnfs_put_lseg(pgio->pg_lseg);
-	pgio->pg_lseg = NULL;
 	pgio->pg_maxretrans = 0;
 	nfs_pageio_reset_read_mds(pgio);
 }
@@ -916,8 +926,7 @@ retry:
 		if (!ds) {
 			if (!ff_layout_no_fallback_to_mds(pgio->pg_lseg))
 				goto out_mds;
-			pnfs_put_lseg(pgio->pg_lseg);
-			pgio->pg_lseg = NULL;
+			pnfs_generic_pg_cleanup(pgio);
 			/* Sleep for 1 second before retrying */
 			ssleep(1);
 			goto retry;
@@ -939,8 +948,6 @@ out_mds:
 			0, NFS4_MAX_UINT64, IOMODE_RW,
 			NFS_I(pgio->pg_inode)->layout,
 			pgio->pg_lseg);
-	pnfs_put_lseg(pgio->pg_lseg);
-	pgio->pg_lseg = NULL;
 	pgio->pg_maxretrans = 0;
 	nfs_pageio_reset_write_mds(pgio);
 	pgio->pg_error = -EAGAIN;
@@ -953,8 +960,8 @@ ff_layout_pg_get_mirror_count_write(struct nfs_pageio_descriptor *pgio,
 	if (!pgio->pg_lseg) {
 		pgio->pg_lseg = pnfs_update_layout(pgio->pg_inode,
 						   nfs_req_openctx(req),
-						   0,
-						   NFS4_MAX_UINT64,
+						   req_offset(req),
+						   req->wb_bytes,
 						   IOMODE_RW,
 						   false,
 						   GFP_NOFS);
@@ -1028,11 +1035,24 @@ static void ff_layout_reset_write(struct nfs_pgio_header *hdr, bool retry_pnfs)
 	}
 }
 
+static void ff_layout_resend_pnfs_read(struct nfs_pgio_header *hdr)
+{
+	u32 idx = hdr->pgio_mirror_idx + 1;
+	int new_idx = 0;
+
+	if (ff_layout_choose_any_ds_for_read(hdr->lseg, idx + 1, &new_idx))
+		ff_layout_send_layouterror(hdr->lseg);
+	else
+		pnfs_error_mark_layout_for_return(hdr->inode, hdr->lseg);
+	pnfs_read_resend_pnfs(hdr, new_idx);
+}
+
 static void ff_layout_reset_read(struct nfs_pgio_header *hdr)
 {
 	struct rpc_task *task = &hdr->task;
 
 	pnfs_layoutcommit_inode(hdr->inode, false);
+	pnfs_error_mark_layout_for_return(hdr->inode, hdr->lseg);
 
 	if (!test_and_set_bit(NFS_IOHDR_REDO, &hdr->flags)) {
 		dprintk("%s Reset task %5u for i/o through MDS "
@@ -1234,6 +1254,12 @@ static void ff_layout_io_track_ds_error(struct pnfs_layout_segment *lseg,
 		break;
 	case NFS4ERR_NXIO:
 		ff_layout_mark_ds_unreachable(lseg, idx);
+		/*
+		 * Don't return the layout if this is a read and we still
+		 * have layouts to try
+		 */
+		if (opnum == OP_READ)
+			break;
 		/* Fallthrough */
 	default:
 		pnfs_error_mark_layout_for_return(lseg->pls_layout->plh_inode,
@@ -1247,7 +1273,6 @@ static void ff_layout_io_track_ds_error(struct pnfs_layout_segment *lseg,
 static int ff_layout_read_done_cb(struct rpc_task *task,
 				struct nfs_pgio_header *hdr)
 {
-	int new_idx = hdr->pgio_mirror_idx;
 	int err;
 
 	if (task->tk_status < 0) {
@@ -1267,10 +1292,6 @@ static int ff_layout_read_done_cb(struct rpc_task *task,
 	clear_bit(NFS_IOHDR_RESEND_MDS, &hdr->flags);
 	switch (err) {
 	case -NFS4ERR_RESET_TO_PNFS:
-		if (ff_layout_choose_best_ds_for_read(hdr->lseg,
-					hdr->pgio_mirror_idx + 1,
-					&new_idx))
-			goto out_layouterror;
 		set_bit(NFS_IOHDR_RESEND_PNFS, &hdr->flags);
 		return task->tk_status;
 	case -NFS4ERR_RESET_TO_MDS:
@@ -1281,10 +1302,6 @@ static int ff_layout_read_done_cb(struct rpc_task *task,
 	}
 
 	return 0;
-out_layouterror:
-	ff_layout_read_record_layoutstats_done(task, hdr);
-	ff_layout_send_layouterror(hdr->lseg);
-	hdr->pgio_mirror_idx = new_idx;
 out_eagain:
 	rpc_restart_call_prepare(task);
 	return -EAGAIN;
@@ -1411,10 +1428,9 @@ static void ff_layout_read_release(void *data)
 	struct nfs_pgio_header *hdr = data;
 
 	ff_layout_read_record_layoutstats_done(&hdr->task, hdr);
-	if (test_bit(NFS_IOHDR_RESEND_PNFS, &hdr->flags)) {
-		ff_layout_send_layouterror(hdr->lseg);
-		pnfs_read_resend_pnfs(hdr);
-	} else if (test_bit(NFS_IOHDR_RESEND_MDS, &hdr->flags))
+	if (test_bit(NFS_IOHDR_RESEND_PNFS, &hdr->flags))
+		ff_layout_resend_pnfs_read(hdr);
+	else if (test_bit(NFS_IOHDR_RESEND_MDS, &hdr->flags))
 		ff_layout_reset_read(hdr);
 	pnfs_generic_rw_release(data);
 }
diff --git a/fs/nfs/fs_context.c b/fs/nfs/fs_context.c
index ccc88be88d6a..66949da0e827 100644
--- a/fs/nfs/fs_context.c
+++ b/fs/nfs/fs_context.c
@@ -982,7 +982,7 @@ static int nfs23_parse_monolithic(struct fs_context *fc,
 		/*
 		 * The legacy version 6 binary mount data from userspace has a
 		 * field used only to transport selinux information into the
-		 * the kernel.  To continue to support that functionality we
+		 * kernel.  To continue to support that functionality we
 		 * have a touch of selinux knowledge here in the NFS code. The
 		 * userspace code converted context=blah to just blah so we are
 		 * converting back to the full string selinux understands.
diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index 0bf1f835de01..aa6493905bbe 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -193,6 +193,7 @@ bool nfs_check_cache_invalid(struct inode *inode, unsigned long flags)
 
 	return nfs_check_cache_invalid_not_delegated(inode, flags);
 }
+EXPORT_SYMBOL_GPL(nfs_check_cache_invalid);
 
 static void nfs_set_cache_invalid(struct inode *inode, unsigned long flags)
 {
@@ -204,7 +205,8 @@ static void nfs_set_cache_invalid(struct inode *inode, unsigned long flags)
 			flags &= ~NFS_INO_INVALID_OTHER;
 		flags &= ~(NFS_INO_INVALID_CHANGE
 				| NFS_INO_INVALID_SIZE
-				| NFS_INO_REVAL_PAGECACHE);
+				| NFS_INO_REVAL_PAGECACHE
+				| NFS_INO_INVALID_XATTR);
 	}
 
 	if (inode->i_mapping->nrpages == 0)
@@ -233,11 +235,13 @@ static void nfs_zap_caches_locked(struct inode *inode)
 					| NFS_INO_INVALID_DATA
 					| NFS_INO_INVALID_ACCESS
 					| NFS_INO_INVALID_ACL
+					| NFS_INO_INVALID_XATTR
 					| NFS_INO_REVAL_PAGECACHE);
 	} else
 		nfs_set_cache_invalid(inode, NFS_INO_INVALID_ATTR
 					| NFS_INO_INVALID_ACCESS
 					| NFS_INO_INVALID_ACL
+					| NFS_INO_INVALID_XATTR
 					| NFS_INO_REVAL_PAGECACHE);
 	nfs_zap_label_cache_locked(nfsi);
 }
@@ -542,6 +546,8 @@ nfs_fhget(struct super_block *sb, struct nfs_fh *fh, struct nfs_fattr *fattr, st
 			inode->i_gid = fattr->gid;
 		else if (nfs_server_capable(inode, NFS_CAP_OWNER_GROUP))
 			nfs_set_cache_invalid(inode, NFS_INO_INVALID_OTHER);
+		if (nfs_server_capable(inode, NFS_CAP_XATTR))
+			nfs_set_cache_invalid(inode, NFS_INO_INVALID_XATTR);
 		if (fattr->valid & NFS_ATTR_FATTR_BLOCKS_USED)
 			inode->i_blocks = fattr->du.nfs2.blocks;
 		if (fattr->valid & NFS_ATTR_FATTR_SPACE_USED) {
@@ -794,8 +800,10 @@ int nfs_getattr(const struct path *path, struct kstat *stat,
 
 	trace_nfs_getattr_enter(inode);
 
-	if ((query_flags & AT_STATX_DONT_SYNC) && !force_sync)
+	if ((query_flags & AT_STATX_DONT_SYNC) && !force_sync) {
+		nfs_readdirplus_parent_cache_hit(path->dentry);
 		goto out_no_update;
+	}
 
 	/* Flush out writes to the server in order to update c/mtime.  */
 	if ((request_mask & (STATX_CTIME|STATX_MTIME)) &&
@@ -1375,6 +1383,8 @@ static void nfs_wcc_update_inode(struct inode *inode, struct nfs_fattr *fattr)
 		inode_set_iversion_raw(inode, fattr->change_attr);
 		if (S_ISDIR(inode->i_mode))
 			nfs_set_cache_invalid(inode, NFS_INO_INVALID_DATA);
+		else if (nfs_server_capable(inode, NFS_CAP_XATTR))
+			nfs_set_cache_invalid(inode, NFS_INO_INVALID_XATTR);
 	}
 	/* If we have atomic WCC data, we may update some attributes */
 	ts = inode->i_ctime;
@@ -1892,7 +1902,8 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
 			if (!(have_writers || have_delegation)) {
 				invalid |= NFS_INO_INVALID_DATA
 					| NFS_INO_INVALID_ACCESS
-					| NFS_INO_INVALID_ACL;
+					| NFS_INO_INVALID_ACL
+					| NFS_INO_INVALID_XATTR;
 				/* Force revalidate of all attributes */
 				save_cache_validity |= NFS_INO_INVALID_CTIME
 					| NFS_INO_INVALID_MTIME
@@ -2095,6 +2106,9 @@ struct inode *nfs_alloc_inode(struct super_block *sb)
 #if IS_ENABLED(CONFIG_NFS_V4)
 	nfsi->nfs4_acl = NULL;
 #endif /* CONFIG_NFS_V4 */
+#ifdef CONFIG_NFS_V4_2
+	nfsi->xattr_cache = NULL;
+#endif
 	return &nfsi->vfs_inode;
 }
 EXPORT_SYMBOL_GPL(nfs_alloc_inode);
diff --git a/fs/nfs/nfs42.h b/fs/nfs/nfs42.h
index c891af949886..0fe5aacbcfdf 100644
--- a/fs/nfs/nfs42.h
+++ b/fs/nfs/nfs42.h
@@ -6,6 +6,8 @@
 #ifndef __LINUX_FS_NFS_NFS4_2_H
 #define __LINUX_FS_NFS_NFS4_2_H
 
+#include <linux/xattr.h>
+
 /*
  * FIXME:  four LAYOUTSTATS calls per compound at most! Do we need to support
  * more? Need to consider not to pre-alloc too much for a compound.
@@ -36,5 +38,27 @@ static inline bool nfs42_files_from_same_server(struct file *in,
 	return nfs4_check_serverowner_major_id(c_in->cl_serverowner,
 					       c_out->cl_serverowner);
 }
+
+ssize_t nfs42_proc_getxattr(struct inode *inode, const char *name,
+			    void *buf, size_t buflen);
+int nfs42_proc_setxattr(struct inode *inode, const char *name,
+			const void *buf, size_t buflen, int flags);
+ssize_t nfs42_proc_listxattrs(struct inode *inode, void *buf,
+			       size_t buflen, u64 *cookiep, bool *eofp);
+int nfs42_proc_removexattr(struct inode *inode, const char *name);
+
+/*
+ * Maximum XDR buffer size needed for a listxattr buffer of buflen size.
+ *
+ * The upper boundary is a buffer with all 1-byte sized attribute names.
+ * They would be 7 bytes long in the eventual buffer ("user.x\0"), and
+ * 8 bytes long XDR-encoded.
+ *
+ * Include the trailing eof word as well.
+ */
+static inline u32 nfs42_listxattr_xdrsize(u32 buflen)
+{
+	return ((buflen / (XATTR_USER_PREFIX_LEN + 2)) * 8) + 4;
+}
 #endif /* CONFIG_NFS_V4_2 */
 #endif /* __LINUX_FS_NFS_NFS4_2_H */
diff --git a/fs/nfs/nfs42proc.c b/fs/nfs/nfs42proc.c
index e2ae54b35dfe..142225f0af59 100644
--- a/fs/nfs/nfs42proc.c
+++ b/fs/nfs/nfs42proc.c
@@ -17,6 +17,7 @@
 #include "nfs4session.h"
 #include "internal.h"
 #include "delegation.h"
+#include "nfs4trace.h"
 
 #define NFSDBG_FACILITY NFSDBG_PROC
 static int nfs42_do_offload_cancel_async(struct file *dst, nfs4_stateid *std);
@@ -714,7 +715,7 @@ nfs42_layoutstat_done(struct rpc_task *task, void *calldata)
 
 	switch (task->tk_status) {
 	case 0:
-		break;
+		return;
 	case -NFS4ERR_BADHANDLE:
 	case -ESTALE:
 		pnfs_destroy_layout(NFS_I(inode));
@@ -760,6 +761,8 @@ nfs42_layoutstat_done(struct rpc_task *task, void *calldata)
 	case -EOPNOTSUPP:
 		NFS_SERVER(inode)->caps &= ~NFS_CAP_LAYOUTSTATS;
 	}
+
+	trace_nfs4_layoutstats(inode, &data->args.stateid, task->tk_status);
 }
 
 static void
@@ -882,7 +885,7 @@ nfs42_layouterror_done(struct rpc_task *task, void *calldata)
 
 	switch (task->tk_status) {
 	case 0:
-		break;
+		return;
 	case -NFS4ERR_BADHANDLE:
 	case -ESTALE:
 		pnfs_destroy_layout(NFS_I(inode));
@@ -926,6 +929,9 @@ nfs42_layouterror_done(struct rpc_task *task, void *calldata)
 	case -EOPNOTSUPP:
 		NFS_SERVER(inode)->caps &= ~NFS_CAP_LAYOUTERROR;
 	}
+
+	trace_nfs4_layouterror(inode, &data->args.errors[0].stateid,
+			       task->tk_status);
 }
 
 static void
@@ -1088,3 +1094,251 @@ out_put_src_lock:
 	nfs_put_lock_context(src_lock);
 	return err;
 }
+
+#define NFS4XATTR_MAXPAGES DIV_ROUND_UP(XATTR_SIZE_MAX, PAGE_SIZE)
+
+static int _nfs42_proc_removexattr(struct inode *inode, const char *name)
+{
+	struct nfs_server *server = NFS_SERVER(inode);
+	struct nfs42_removexattrargs args = {
+		.fh = NFS_FH(inode),
+		.xattr_name = name,
+	};
+	struct nfs42_removexattrres res;
+	struct rpc_message msg = {
+		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_REMOVEXATTR],
+		.rpc_argp = &args,
+		.rpc_resp = &res,
+	};
+	int ret;
+	unsigned long timestamp = jiffies;
+
+	ret = nfs4_call_sync(server->client, server, &msg, &args.seq_args,
+	    &res.seq_res, 1);
+	if (!ret)
+		nfs4_update_changeattr(inode, &res.cinfo, timestamp, 0);
+
+	return ret;
+}
+
+static int _nfs42_proc_setxattr(struct inode *inode, const char *name,
+				const void *buf, size_t buflen, int flags)
+{
+	struct nfs_server *server = NFS_SERVER(inode);
+	struct page *pages[NFS4XATTR_MAXPAGES];
+	struct nfs42_setxattrargs arg = {
+		.fh		= NFS_FH(inode),
+		.xattr_pages	= pages,
+		.xattr_len	= buflen,
+		.xattr_name	= name,
+		.xattr_flags	= flags,
+	};
+	struct nfs42_setxattrres res;
+	struct rpc_message msg = {
+		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_SETXATTR],
+		.rpc_argp	= &arg,
+		.rpc_resp	= &res,
+	};
+	int ret, np;
+	unsigned long timestamp = jiffies;
+
+	if (buflen > server->sxasize)
+		return -ERANGE;
+
+	if (buflen > 0) {
+		np = nfs4_buf_to_pages_noslab(buf, buflen, arg.xattr_pages);
+		if (np < 0)
+			return np;
+	} else
+		np = 0;
+
+	ret = nfs4_call_sync(server->client, server, &msg, &arg.seq_args,
+	    &res.seq_res, 1);
+
+	for (; np > 0; np--)
+		put_page(pages[np - 1]);
+
+	if (!ret)
+		nfs4_update_changeattr(inode, &res.cinfo, timestamp, 0);
+
+	return ret;
+}
+
+static ssize_t _nfs42_proc_getxattr(struct inode *inode, const char *name,
+				void *buf, size_t buflen)
+{
+	struct nfs_server *server = NFS_SERVER(inode);
+	struct page *pages[NFS4XATTR_MAXPAGES] = {};
+	struct nfs42_getxattrargs arg = {
+		.fh		= NFS_FH(inode),
+		.xattr_pages	= pages,
+		.xattr_len	= buflen,
+		.xattr_name	= name,
+	};
+	struct nfs42_getxattrres res;
+	struct rpc_message msg = {
+		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_GETXATTR],
+		.rpc_argp	= &arg,
+		.rpc_resp	= &res,
+	};
+	int ret, np;
+
+	ret = nfs4_call_sync(server->client, server, &msg, &arg.seq_args,
+	    &res.seq_res, 0);
+	if (ret < 0)
+		return ret;
+
+	/*
+	 * Normally, the caching is done one layer up, but for successful
+	 * RPCS, always cache the result here, even if the caller was
+	 * just querying the length, or if the reply was too big for
+	 * the caller. This avoids a second RPC in the case of the
+	 * common query-alloc-retrieve cycle for xattrs.
+	 *
+	 * Note that xattr_len is always capped to XATTR_SIZE_MAX.
+	 */
+
+	nfs4_xattr_cache_add(inode, name, NULL, pages, res.xattr_len);
+
+	if (buflen) {
+		if (res.xattr_len > buflen)
+			return -ERANGE;
+		_copy_from_pages(buf, pages, 0, res.xattr_len);
+	}
+
+	np = DIV_ROUND_UP(res.xattr_len, PAGE_SIZE);
+	while (--np >= 0)
+		__free_page(pages[np]);
+
+	return res.xattr_len;
+}
+
+static ssize_t _nfs42_proc_listxattrs(struct inode *inode, void *buf,
+				 size_t buflen, u64 *cookiep, bool *eofp)
+{
+	struct nfs_server *server = NFS_SERVER(inode);
+	struct page **pages;
+	struct nfs42_listxattrsargs arg = {
+		.fh		= NFS_FH(inode),
+		.cookie		= *cookiep,
+	};
+	struct nfs42_listxattrsres res = {
+		.eof = false,
+		.xattr_buf = buf,
+		.xattr_len = buflen,
+	};
+	struct rpc_message msg = {
+		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_LISTXATTRS],
+		.rpc_argp	= &arg,
+		.rpc_resp	= &res,
+	};
+	u32 xdrlen;
+	int ret, np;
+
+
+	res.scratch = alloc_page(GFP_KERNEL);
+	if (!res.scratch)
+		return -ENOMEM;
+
+	xdrlen = nfs42_listxattr_xdrsize(buflen);
+	if (xdrlen > server->lxasize)
+		xdrlen = server->lxasize;
+	np = xdrlen / PAGE_SIZE + 1;
+
+	pages = kcalloc(np, sizeof(struct page *), GFP_KERNEL);
+	if (pages == NULL) {
+		__free_page(res.scratch);
+		return -ENOMEM;
+	}
+
+	arg.xattr_pages = pages;
+	arg.count = xdrlen;
+
+	ret = nfs4_call_sync(server->client, server, &msg, &arg.seq_args,
+	    &res.seq_res, 0);
+
+	if (ret >= 0) {
+		ret = res.copied;
+		*cookiep = res.cookie;
+		*eofp = res.eof;
+	}
+
+	while (--np >= 0) {
+		if (pages[np])
+			__free_page(pages[np]);
+	}
+
+	__free_page(res.scratch);
+	kfree(pages);
+
+	return ret;
+
+}
+
+ssize_t nfs42_proc_getxattr(struct inode *inode, const char *name,
+			      void *buf, size_t buflen)
+{
+	struct nfs4_exception exception = { };
+	ssize_t err;
+
+	do {
+		err = _nfs42_proc_getxattr(inode, name, buf, buflen);
+		if (err >= 0)
+			break;
+		err = nfs4_handle_exception(NFS_SERVER(inode), err,
+				&exception);
+	} while (exception.retry);
+
+	return err;
+}
+
+int nfs42_proc_setxattr(struct inode *inode, const char *name,
+			      const void *buf, size_t buflen, int flags)
+{
+	struct nfs4_exception exception = { };
+	int err;
+
+	do {
+		err = _nfs42_proc_setxattr(inode, name, buf, buflen, flags);
+		if (!err)
+			break;
+		err = nfs4_handle_exception(NFS_SERVER(inode), err,
+				&exception);
+	} while (exception.retry);
+
+	return err;
+}
+
+ssize_t nfs42_proc_listxattrs(struct inode *inode, void *buf,
+			      size_t buflen, u64 *cookiep, bool *eofp)
+{
+	struct nfs4_exception exception = { };
+	ssize_t err;
+
+	do {
+		err = _nfs42_proc_listxattrs(inode, buf, buflen,
+		    cookiep, eofp);
+		if (err >= 0)
+			break;
+		err = nfs4_handle_exception(NFS_SERVER(inode), err,
+				&exception);
+	} while (exception.retry);
+
+	return err;
+}
+
+int nfs42_proc_removexattr(struct inode *inode, const char *name)
+{
+	struct nfs4_exception exception = { };
+	int err;
+
+	do {
+		err = _nfs42_proc_removexattr(inode, name);
+		if (!err)
+			break;
+		err = nfs4_handle_exception(NFS_SERVER(inode), err,
+				&exception);
+	} while (exception.retry);
+
+	return err;
+}
diff --git a/fs/nfs/nfs42xattr.c b/fs/nfs/nfs42xattr.c
new file mode 100644
index 000000000000..86777996cfec
--- /dev/null
+++ b/fs/nfs/nfs42xattr.c
@@ -0,0 +1,1056 @@
+// SPDX-License-Identifier: GPL-2.0
+
+/*
+ * Copyright 2019, 2020 Amazon.com, Inc. or its affiliates. All rights reserved.
+ *
+ * User extended attribute client side cache functions.
+ *
+ * Author: Frank van der Linden <fllinden@amazon.com>
+ */
+#include <linux/errno.h>
+#include <linux/nfs_fs.h>
+#include <linux/hashtable.h>
+#include <linux/refcount.h>
+#include <uapi/linux/xattr.h>
+
+#include "nfs4_fs.h"
+#include "internal.h"
+
+/*
+ * User extended attributes client side caching is implemented by having
+ * a cache structure attached to NFS inodes. This structure is allocated
+ * when needed, and freed when the cache is zapped.
+ *
+ * The cache structure contains as hash table of entries, and a pointer
+ * to a special-cased entry for the listxattr cache.
+ *
+ * Accessing and allocating / freeing the caches is done via reference
+ * counting. The cache entries use a similar refcounting scheme.
+ *
+ * This makes freeing a cache, both from the shrinker and from the
+ * zap cache path, easy. It also means that, in current use cases,
+ * the large majority of inodes will not waste any memory, as they
+ * will never have any user extended attributes assigned to them.
+ *
+ * Attribute entries are hashed in to a simple hash table. They are
+ * also part of an LRU.
+ *
+ * There are three shrinkers.
+ *
+ * Two shrinkers deal with the cache entries themselves: one for
+ * large entries (> PAGE_SIZE), and one for smaller entries. The
+ * shrinker for the larger entries works more aggr