NFS client updates for Linux 4.13

Stable bugfixes:
 - Fix -EACCESS on commit to DS handling
 - Fix initialization of nfs_page_array->npages
 - Only invalidate dentries that are actually invalid
 
 Features:
 - Enable NFSoRDMA transparent state migration
 - Add support for lookup-by-filehandle
 - Add support for nfs re-exporting
 
 Other bugfixes and cleanups:
 - Christoph cleaned up the way we declare NFS operations
 - Clean up various internal structures
 - Various cleanups to commits
 - Various improvements to error handling
 - Set the dt_type of . and .. entries in NFS v4
 - Make slot allocation more reliable
 - Fix fscache stat printing
 - Fix uninitialized variable warnings
 - Fix potential list overrun in nfs_atomic_open()
 - Fix a race in NFSoRDMA RPC reply handler
 - Fix return size for nfs42_proc_copy()
 - Fix against MAC forgery timing attacks
 -----BEGIN PGP SIGNATURE-----
 
 iQIzBAABCAAdFiEEnZ5MQTpR7cLU7KEp18tUv7ClQOsFAlln4jEACgkQ18tUv7Cl
 QOv2ZxAAwbQN9Dtx4rOZmPe0Xszua23sNN0ja891PodkCjIiZrRelZhLIBAf1rfP
 uSR+jTD8EsBHGt3bzTXg2DHz+o8cGDZuH+uuZX+wRWJPQcKA2pC7zElqnse8nmn5
 4Z1UUdzf42vE4NZ/G1ucqpEiAmOqGJ3s7pCRLLXPvOSSQXqOhiomNDAcGxX05FIv
 Ly4Kr6RIfg/O4oNOZBuuL/tZHodeyOj1vbyjt/4bDQ5MEXlUQfcjJZEsz/2EcNh6
 rAgbquxr1pGCD072pPBwYNH2vLGbgNN41KDDMGI0clp+8p6EhV6BOlgcEoGtZM86
 c0yro2oBOB2vPCv9nGr6JgTOHPKG6ksJ7vWVXrtQEjBGP82AbFfAawLgqZ6Ae8dP
 Sqpx55j4xdm4nyNglCuhq5PlPAogARq/eibR+RbY973Lhzr5bZb3XqlairCkNNEv
 4RbTlxbWjhgrKJ56jVf+KpUDJAVG5viKMD7YDx/bOfLtvPwALbozD7ONrunz5v43
 PgQEvWvVtnQAKp27pqHemTsLFhU6M6eGUEctRnAfB/0ogWZh1X8QXgulpDlqG3kb
 g12kr5hfA0pSfcB0aGXVzJNnHKfW3IY3WBWtxq4xaMY22YkHtuB+78+9/yk3jCAi
 dvimjT2Ko9fE9MnltJ/hC5BU+T+xUxg+1vfwWnKMvMH8SIqjyu4=
 =OpLj
 -----END PGP SIGNATURE-----

Merge tag 'nfs-for-4.13-1' of git://git.linux-nfs.org/projects/anna/linux-nfs

Pull NFS client updates from Anna Schumaker:
 "Stable bugfixes:
   - Fix -EACCESS on commit to DS handling
   - Fix initialization of nfs_page_array->npages
   - Only invalidate dentries that are actually invalid

  Features:
   - Enable NFSoRDMA transparent state migration
   - Add support for lookup-by-filehandle
   - Add support for nfs re-exporting

  Other bugfixes and cleanups:
   - Christoph cleaned up the way we declare NFS operations
   - Clean up various internal structures
   - Various cleanups to commits
   - Various improvements to error handling
   - Set the dt_type of . and .. entries in NFS v4
   - Make slot allocation more reliable
   - Fix fscache stat printing
   - Fix uninitialized variable warnings
   - Fix potential list overrun in nfs_atomic_open()
   - Fix a race in NFSoRDMA RPC reply handler
   - Fix return size for nfs42_proc_copy()
   - Fix against MAC forgery timing attacks"

* tag 'nfs-for-4.13-1' of git://git.linux-nfs.org/projects/anna/linux-nfs: (68 commits)
  NFS: Don't run wake_up_bit() when nobody is waiting...
  nfs: add export operations
  nfs4: add NFSv4 LOOKUPP handlers
  nfs: add a nfs_ilookup helper
  nfs: replace d_add with d_splice_alias in atomic_open
  sunrpc: use constant time memory comparison for mac
  NFSv4.2 fix size storage for nfs42_proc_copy
  xprtrdma: Fix documenting comments in frwr_ops.c
  xprtrdma: Replace PAGE_MASK with offset_in_page()
  xprtrdma: FMR does not need list_del_init()
  xprtrdma: Demote "connect" log messages
  NFSv4.1: Use seqid returned by EXCHANGE_ID after state migration
  NFSv4.1: Handle EXCHGID4_FLAG_CONFIRMED_R during NFSv4.1 migration
  xprtrdma: Don't defer MR recovery if ro_map fails
  xprtrdma: Fix FRWR invalidation error recovery
  xprtrdma: Fix client lock-up after application signal fires
  xprtrdma: Rename rpcrdma_req::rl_free
  xprtrdma: Pass only the list of registered MRs to ro_unmap_sync
  xprtrdma: Pre-mark remotely invalidated MRs
  xprtrdma: On invalidation failure, remove MWs from rl_registered
  ...

fs/nfs/Makefile

@@ -7,7 +7,7 @@ obj-$(CONFIG_NFS_FS) += nfs.o
 CFLAGS_nfstrace.o += -I$(src)
 nfs-y 			:= client.o dir.o file.o getroot.o inode.o super.o \
 			   io.o direct.o pagelist.o read.o symlink.o unlink.o \
-			   write.o namespace.o mount_clnt.o nfstrace.o
+			   write.o namespace.o mount_clnt.o nfstrace.o export.o
 nfs-$(CONFIG_ROOT_NFS)	+= nfsroot.o
 nfs-$(CONFIG_SYSCTL)	+= sysctl.o
 nfs-$(CONFIG_NFS_FSCACHE) += fscache.o fscache-index.o

fs/nfs/dir.c

@@ -151,7 +151,7 @@ struct nfs_cache_array {
 	struct nfs_cache_array_entry array[0];
 };
 
-typedef int (*decode_dirent_t)(struct xdr_stream *, struct nfs_entry *, int);
+typedef int (*decode_dirent_t)(struct xdr_stream *, struct nfs_entry *, bool);
 typedef struct {
 	struct file	*file;
 	struct page	*page;
@@ -165,8 +165,8 @@ typedef struct {
 	unsigned long	timestamp;
 	unsigned long	gencount;
 	unsigned int	cache_entry_index;
-	unsigned int	plus:1;
+	bool plus;
-	unsigned int	eof:1;
+	bool eof;
 } nfs_readdir_descriptor_t;
 
 /*
@@ -355,7 +355,7 @@ int nfs_readdir_xdr_filler(struct page **pages, nfs_readdir_descriptor_t *desc,
 		if (error == -ENOTSUPP && desc->plus) {
 			NFS_SERVER(inode)->caps &= ~NFS_CAP_READDIRPLUS;
 			clear_bit(NFS_INO_ADVISE_RDPLUS, &NFS_I(inode)->flags);
-			desc->plus = 0;
+			desc->plus = false;
 			goto again;
 		}
 		goto error;
@@ -557,7 +557,7 @@ int nfs_readdir_page_filler(nfs_readdir_descriptor_t *desc, struct nfs_entry *en
 
 		count++;
 
-		if (desc->plus != 0)
+		if (desc->plus)
 			nfs_prime_dcache(file_dentry(desc->file), entry);
 
 		status = nfs_readdir_add_to_array(entry, page);
@@ -860,7 +860,7 @@ static int nfs_readdir(struct file *file, struct dir_context *ctx)
 	desc->ctx = ctx;
 	desc->dir_cookie = &dir_ctx->dir_cookie;
 	desc->decode = NFS_PROTO(inode)->decode_dirent;
-	desc->plus = nfs_use_readdirplus(inode, ctx) ? 1 : 0;
+	desc->plus = nfs_use_readdirplus(inode, ctx);
 
 	if (ctx->pos == 0 || nfs_attribute_cache_expired(inode))
 		res = nfs_revalidate_mapping(inode, file->f_mapping);
@@ -885,8 +885,8 @@ static int nfs_readdir(struct file *file, struct dir_context *ctx)
 			clear_bit(NFS_INO_ADVISE_RDPLUS, &NFS_I(inode)->flags);
 			nfs_zap_caches(inode);
 			desc->page_index = 0;
-			desc->plus = 0;
+			desc->plus = false;
-			desc->eof = 0;
+			desc->eof = false;
 			continue;
 		}
 		if (res < 0)
@@ -1115,11 +1115,13 @@ static int nfs_lookup_revalidate(struct dentry *dentry, unsigned int flags)
 	/* Force a full look up iff the parent directory has changed */
 	if (!nfs_is_exclusive_create(dir, flags) &&
 	    nfs_check_verifier(dir, dentry, flags & LOOKUP_RCU)) {
-
+		error = nfs_lookup_verify_inode(inode, flags);
-		if (nfs_lookup_verify_inode(inode, flags)) {
+		if (error) {
 			if (flags & LOOKUP_RCU)
 				return -ECHILD;
-			goto out_zap_parent;
+			if (error == -ESTALE)
+				goto out_zap_parent;
+			goto out_error;
 		}
 		nfs_advise_use_readdirplus(dir);
 		goto out_valid;
@@ -1144,8 +1146,10 @@ static int nfs_lookup_revalidate(struct dentry *dentry, unsigned int flags)
 	trace_nfs_lookup_revalidate_enter(dir, dentry, flags);
 	error = NFS_PROTO(dir)->lookup(dir, &dentry->d_name, fhandle, fattr, label);
 	trace_nfs_lookup_revalidate_exit(dir, dentry, flags, error);
-	if (error)
+	if (error == -ESTALE || error == -ENOENT)
 		goto out_bad;
+	if (error)
+		goto out_error;
 	if (nfs_compare_fh(NFS_FH(inode), fhandle))
 		goto out_bad;
 	if ((error = nfs_refresh_inode(inode, fattr)) != 0)
@@ -1427,8 +1431,10 @@ static int nfs_finish_open(struct nfs_open_context *ctx,
 	err = finish_open(file, dentry, do_open, opened);
 	if (err)
 		goto out;
-	nfs_file_set_open_context(file, ctx);
+	if (S_ISREG(file->f_path.dentry->d_inode->i_mode))
-
+		nfs_file_set_open_context(file, ctx);
+	else
+		err = -ESTALE;
 out:
 	return err;
 }
@@ -1512,7 +1518,7 @@ int nfs_atomic_open(struct inode *dir, struct dentry *dentry,
 		d_drop(dentry);
 		switch (err) {
 		case -ENOENT:
-			d_add(dentry, NULL);
+			d_splice_alias(NULL, dentry);
 			nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
 			break;
 		case -EISDIR:
@@ -2035,7 +2041,11 @@ int nfs_rename(struct inode *old_dir, struct dentry *old_dentry,
 	}
 
 	error = rpc_wait_for_completion_task(task);
-	if (error == 0)
+	if (error != 0) {
+		((struct nfs_renamedata *)task->tk_calldata)->cancelled = 1;
+		/* Paired with the atomic_dec_and_test() barrier in rpc_do_put_task() */
+		smp_wmb();
+	} else
 		error = task->tk_status;
 	rpc_put_task(task);
 	nfs_mark_for_revalidate(old_inode);

fs/nfs/export.c

@@ -0,0 +1,177 @@
+/*
+ * Copyright (c) 2015, Primary Data, Inc. All rights reserved.
+ *
+ * Tao Peng <bergwolf@primarydata.com>
+ */
+#include <linux/dcache.h>
+#include <linux/exportfs.h>
+#include <linux/nfs.h>
+#include <linux/nfs_fs.h>
+
+#include "internal.h"
+#include "nfstrace.h"
+
+#define NFSDBG_FACILITY		NFSDBG_VFS
+
+enum {
+	FILEID_HIGH_OFF = 0,	/* inode fileid high */
+	FILEID_LOW_OFF,		/* inode fileid low */
+	FILE_I_TYPE_OFF,	/* inode type */
+	EMBED_FH_OFF		/* embeded server fh */
+};
+
+
+static struct nfs_fh *nfs_exp_embedfh(__u32 *p)
+{
+	return (struct nfs_fh *)(p + EMBED_FH_OFF);
+}
+
+/*
+ * Let's break subtree checking for now... otherwise we'll have to embed parent fh
+ * but there might not be enough space.
+ */
+static int
+nfs_encode_fh(struct inode *inode, __u32 *p, int *max_len, struct inode *parent)
+{
+	struct nfs_fh *server_fh = NFS_FH(inode);
+	struct nfs_fh *clnt_fh = nfs_exp_embedfh(p);
+	size_t fh_size = offsetof(struct nfs_fh, data) + server_fh->size;
+	int len = EMBED_FH_OFF + XDR_QUADLEN(fh_size);
+
+	dprintk("%s: max fh len %d inode %p parent %p",
+		__func__, *max_len, inode, parent);
+
+	if (*max_len < len || IS_AUTOMOUNT(inode)) {
+		dprintk("%s: fh len %d too small, required %d\n",
+			__func__, *max_len, len);
+		*max_len = len;
+		return FILEID_INVALID;
+	}
+	if (IS_AUTOMOUNT(inode)) {
+		*max_len = FILEID_INVALID;
+		goto out;
+	}
+
+	p[FILEID_HIGH_OFF] = NFS_FILEID(inode) >> 32;
+	p[FILEID_LOW_OFF] = NFS_FILEID(inode);
+	p[FILE_I_TYPE_OFF] = inode->i_mode & S_IFMT;
+	p[len - 1] = 0; /* Padding */
+	nfs_copy_fh(clnt_fh, server_fh);
+	*max_len = len;
+out:
+	dprintk("%s: result fh fileid %llu mode %u size %d\n",
+		__func__, NFS_FILEID(inode), inode->i_mode, *max_len);
+	return *max_len;
+}
+
+static struct dentry *
+nfs_fh_to_dentry(struct super_block *sb, struct fid *fid,
+		 int fh_len, int fh_type)
+{
+	struct nfs4_label *label = NULL;
+	struct nfs_fattr *fattr = NULL;
+	struct nfs_fh *server_fh = nfs_exp_embedfh(fid->raw);
+	size_t fh_size = offsetof(struct nfs_fh, data) + server_fh->size;
+	const struct nfs_rpc_ops *rpc_ops;
+	struct dentry *dentry;
+	struct inode *inode;
+	int len = EMBED_FH_OFF + XDR_QUADLEN(fh_size);
+	u32 *p = fid->raw;
+	int ret;
+
+	/* NULL translates to ESTALE */
+	if (fh_len < len || fh_type != len)
+		return NULL;
+
+	fattr = nfs_alloc_fattr();
+	if (fattr == NULL) {
+		dentry = ERR_PTR(-ENOMEM);
+		goto out;
+	}
+
+	fattr->fileid = ((u64)p[FILEID_HIGH_OFF] << 32) + p[FILEID_LOW_OFF];
+	fattr->mode = p[FILE_I_TYPE_OFF];
+	fattr->valid |= NFS_ATTR_FATTR_FILEID | NFS_ATTR_FATTR_TYPE;
+
+	dprintk("%s: fileid %llu mode %d\n", __func__, fattr->fileid, fattr->mode);
+
+	inode = nfs_ilookup(sb, fattr, server_fh);
+	if (inode)
+		goto out_found;
+
+	label = nfs4_label_alloc(NFS_SB(sb), GFP_KERNEL);
+	if (IS_ERR(label)) {
+		dentry = ERR_CAST(label);
+		goto out_free_fattr;
+	}
+
+	rpc_ops = NFS_SB(sb)->nfs_client->rpc_ops;
+	ret = rpc_ops->getattr(NFS_SB(sb), server_fh, fattr, label);
+	if (ret) {
+		dprintk("%s: getattr failed %d\n", __func__, ret);
+		dentry = ERR_PTR(ret);
+		goto out_free_label;
+	}
+
+	inode = nfs_fhget(sb, server_fh, fattr, label);
+
+out_found:
+	dentry = d_obtain_alias(inode);
+
+out_free_label:
+	nfs4_label_free(label);
+out_free_fattr:
+	nfs_free_fattr(fattr);
+out:
+	return dentry;
+}
+
+static struct dentry *
+nfs_get_parent(struct dentry *dentry)
+{
+	int ret;
+	struct inode *inode = d_inode(dentry), *pinode;
+	struct super_block *sb = inode->i_sb;
+	struct nfs_server *server = NFS_SB(sb);
+	struct nfs_fattr *fattr = NULL;
+	struct nfs4_label *label = NULL;
+	struct dentry *parent;
+	struct nfs_rpc_ops const *ops = server->nfs_client->rpc_ops;
+	struct nfs_fh fh;
+
+	if (!ops->lookupp)
+		return ERR_PTR(-EACCES);
+
+	fattr = nfs_alloc_fattr();
+	if (fattr == NULL) {
+		parent = ERR_PTR(-ENOMEM);
+		goto out;
+	}
+
+	label = nfs4_label_alloc(server, GFP_KERNEL);
+	if (IS_ERR(label)) {
+		parent = ERR_CAST(label);
+		goto out_free_fattr;
+	}
+
+	ret = ops->lookupp(inode, &fh, fattr, label);
+	if (ret) {
+		parent = ERR_PTR(ret);
+		goto out_free_label;
+	}
+
+	pinode = nfs_fhget(sb, &fh, fattr, label);
+	parent = d_obtain_alias(pinode);
+out_free_label:
+	nfs4_label_free(label);
+out_free_fattr:
+	nfs_free_fattr(fattr);
+out:
+	return parent;
+}
+
+const struct export_operations nfs_export_ops = {
+	.encode_fh = nfs_encode_fh,
+	.fh_to_dentry = nfs_fh_to_dentry,
+	.get_parent = nfs_get_parent,
+};

fs/nfs/filelayout/filelayout.c

@@ -126,32 +126,13 @@ static int filelayout_async_handle_error(struct rpc_task *task,
 {
 	struct pnfs_layout_hdr *lo = lseg->pls_layout;
 	struct inode *inode = lo->plh_inode;
-	struct nfs_server *mds_server = NFS_SERVER(inode);
 	struct nfs4_deviceid_node *devid = FILELAYOUT_DEVID_NODE(lseg);
-	struct nfs_client *mds_client = mds_server->nfs_client;
 	struct nfs4_slot_table *tbl = &clp->cl_session->fc_slot_table;
 
 	if (task->tk_status >= 0)
 		return 0;
 
 	switch (task->tk_status) {
-	/* MDS state errors */
-	case -NFS4ERR_DELEG_REVOKED:
-	case -NFS4ERR_ADMIN_REVOKED:
-	case -NFS4ERR_BAD_STATEID:
-	case -NFS4ERR_OPENMODE:
-		if (state == NULL)
-			break;
-		if (nfs4_schedule_stateid_recovery(mds_server, state) < 0)
-			goto out_bad_stateid;
-		goto wait_on_recovery;
-	case -NFS4ERR_EXPIRED:
-		if (state != NULL) {
-			if (nfs4_schedule_stateid_recovery(mds_server, state) < 0)
-				goto out_bad_stateid;
-		}
-		nfs4_schedule_lease_recovery(mds_client);
-		goto wait_on_recovery;
 	/* DS session errors */
 	case -NFS4ERR_BADSESSION:
 	case -NFS4ERR_BADSLOT:
@@ -172,6 +153,7 @@ static int filelayout_async_handle_error(struct rpc_task *task,
 	case -NFS4ERR_RETRY_UNCACHED_REP:
 		break;
 	/* Invalidate Layout errors */
+	case -NFS4ERR_ACCESS:
 	case -NFS4ERR_PNFS_NO_LAYOUT:
 	case -ESTALE:           /* mapped NFS4ERR_STALE */
 	case -EBADHANDLE:       /* mapped NFS4ERR_BADHANDLE */
@@ -202,26 +184,17 @@ static int filelayout_async_handle_error(struct rpc_task *task,
 			task->tk_status);
 		nfs4_mark_deviceid_unavailable(devid);
 		pnfs_error_mark_layout_for_return(inode, lseg);
+		pnfs_set_lo_fail(lseg);
 		rpc_wake_up(&tbl->slot_tbl_waitq);
 		/* fall through */
 	default:
-		pnfs_set_lo_fail(lseg);
 reset:
 		dprintk("%s Retry through MDS. Error %d\n", __func__,
 			task->tk_status);
 		return -NFS4ERR_RESET_TO_MDS;
 	}
-out:
 	task->tk_status = 0;
 	return -EAGAIN;
-out_bad_stateid:
-	task->tk_status = -EIO;
-	return 0;
-wait_on_recovery:
-	rpc_sleep_on(&mds_client->cl_rpcwaitq, task, NULL);
-	if (test_bit(NFS4CLNT_MANAGER_RUNNING, &mds_client->cl_state) == 0)
-		rpc_wake_up_queued_task(&mds_client->cl_rpcwaitq, task);
-	goto out;
 }
 
 /* NFS_PROTO call done callback routines */

fs/nfs/flexfilelayout/flexfilelayout.c

@@ -1050,34 +1050,10 @@ static int ff_layout_async_handle_error_v4(struct rpc_task *task,
 {
 	struct pnfs_layout_hdr *lo = lseg->pls_layout;
 	struct inode *inode = lo->plh_inode;
-	struct nfs_server *mds_server = NFS_SERVER(inode);
-
 	struct nfs4_deviceid_node *devid = FF_LAYOUT_DEVID_NODE(lseg, idx);
-	struct nfs_client *mds_client = mds_server->nfs_client;
 	struct nfs4_slot_table *tbl = &clp->cl_session->fc_slot_table;
 
 	switch (task->tk_status) {
-	/* MDS state errors */
-	case -NFS4ERR_DELEG_REVOKED:
-	case -NFS4ERR_ADMIN_REVOKED:
-	case -NFS4ERR_BAD_STATEID:
-		if (state == NULL)
-			break;
-		nfs_remove_bad_delegation(state->inode, NULL);
-	case -NFS4ERR_OPENMODE:
-		if (state == NULL)
-			break;
-		if (nfs4_schedule_stateid_recovery(mds_server, state) < 0)
-			goto out_bad_stateid;
-		goto wait_on_recovery;
-	case -NFS4ERR_EXPIRED:
-		if (state != NULL) {
-			if (nfs4_schedule_stateid_recovery(mds_server, state) < 0)
-				goto out_bad_stateid;
-		}
-		nfs4_schedule_lease_recovery(mds_client);
-		goto wait_on_recovery;
-	/* DS session errors */
 	case -NFS4ERR_BADSESSION:
 	case -NFS4ERR_BADSLOT:
 	case -NFS4ERR_BAD_HIGH_SLOT:
@@ -1137,17 +1113,8 @@ reset:
 			task->tk_status);
 		return -NFS4ERR_RESET_TO_MDS;
 	}
-out:
 	task->tk_status = 0;
 	return -EAGAIN;
-out_bad_stateid:
-	task->tk_status = -EIO;
-	return 0;
-wait_on_recovery:
-	rpc_sleep_on(&mds_client->cl_rpcwaitq, task, NULL);
-	if (test_bit(NFS4CLNT_MANAGER_RUNNING, &mds_client->cl_state) == 0)
-		rpc_wake_up_queued_task(&mds_client->cl_rpcwaitq, task);
-	goto out;
 }
 
 /* Retry all errors through either pNFS or MDS except for -EJUKEBOX */

fs/nfs/inode.c

@@ -386,6 +386,28 @@ void nfs_setsecurity(struct inode *inode, struct nfs_fattr *fattr,
 #endif
 EXPORT_SYMBOL_GPL(nfs_setsecurity);
 
+/* Search for inode identified by fh, fileid and i_mode in inode cache. */
+struct inode *
+nfs_ilookup(struct super_block *sb, struct nfs_fattr *fattr, struct nfs_fh *fh)
+{
+	struct nfs_find_desc desc = {
+		.fh	= fh,
+		.fattr	= fattr,
+	};
+	struct inode *inode;
+	unsigned long hash;
+
+	if (!(fattr->valid & NFS_ATTR_FATTR_FILEID) ||
+	    !(fattr->valid & NFS_ATTR_FATTR_TYPE))
+		return NULL;
+
+	hash = nfs_fattr_to_ino_t(fattr);
+	inode = ilookup5(sb, hash, nfs_find_actor, &desc);
+
+	dprintk("%s: returning %p\n", __func__, inode);
+	return inode;
+}
+
 /*
  * This is our front-end to iget that looks up inodes by file handle
  * instead of inode number.
@@ -525,8 +547,14 @@ nfs_fhget(struct super_block *sb, struct nfs_fh *fh, struct nfs_fattr *fattr, st
 		nfs_fscache_init_inode(inode);
 
 		unlock_new_inode(inode);
-	} else
+	} else {
-		nfs_refresh_inode(inode, fattr);
+		int err = nfs_refresh_inode(inode, fattr);
+		if (err < 0) {
+			iput(inode);
+			inode = ERR_PTR(err);
+			goto out_no_inode;
+		}
+	}
 	dprintk("NFS: nfs_fhget(%s/%Lu fh_crc=0x%08x ct=%d)\n",
 		inode->i_sb->s_id,
 		(unsigned long long)NFS_FILEID(inode),
@@ -1315,9 +1343,9 @@ static int nfs_check_inode_attributes(struct inode *inode, struct nfs_fattr *fat
 		return 0;
 	/* Has the inode gone and changed behind our back? */
 	if ((fattr->valid & NFS_ATTR_FATTR_FILEID) && nfsi->fileid != fattr->fileid)
-		return -EIO;
+		return -ESTALE;
 	if ((fattr->valid & NFS_ATTR_FATTR_TYPE) && (inode->i_mode & S_IFMT) != (fattr->mode & S_IFMT))
-		return -EIO;
+		return -ESTALE;
 
 	if (!nfs_file_has_buffered_writers(nfsi)) {
 		/* Verify a few of the more important attributes */

fs/nfs/internal.h

@@ -11,6 +11,8 @@
 
 #define NFS_MS_MASK (MS_RDONLY|MS_NOSUID|MS_NODEV|MS_NOEXEC|MS_SYNCHRONOUS)
 
+extern const struct export_operations nfs_export_ops;
+
 struct nfs_string;
 
 /* Maximum number of readahead requests
@@ -273,17 +275,17 @@ static inline bool nfs_match_open_context(const struct nfs_open_context *ctx1,
 /* nfs2xdr.c */
 extern const struct rpc_procinfo nfs_procedures[];
 extern int nfs2_decode_dirent(struct xdr_stream *,
-				struct nfs_entry *, int);
+				struct nfs_entry *, bool);
 
 /* nfs3xdr.c */
 extern const struct rpc_procinfo nfs3_procedures[];
 extern int nfs3_decode_dirent(struct xdr_stream *,
-				struct nfs_entry *, int);
+				struct nfs_entry *, bool);
 
 /* nfs4xdr.c */
 #if IS_ENABLED(CONFIG_NFS_V4)
 extern int nfs4_decode_dirent(struct xdr_stream *,
-				struct nfs_entry *, int);
+				struct nfs_entry *, bool);
 #endif
 #ifdef CONFIG_NFS_V4_1
 extern const u32 nfs41_maxread_overhead;

fs/nfs/nfs2xdr.c

@@ -939,7 +939,7 @@ static int nfs2_xdr_dec_writeres(struct rpc_rqst *req, struct xdr_stream *xdr,
  *	};
  */
 int nfs2_decode_dirent(struct xdr_stream *xdr, struct nfs_entry *entry,
-		       int plus)
+		       bool plus)
 {
 	__be32 *p;
 	int error;

fs/nfs/nfs3proc.c

@@ -621,7 +621,7 @@ out:
  */
 static int
 nfs3_proc_readdir(struct dentry *dentry, struct rpc_cred *cred,
-		  u64 cookie, struct page **pages, unsigned int count, int plus)
+		  u64 cookie, struct page **pages, unsigned int count, bool plus)
 {
 	struct inode		*dir = d_inode(dentry);
 	__be32			*verf = NFS_I(dir)->cookieverf;

fs/nfs/nfs3xdr.c

@@ -1991,7 +1991,7 @@ out_status:
  *	};
  */
 int nfs3_decode_dirent(struct xdr_stream *xdr, struct nfs_entry *entry,
-		       int plus)
+		       bool plus)
 {
 	struct nfs_entry old = *entry;
 	__be32 *p;

fs/nfs/nfs42proc.c

@@ -146,7 +146,7 @@ static ssize_t _nfs42_proc_copy(struct file *src,
 	loff_t pos_src = args->src_pos;
 	loff_t pos_dst = args->dst_pos;
 	size_t count = args->count;
-	int status;
+	ssize_t status;
 
 	status = nfs4_set_rw_stateid(&args->src_stateid, src_lock->open_context,
 				     src_lock, FMODE_READ);

fs/nfs/nfs4client.c

@@ -414,6 +414,7 @@ struct nfs_client *nfs4_init_client(struct nfs_client *clp,
 	if (clp != old)
 		clp->cl_preserve_clid = true;
 	nfs_put_client(clp);
+	clear_bit(NFS_CS_TSM_POSSIBLE, &clp->cl_flags);
 	return old;
 
 error:
@@ -852,6 +853,8 @@ static int nfs4_set_client(struct nfs_server *server,
 		set_bit(NFS_CS_NORESVPORT, &cl_init.init_flags);
 	if (server->options & NFS_OPTION_MIGRATION)
 		set_bit(NFS_CS_MIGRATION, &cl_init.init_flags);
+	if (test_bit(NFS_MIG_TSM_POSSIBLE, &server->mig_status))
+		set_bit(NFS_CS_TSM_POSSIBLE, &cl_init.init_flags);
 
 	/* Allocate or find a client reference we can use */
 	clp = nfs_get_client(&cl_init);
@@ -1212,9 +1215,11 @@ int nfs4_update_server(struct nfs_server *server, const char *hostname,
 		return -EAFNOSUPPORT;
 
 	nfs_server_remove_lists(server);
+	set_bit(NFS_MIG_TSM_POSSIBLE, &server->mig_status);
 	error = nfs4_set_client(server, hostname, sap, salen, buf,
 				clp->cl_proto, clnt->cl_timeout,
 				clp->cl_minorversion, net);
+	clear_bit(NFS_MIG_TSM_POSSIBLE, &server->mig_status);
 	nfs_put_client(clp);
 	if (error != 0) {
 		nfs_server_insert_lists(server);

fs/nfs/nfs4idmap.c

@@ -364,7 +364,8 @@ static int nfs_idmap_lookup_id(const char *name, size_t namelen, const char *typ
 		ret = -EINVAL;
 	} else {
 		ret = kstrtol(id_str, 10, &id_long);
-		*id = (__u32)id_long;
+		if (!ret)
+			*id = (__u32)id_long;
 	}
 	return ret;
 }

fs/nfs/nfs4proc.c

@@ -275,6 +275,7 @@ const u32 nfs4_fs_locations_bitmap[3] = {
 static void nfs4_setup_readdir(u64 cookie, __be32 *verifier, struct dentry *dentry,
 		struct nfs4_readdir_arg *readdir)
 {
+	unsigned int attrs = FATTR4_WORD0_FILEID | FATTR4_WORD0_TYPE;
 	__be32 *start, *p;
 
 	if (cookie > 2) {
@@ -305,8 +306,9 @@ static void nfs4_setup_readdir(u64 cookie, __be32 *verifier, struct dentry *dent
 		memcpy(p, ".\0\0\0", 4);                        /* entry */
 		p++;
 		*p++ = xdr_one;                         /* bitmap length */
-		*p++ = htonl(FATTR4_WORD0_FILEID);             /* bitmap */
+		*p++ = htonl(attrs);                           /* bitmap */
-		*p++ = htonl(8);              /* attribute buffer length */
+		*p++ = htonl(12);             /* attribute buffer length */
+		*p++ = htonl(NF4DIR);
 		p = xdr_encode_hyper(p, NFS_FILEID(d_inode(dentry)));
 	}
 	
@@ -317,8 +319,9 @@ static void nfs4_setup_readdir(u64 cookie, __be32 *verifier, struct dentry *dent
 	memcpy(p, "..\0\0", 4);                         /* entry */
 	p++;
 	*p++ = xdr_one;                         /* bitmap length */
-	*p++ = htonl(FATTR4_WORD0_FILEID);             /* bitmap */
+	*p++ = htonl(attrs);                           /* bitmap */
-	*p++ = htonl(8);              /* attribute buffer length */
+	*p++ = htonl(12);             /* attribute buffer length */
+	*p++ = htonl(NF4DIR);
 	p = xdr_encode_hyper(p, NFS_FILEID(d_inode(dentry->d_parent)));
 
 	readdir->pgbase = (char *)p - (char *)start;
@@ -1034,11 +1037,11 @@ struct nfs4_opendata {
 	struct nfs4_state *state;
 	struct iattr attrs;
 	unsigned long timestamp;
-	unsigned int rpc_done : 1;
+	bool rpc_done;
-	unsigned int file_created : 1;
+	bool file_created;
-	unsigned int is_recover : 1;
+	bool is_recover;
+	bool cancelled;
 	int rpc_status;
-	int cancelled;
 };
 
 static bool nfs4_clear_cap_atomic_open_v1(struct nfs_server *server,
@@ -1962,7 +1965,7 @@ static void nfs4_open_confirm_done(struct rpc_task *task, void *calldata)
 		nfs4_stateid_copy(&data->o_res.stateid, &data->c_res.stateid);
 		nfs_confirm_seqid(&data->owner->so_seqid, 0);
 		renew_lease(data->o_res.server, data->timestamp);
-		data->rpc_done = 1;
+		data->rpc_done = true;
 	}
 }
 
@@ -1972,7 +1975,7 @@ static void nfs4_open_confirm_release(void *calldata)
 	struct nfs4_state *state = NULL;
 
 	/* If this request hasn't been cancelled, do nothing */
-	if (data->cancelled == 0)
+	if (!data->cancelled)
 		goto out_free;
 	/* In case of error, no cleanup! */
 	if (!data->rpc_done)
@@ -2015,7 +2018,7 @@ static int _nfs4_proc_open_confirm(struct nfs4_opendata *data)
 
 	nfs4_init_sequence(&data->c_arg.seq_args, &data->c_res.seq_res, 1);
 	kref_get(&data->kref);
-	data->rpc_done = 0;
+	data->rpc_done = false;
 	data->rpc_status = 0;
 	data->timestamp = jiffies;
 	if (data->is_recover)
@@ -2025,7 +2028,7 @@ static int _nfs4_proc_open_confirm(struct nfs4_opendata *data)
 		return PTR_ERR(task);
 	status = rpc_wait_for_completion_task(task);
 	if (status != 0) {
-		data->cancelled = 1;
+		data->cancelled = true;
 		smp_wmb();
 	} else
 		status = data->rpc_status;
@@ -2124,7 +2127,7 @@ static void nfs4_open_done(struct rpc_task *task, void *calldata)
 		if (!(data->o_res.rflags & NFS4_OPEN_RESULT_CONFIRM))
 			nfs_confirm_seqid(&data->owner->so_seqid, 0);
 	}
-	data->rpc_done = 1;
+	data->rpc_done = true;
 }
 
 static void nfs4_open_release(void *calldata)
@@ -2133,7 +2136,7 @@ static void nfs4_open_release(void *calldata)
 	struct nfs4_state *state = NULL;
 
 	/* If this request hasn't been cancelled, do nothing */
-	if (data->cancelled == 0)
+	if (!data->cancelled)
 		goto out_free;
 	/* In case of error, no cleanup! */
 	if (data->rpc_status != 0 || !data->rpc_done)
@@ -2179,20 +2182,20 @@ static int nfs4_run_open_task(struct nfs4_opendata *data, int isrecover)
 
 	nfs4_init_sequence(&o_arg->seq_args, &o_res->seq_res, 1);
 	kref_get(&data->kref);
-	data->rpc_done = 0;
+	data->rpc_done = false;
 	data->rpc_status = 0;
-	data->cancelled = 0;
+	data->cancelled = false;
-	data->is_recover = 0;
+	data->is_recover = false;
 	if (isrecover) {
 		nfs4_set_sequence_privileged(&o_arg->seq_args);
-		data->is_recover = 1;
+		data->is_recover = true;
 	}
 	task = rpc_run_task(&task_setup_data);
 	if (IS_ERR(task))
 		return PTR_ERR(task);
 	status = rpc_wait_for_completion_task(task);
 	if (status != 0) {
-		data->cancelled = 1;
+		data->cancelled = true;
 		smp_wmb();
 	} else
 		status = data->rpc_status;
@@ -2287,9 +2290,9 @@ static int _nfs4_proc_open(struct nfs4_opendata *data)
 
 	if (o_arg->open_flags & O_CREAT) {
 		if (o_arg->open_flags & O_EXCL)
-			data->file_created = 1;
+			data->file_created = true;
 		else if (o_res->cinfo.before != o_res->cinfo.after)
-			data->file_created = 1;
+			data->file_created = true;
 		if (data->file_created || dir->i_version != o_res->cinfo.after)
 			update_changeattr(dir, &o_res->cinfo,
 					o_res->f_attr->time_start);
@@ -3803,6 +3806,54 @@ nfs4_proc_lookup_mountpoint(struct inode *dir, const struct qstr *name,
 	return (client == NFS_CLIENT(dir)) ? rpc_clone_client(client) : client;
 }
 
+static int _nfs4_proc_lookupp(struct inode *inode,
+		struct nfs_fh *fhandle, struct nfs_fattr *fattr,
+		struct nfs4_label *label)
+{
+	struct rpc_clnt *clnt = NFS_CLIENT(inode);
+	struct nfs_server *server = NFS_SERVER(inode);
+	int		       status;
+	struct nfs4_lookupp_arg args = {
+		.bitmask = server->attr_bitmask,
+		.fh = NFS_FH(inode),
+	};
+	struct nfs4_lookupp_res res = {
+		.server = server,
+		.fattr = fattr,
+		.label = label,
+		.fh = fhandle,
+	};
+	struct rpc_message msg = {
+		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOOKUPP],
+		.rpc_argp = &args,
+		.rpc_resp = &res,
+	};
+
+	args.bitmask = nfs4_bitmask(server, label);
+
+	nfs_fattr_init(fattr);
+
+	dprintk("NFS call  lookupp ino=0x%lx\n", inode->i_ino);
+	status = nfs4_call_sync(clnt, server, &msg, &args.seq_args,
+				&res.seq_res, 0);
+	dprintk("NFS reply lookupp: %d\n", status);
+	return status;
+}
+
+static int nfs4_proc_lookupp(struct inode *inode, struct nfs_fh *fhandle,
+			     struct nfs_fattr *fattr, struct nfs4_label *label)
+{
+	struct nfs4_exception exception = { };
+	int err;
+	do {
+		err = _nfs4_proc_lookupp(inode, fhandle, fattr, label);
+		trace_nfs4_lookupp(inode, err);
+		err = nfs4_handle_exception(NFS_SERVER(inode), err,
+				&exception);
+	} while (exception.retry);
+	return err;
+}
+
 static int _nfs4_proc_access(struct inode *inode, struct nfs_access_entry *entry)
 {
 	struct nfs_server *server = NFS_SERVER(inode);
@@ -4273,7 +4324,7 @@ static int nfs4_proc_mkdir(struct inode *dir, struct dentry *dentry,
 }
 
 static int _nfs4_proc_readdir(struct dentry *dentry, struct rpc_cred *cred,
-		u64 cookie, struct page **pages, unsigned int count, int plus)
+		u64 cookie, struct page **pages, unsigned int count, bool plus)
 {
 	struct inode		*dir = d_inode(dentry);
 	struct nfs4_readdir_arg args = {
@@ -4311,7 +4362,7 @@ static int _nfs4_proc_readdir(struct dentry *dentry, struct rpc_cred *cred,
 }
 
 static int nfs4_proc_readdir(struct dentry *dentry, struct rpc_cred *cred,
-		u64 cookie, struct page **pages, unsigned int count, int plus)
+		u64 cookie, struct page **pages, unsigned int count, bool plus)
 {
 	struct nfs4_exception exception = { };
 	int err;
@@ -6135,7 +6186,7 @@ static void nfs4_lock_release(void *calldata)
 
 	dprintk("%s: begin!\n", __func__);
 	nfs_free_seqid(data->arg.open_seqid);
-	if (data->cancelled != 0) {
+	if (data->cancelled) {
 		struct rpc_task *task;
 		task = nfs4_do_unlck(&data->fl, data->ctx, data->lsp,
 				data->arg.lock_seqid);
@@ -6218,7 +6269,7 @@ static int _nfs4_do_setlk(struct nfs4_state *state, int cmd, struct file_lock *f
 			nfs4_handle_setlk_error(data->server, data->lsp,
 					data->arg.new_lock_owner, ret);
 	} else
-		data->cancelled = 1;
+		data->cancelled = true;
 	rpc_put_task(task);
 	dprintk("%s: done, ret = %d!\n", __func__, ret);
 	trace_nfs4_set_lock(fl, state, &data->res.stateid, cmd, ret);
@@ -7376,12 +7427,11 @@ static void nfs4_exchange_id_done(struct rpc_task *task, void *data)
 	if (status == 0) {
 		clp->cl_clientid = cdata->res.clientid;
 		clp->cl_exchange_flags = cdata->res.flags;
+		clp->cl_seqid = cdata->res.seqid;
 		/* Client ID is not confirmed */
-		if (!(cdata->res.flags & EXCHGID4_FLAG_CONFIRMED_R)) {
+		if (!(cdata->res.flags & EXCHGID4_FLAG_CONFIRMED_R))
 			clear_bit(NFS4_SESSION_ESTABLISHED,
-			&clp->cl_session->session_state);
+				  &clp->cl_session->session_state);
-			clp->cl_seqid = cdata->res.seqid;
-		}
 
 		kfree(clp->cl_serverowner);
 		clp->cl_serverowner = cdata->res.server_owner;
@@ -9313,6 +9363,7 @@ const struct nfs_rpc_ops nfs_v4_clientops = {
 	.getattr	= nfs4_proc_getattr,
 	.setattr	= nfs4_proc_setattr,
 	.lookup		= nfs4_proc_lookup,
+	.lookupp	= nfs4_proc_lookupp,
 	.access		= nfs4_proc_access,
 	.readlink	= nfs4_proc_readlink,
 	.create		= nfs4_proc_create,

fs/nfs/nfs4state.c

@@ -352,11 +352,17 @@ int nfs41_discover_server_trunking(struct nfs_client *clp,
 	if (clp != *result)
 		return 0;
 
-	/* Purge state if the client id was established in a prior instance */
+	/*
-	if (clp->cl_exchange_flags & EXCHGID4_FLAG_CONFIRMED_R)
+	 * Purge state if the client id was established in a prior
-		set_bit(NFS4CLNT_PURGE_STATE, &clp->cl_state);
+	 * instance and the client id could not have arrived on the
-	else
+	 * server via Transparent State Migration.
-		set_bit(NFS4CLNT_LEASE_CONFIRM, &clp->cl_state);
+	 */
+	if (clp->cl_exchange_flags & EXCHGID4_FLAG_CONFIRMED_R) {
+		if (!test_bit(NFS_CS_TSM_POSSIBLE, &clp->cl_flags))
+			set_bit(NFS4CLNT_PURGE_STATE, &clp->cl_state);
+		else
+			set_bit(NFS4CLNT_LEASE_CONFIRM, &clp->cl_state);
+	}
 	nfs4_schedule_state_manager(clp);
 	status = nfs_wait_client_init_complete(clp);
 	if (status < 0)

fs/nfs/nfs4trace.h

@@ -891,6 +891,35 @@ DEFINE_NFS4_LOOKUP_EVENT(nfs4_remove);
 DEFINE_NFS4_LOOKUP_EVENT(nfs4_get_fs_locations);
 DEFINE_NFS4_LOOKUP_EVENT(nfs4_secinfo);
 
+TRACE_EVENT(nfs4_lookupp,
+		TP_PROTO(
+			const struct inode *inode,
+			int error
+		),
+
+		TP_ARGS(inode, error),
+
+		TP_STRUCT__entry(
+			__field(dev_t, dev)
+			__field(u64, ino)
+			__field(int, error)
+		),
+
+		TP_fast_assign(
+			__entry->dev = inode->i_sb->s_dev;
+			__entry->ino = NFS_FILEID(inode);
+			__entry->error = error;
+		),
+
+		TP_printk(
+			"error=%d (%s) inode=%02x:%02x:%llu",
+			__entry->error,
+			show_nfsv4_errors(__entry->error),
+			MAJOR(__entry->dev), MINOR(__entry->dev),
+			(unsigned long long)__entry->ino
+		)
+);
+
 TRACE_EVENT(nfs4_rename,
 		TP_PROTO(
 			const struct inode *olddir,

fs/nfs/nfs4xdr.c

@@ -159,6 +159,8 @@ static int nfs4_stat_to_errno(int);
 				(op_decode_hdr_maxsz)
 #define encode_lookup_maxsz	(op_encode_hdr_maxsz + nfs4_name_maxsz)
 #define decode_lookup_maxsz	(op_decode_hdr_maxsz)
+#define encode_lookupp_maxsz	(op_encode_hdr_maxsz)
+#define decode_lookupp_maxsz	(op_decode_hdr_maxsz)
 #define encode_share_access_maxsz \
 				(2)
 #define encode_createmode_maxsz	(1 + encode_attrs_maxsz + encode_verifier_maxsz)
@@ -618,6 +620,18 @@ static int nfs4_stat_to_errno(int);
 				decode_lookup_maxsz + \
 				decode_getattr_maxsz + \
 				decode_getfh_maxsz)
+#define NFS4_enc_lookupp_sz	(compound_encode_hdr_maxsz + \
+				encode_sequence_maxsz + \
+				encode_putfh_maxsz + \
+				encode_lookupp_maxsz + \
+				encode_getattr_maxsz + \
+				encode_getfh_maxsz)
+#define NFS4_dec_lookupp_sz	(compound_decode_hdr_maxsz + \
+				decode_sequence_maxsz + \
+				decode_putfh_maxsz + \
+				decode_lookupp_maxsz + \
+				decode_getattr_maxsz + \
+				decode_getfh_maxsz)
 #define NFS4_enc_lookup_root_sz (compound_encode_hdr_maxsz + \
 				encode_sequence_maxsz + \
 				encode_putrootfh_maxsz + \
@@ -1368,6 +1382,11 @@ static void encode_lookup(struct xdr_stream *xdr, const struct qstr *name, struc
 	encode_string(xdr, name->len, name->name);
 }
 
+static void encode_lookupp(struct xdr_stream *xdr, struct compound_hdr *hdr)
+{
+	encode_op_hdr(xdr, OP_LOOKUPP, decode_lookupp_maxsz, hdr);
+}
+
 static void encode_share_access(struct xdr_stream *xdr, u32 share_access)
 {
 	__be32 *p;
@@ -2122,6 +2141,26 @@ static void nfs4_xdr_enc_lookup(struct rpc_rqst *req, struct xdr_stream *xdr,
 	encode_nops(&hdr);
 }
 
+/*
+ * Encode LOOKUPP request
+ */
+static void nfs4_xdr_enc_lookupp(struct rpc_rqst *req, struct xdr_stream *xdr,
+		const void *data)
+{
+	const struct nfs4_lookupp_arg *args = data;
+	struct compound_hdr hdr = {
+		.minorversion = nfs4_xdr_minorversion(&args->seq_args),
+	};
+
+	encode_compound_hdr(xdr, req, &hdr);
+	encode_sequence(xdr, &args->seq_args, &hdr);
+	encode_putfh(xdr, args->fh, &hdr);
+	encode_lookupp(xdr, &hdr);
+	encode_getfh(xdr, &hdr);
+	encode_getfattr(xdr, args->bitmask, &hdr);
+	encode_nops(&hdr);
+}
+
 /*
  * Encode LOOKUP_ROOT request
  */
@@ -5058,6 +5097,11 @@ static int decode_lookup(struct xdr_stream *xdr)
 	return decode_op_hdr(xdr, OP_LOOKUP);
 }
 
+static int decode_lookupp(struct xdr_stream *xdr)
+{
+	return decode_op_hdr(xdr, OP_LOOKUPP);
+}
+
 /* This is too sick! */
 static int decode_space_limit(struct xdr_stream *xdr,
 		unsigned long *pagemod_limit)
@@ -6237,6 +6281,36 @@ out:
 	return status;
 }
 
+/*
+ * Decode LOOKUPP response
+ */
+static int nfs4_xdr_dec_lookupp(struct rpc_rqst *rqstp, struct xdr_stream *xdr,
+		void *data)
+{
+	struct nfs4_lookupp_res *res = data;
+	struct compound_hdr hdr;
+	int status;
+
+	status = decode_compound_hdr(xdr, &hdr);
+	if (status)
+		goto out;
+	status = decode_sequence(xdr, &res->seq_res, rqstp);
+	if (status)
+		goto out;
+	status = decode_putfh(xdr);
+	if (status)
+		goto out;
+	status = decode_lookupp(xdr);
+	if (status)
+		goto out;
+	status = decode_getfh(xdr, res->fh);
+	if (status)
+		goto out;
+	status = decode_getfattr_label(xdr, res->fattr, res->label, res->server);
+out:
+	return status;
+}
+
 /*
  * Decode LOOKUP_ROOT response
  */
@@ -7447,7 +7521,7 @@ out:
  * on a directory already in our cache.
  */
 int nfs4_decode_dirent(struct xdr_stream *xdr, struct nfs_entry *entry,
-		       int plus)
+		       bool plus)
 {
 	unsigned int savep;
 	uint32_t bitmap[3] = {0};
@@ -7614,6 +7688,7 @@ const struct rpc_procinfo nfs4_procedures[] = {
 	PROC(ACCESS,		enc_access,		dec_access),
 	PROC(GETATTR,		enc_getattr,		dec_getattr),
 	PROC(LOOKUP,		enc_lookup,		dec_lookup),
+	PROC(LOOKUPP,		enc_lookupp,		dec_lookupp),
 	PROC(LOOKUP_ROOT,	enc_lookup_root,	dec_lookup_root),
 	PROC(REMOVE,		enc_remove,		dec_remove),
 	PROC(RENAME,		enc_rename,		dec_rename),

fs/nfs/pagelist.c

@@ -50,8 +50,8 @@ void nfs_pgheader_init(struct nfs_pageio_descriptor *desc,
 	hdr->cred = hdr->req->wb_context->cred;
 	hdr->io_start = req_offset(hdr->req);
 	hdr->good_bytes = mirror->pg_count;
+	hdr->io_completion = desc->pg_io_completion;
 	hdr->dreq = desc->pg_dreq;
-	hdr->layout_private = desc->pg_layout_private;
 	hdr->release = release;
 	hdr->completion_ops = desc->pg_completion_ops;
 	if (hdr->completion_ops->init_hdr)
@@ -155,9 +155,12 @@ nfs_page_group_lock(struct nfs_page *req, bool nonblock)
 	if (!test_and_set_bit(PG_HEADLOCK, &head->wb_flags))
 		return 0;
 
-	if (!nonblock)
+	if (!nonblock) {
+		set_bit(PG_CONTENDED1, &head->wb_flags);
+		smp_mb__after_atomic();
 		return wait_on_bit_lock(&head->wb_flags, PG_HEADLOCK,
 				TASK_UNINTERRUPTIBLE);
+	}
 
 	return -EAGAIN;
 }
@@ -175,6 +178,10 @@ nfs_page_group_lock_wait(struct nfs_page *req)
 
 	WARN_ON_ONCE(head != head->wb_head);
 
+	if (!test_bit(PG_HEADLOCK, &head->wb_flags))
+		return;
+	set_bit(PG_CONTENDED1, &head->wb_flags);
+	smp_mb__after_atomic();
 	wait_on_bit(&head->wb_flags, PG_HEADLOCK,
 		TASK_UNINTERRUPTIBLE);
 }
@@ -193,6 +200,8 @@ nfs_page_group_unlock(struct nfs_page *req)
 	smp_mb__before_atomic();
 	clear_bit(PG_HEADLOCK, &head->wb_flags);
 	smp_mb__after_atomic();
+	if (!test_bit(PG_CONTENDED1, &head->wb_flags))
+		return;
 	wake_up_bit(&head->wb_flags, PG_HEADLOCK);
 }
 
@@ -383,6 +392,8 @@ void nfs_unlock_request(struct nfs_page *req)
 	smp_mb__before_atomic();
 	clear_bit(PG_BUSY, &req->wb_flags);
 	smp_mb__after_atomic();
+	if (!test_bit(PG_CONTENDED2, &req->wb_flags))
+		return;
 	wake_up_bit(&req->wb_flags, PG_BUSY);
 }
 
@@ -465,6 +476,10 @@ void nfs_release_request(struct nfs_page *req)
 int
 nfs_wait_on_request(struct nfs_page *req)
 {
+	if (!test_bit(PG_BUSY, &req->wb_flags))
+		return 0;
+	set_bit(PG_CONTENDED2, &req->wb_flags);
+	smp_mb__after_atomic();
 	return wait_on_bit_io(&req->wb_flags, PG_BUSY,
 			      TASK_UNINTERRUPTIBLE);
 }
@@ -710,8 +725,8 @@ void nfs_pageio_init(struct nfs_pageio_descriptor *desc,
 	desc->pg_ioflags = io_flags;
 	desc->pg_error = 0;
 	desc->pg_lseg = NULL;
+	desc->pg_io_completion = NULL;
 	desc->pg_dreq = NULL;
-	desc->pg_layout_private = NULL;
 	desc->pg_bsize = bsize;
 
 	desc->pg_mirror_count = 1;
@@ -779,6 +794,7 @@ int nfs_generic_pgio(struct nfs_pageio_descriptor *desc,
 	gfp_t gfp_flags = GFP_KERNEL;
 
 	pagecount = nfs_page_array_len(mirror->pg_base, mirror->pg_count);
+	pg_array->npages = pagecount;
 
 	if (pagecount <= ARRAY_SIZE(pg_array->page_array))
 		pg_array->pagevec = pg_array->page_array;
@@ -1233,6 +1249,7 @@ int nfs_pageio_resend(struct nfs_pageio_descriptor *desc,
 {
 	LIST_HEAD(failed);
 
+	desc->pg_io_completion = hdr->io_completion;
 	desc->pg_dreq = hdr->dreq;
 	while (!list_empty(&hdr->pages)) {
 		struct nfs_page *req = nfs_list_entry(hdr->pages.next);

fs/nfs/proc.c

@@ -485,7 +485,7 @@ nfs_proc_rmdir(struct inode *dir, const struct qstr *name)
  */
 static int
 nfs_proc_readdir(struct dentry *dentry, struct rpc_cred *cred,
-		 u64 cookie, struct page **pages, unsigned int count, int plus)
+		 u64 cookie, struct page **pages, unsigned int count, bool plus)
 {
 	struct inode		*dir = d_inode(dentry);
 	struct nfs_readdirargs	arg = {

fs/nfs/super.c

@@ -879,7 +879,7 @@ int nfs_show_stats(struct seq_file *m, struct dentry *root)
 	if (nfss->options & NFS_OPTION_FSCACHE) {
 		seq_printf(m, "\n\tfsc:\t");
 		for (i = 0; i < __NFSIOS_FSCACHEMAX; i++)
-			seq_printf(m, "%Lu ", totals.bytes[i]);
+			seq_printf(m, "%Lu ", totals.fscache[i]);
 	}
 #endif
 	seq_printf(m, "\n");
@@ -2339,6 +2339,7 @@ void nfs_fill_super(struct super_block *sb, struct nfs_mount_info *mount_info)
 		 */
 		sb->s_flags |= MS_POSIXACL;
 		sb->s_time_gran = 1;
+		sb->s_export_op = &nfs_export_ops;
 	}
 
  	nfs_initialise_sb(sb);
@@ -2360,6 +2361,7 @@ static void nfs_clone_super(struct super_block *sb,
 	sb->s_xattr = old_sb->s_xattr;
 	sb->s_op = old_sb->s_op;
 	sb->s_time_gran = 1;
+	sb->s_export_op = old_sb->s_export_op;
 
 	if (server->nfs_client->rpc_ops->version != 2) {
 		/* The VFS shouldn't apply the umask to mode bits. We will do

fs/nfs/unlink.c

@@ -288,6 +288,19 @@ static void nfs_async_rename_release(void *calldata)
 	if (d_really_is_positive(data->old_dentry))
 		nfs_mark_for_revalidate(d_inode(data->old_dentry));
 
+	/* The result of the rename is unknown. Play it safe by
+	 * forcing a new lookup */
+	if (data->cancelled) {
+		spin_lock(&data->old_dir->i_lock);
+		nfs_force_lookup_revalidate(data->old_dir);
+		spin_unlock(&data->old_dir->i_lock);
+		if (data->new_dir != data->old_dir) {
+			spin_lock(&data->new_dir->i_lock);
+			nfs_force_lookup_revalidate(data->new_dir);
+			spin_unlock(&data->new_dir->i_lock);
+		}
+	}
+
 	dput(data->old_dentry);
 	dput(data->new_dentry);
 	iput(data->old_dir);

fs/nfs/write.c

@@ -40,6 +40,12 @@
 #define MIN_POOL_WRITE		(32)
 #define MIN_POOL_COMMIT		(4)
 
+struct nfs_io_completion {
+	void (*complete)(void *data);
+	void *data;
+	struct kref refcount;
+};
+
 /*
  * Local function declarations
  */
@@ -108,6 +114,39 @@ static void nfs_writehdr_free(struct nfs_pgio_header *hdr)
 	mempool_free(hdr, nfs_wdata_mempool);
 }
 
+static struct nfs_io_completion *nfs_io_completion_alloc(gfp_t gfp_flags)
+{
+	return kmalloc(sizeof(struct nfs_io_completion), gfp_flags);
+}
+
+static void nfs_io_completion_init(struct nfs_io_completion *ioc,
+		void (*complete)(void *), void *data)
+{
+	ioc->complete = complete;
+	ioc->data = data;
+	kref_init(&ioc->refcount);
+}
+
+static void nfs_io_completion_release(struct kref *kref)
+{
+	struct nfs_io_completion *ioc = container_of(kref,
+			struct nfs_io_completion, refcount);
+	ioc->complete(ioc->data);
+	kfree(ioc);
+}
+
+static void nfs_io_completion_get(struct nfs_io_completion *ioc)
+{
+	if (ioc != NULL)
+		kref_get(&ioc->refcount);
+}
+
+static void nfs_io_completion_put(struct nfs_io_completion *ioc)
+{
+	if (ioc != NULL)
+		kref_put(&ioc->refcount, nfs_io_completion_release);
+}
+
 static void nfs_context_set_write_error(struct nfs_open_context *ctx, int error)
 {
 	ctx->error = error;
@@ -681,18 +720,29 @@ static int nfs_writepages_callback(struct page *page, struct writeback_control *
 	return ret;
 }
 
+static void nfs_io_completion_commit(void *inode)
+{
+	nfs_commit_inode(inode, 0);
+}
+
 int nfs_writepages(struct address_space *mapping, struct writeback_control *wbc)
 {
 	struct inode *inode = mapping->host;
 	struct nfs_pageio_descriptor pgio;
+	struct nfs_io_completion *ioc = nfs_io_completion_alloc(GFP_NOFS);
 	int err;
 
 	nfs_inc_stats(inode, NFSIOS_VFSWRITEPAGES);
 
+	if (ioc)
+		nfs_io_completion_init(ioc, nfs_io_completion_commit, inode);
+
 	nfs_pageio_init_write(&pgio, inode, wb_priority(wbc), false,
 				&nfs_async_write_completion_ops);
+	pgio.pg_io_completion = ioc;
 	err = write_cache_pages(mapping, wbc, nfs_writepages_callback, &pgio);
 	nfs_pageio_complete(&pgio);
+	nfs_io_completion_put(ioc);
 
 	if (err < 0)
 		goto out_err;
@@ -940,6 +990,11 @@ int nfs_write_need_commit(struct nfs_pgio_header *hdr)
 	return hdr->verf.committed != NFS_FILE_SYNC;
 }
 
+static void nfs_async_write_init(struct nfs_pgio_header *hdr)
+{
+	nfs_io_completion_get(hdr->io_completion);
+}
+
 static void nfs_write_completion(struct nfs_pgio_header *hdr)
 {
 	struct nfs_commit_info cinfo;
@@ -973,6 +1028,7 @@ next:
 		nfs_release_request(req);
 	}
 out:
+	nfs_io_completion_put(hdr->io_completion);
 	hdr->release(hdr);
 }
 
@@ -1378,6 +1434,7 @@ static void nfs_async_write_reschedule_io(struct nfs_pgio_header *hdr)
 }
 
 static const struct nfs_pgio_completion_ops nfs_async_write_completion_ops = {
+	.init_hdr = nfs_async_write_init,
 	.error_cleanup = nfs_async_write_error,
 	.completion = nfs_write_completion,
 	.reschedule_io = nfs_async_write_reschedule_io,
@@ -1884,7 +1941,7 @@ int nfs_write_inode(struct inode *inode, struct writeback_control *wbc)
 		/* Don't commit yet if this is a non-blocking flush and there
 		 * are a lot of outstanding writes for this mapping.
 		 */
-		if (nfsi->commit_info.ncommit <= (nfsi->nrequests >> 1))
+		if (mapping_tagged(inode->i_mapping, PAGECACHE_TAG_WRITEBACK))
 			goto out_mark_dirty;
 
 		/* don't wait for the COMMIT response */

include/linux/nfs4.h

@@ -479,6 +479,7 @@ enum {
 	NFSPROC4_CLNT_ACCESS,
 	NFSPROC4_CLNT_GETATTR,
 	NFSPROC4_CLNT_LOOKUP,
+	NFSPROC4_CLNT_LOOKUPP,
 	NFSPROC4_CLNT_LOOKUP_ROOT,
 	NFSPROC4_CLNT_REMOVE,
 	NFSPROC4_CLNT_RENAME,

include/linux/nfs_fs.h

@@ -332,6 +332,7 @@ extern void nfs_zap_caches(struct inode *);
 extern void nfs_invalidate_atime(struct inode *);
 extern struct inode *nfs_fhget(struct super_block *, struct nfs_fh *,
 				struct nfs_fattr *, struct nfs4_label *);
+struct inode *nfs_ilookup(struct super_block *sb, struct nfs_fattr *, struct nfs_fh *);
 extern int nfs_refresh_inode(struct inode *, struct nfs_fattr *);
 extern int nfs_post_op_update_inode(struct inode *inode, struct nfs_fattr *fattr);
 extern int nfs_post_op_update_inode_force_wcc(struct inode *inode, struct nfs_fattr *fattr);

include/linux/nfs_fs_sb.h

@@ -42,6 +42,7 @@ struct nfs_client {
 #define NFS_CS_MIGRATION	2		/* - transparent state migr */
 #define NFS_CS_INFINITE_SLOTS	3		/* - don't limit TCP slots */
 #define NFS_CS_NO_RETRANS_TIMEOUT	4	/* - Disable retransmit timeouts */
+#define NFS_CS_TSM_POSSIBLE	5		/* - Maybe state migration */
 	struct sockaddr_storage	cl_addr;	/* server identifier */
 	size_t			cl_addrlen;
 	char *			cl_hostname;	/* hostname of server */
@@ -210,6 +211,7 @@ struct nfs_server {
 	unsigned long		mig_status;
 #define NFS_MIG_IN_TRANSITION		(1)
 #define NFS_MIG_FAILED			(2)
+#define NFS_MIG_TSM_POSSIBLE		(3)
 
 	void (*destroy)(struct nfs_server *);
 

include/linux/nfs_page.h

@@ -33,6 +33,8 @@ enum {
 	PG_UPTODATE,		/* page group sync bit in read path */
 	PG_WB_END,		/* page group sync bit in write path */
 	PG_REMOVE,		/* page group sync bit in write path */
+	PG_CONTENDED1,		/* Is someone waiting for a lock? */
+	PG_CONTENDED2,		/* Is someone waiting for a lock? */
 };
 
 struct nfs_inode;
@@ -93,8 +95,8 @@ struct nfs_pageio_descriptor {
 	const struct rpc_call_ops *pg_rpc_callops;
 	const struct nfs_pgio_completion_ops *pg_completion_ops;
 	struct pnfs_layout_segment *pg_lseg;
+	struct nfs_io_completion *pg_io_completion;
 	struct nfs_direct_req	*pg_dreq;
-	void			*pg_layout_private;
 	unsigned int		pg_bsize;	/* default bsize for mirrors */
 
 	u32			pg_mirror_count;

include/linux/nfs_xdr.h

@@ -878,7 +878,7 @@ struct nfs3_readdirargs {
 	struct nfs_fh *		fh;
 	__u64			cookie;
 	__be32			verf[2];
-	int			plus;
+	bool			plus;
 	unsigned int            count;
 	struct page **		pages;
 };
@@ -909,7 +909,7 @@ struct nfs3_linkres {
 struct nfs3_readdirres {
 	struct nfs_fattr *	dir_attr;
 	__be32 *		verf;
-	int			plus;
+	bool			plus;
 };
 
 struct nfs3_getaclres {
@@ -1012,7 +1012,6 @@ struct nfs4_link_res {
 	struct nfs_fattr *		dir_attr;
 };
 
-
 struct nfs4_lookup_arg {
 	struct nfs4_sequence_args	seq_args;
 	const struct nfs_fh *		dir_fh;
@@ -1028,6 +1027,20 @@ struct nfs4_lookup_res {
 	struct nfs4_label		*label;
 };
 
+struct nfs4_lookupp_arg {
+	struct nfs4_sequence_args	seq_args;
+	const struct nfs_fh		*fh;
+	const u32			*bitmask;
+};
+
+struct nfs4_lookupp_res {
+	struct nfs4_sequence_res	seq_res;
+	const struct nfs_server		*server;
+	struct nfs_fattr		*fattr;
+	struct nfs_fh			*fh;
+	struct nfs4_label		*label;
+};
+
 struct nfs4_lookup_root_arg {
 	struct nfs4_sequence_args	seq_args;
 	const u32 *			bitmask;
@@ -1053,7 +1066,7 @@ struct nfs4_readdir_arg {
 	struct page **			pages;	/* zero-copy data */
 	unsigned int			pgbase;	/* zero-copy data */
 	const u32 *			bitmask;
-	int				plus;
+	bool				plus;
 };
 
 struct nfs4_readdir_res {
@@ -1422,6 +1435,7 @@ enum {
 	NFS_IOHDR_STAT,
 };
 
+struct nfs_io_completion;
 struct nfs_pgio_header {
 	struct inode		*inode;
 	struct rpc_cred		*cred;
@@ -1435,8 +1449,8 @@ struct nfs_pgio_header {
 	void (*release) (struct nfs_pgio_header *hdr);
 	const struct nfs_pgio_completion_ops *completion_ops;
 	const struct nfs_rw_ops	*rw_ops;
+	struct nfs_io_completion *io_completion;
 	struct nfs_direct_req	*dreq;
-	void			*layout_private;
 	spinlock_t		lock;
 	/* fields protected by lock */
 	int			pnfs_error;
@@ -1533,6 +1547,7 @@ struct nfs_renamedata {
 	struct nfs_fattr	new_fattr;
 	void (*complete)(struct rpc_task *, struct nfs_renamedata *);
 	long timeout;
+	bool cancelled;
 };
 
 struct nfs_access_entry;
@@ -1567,6 +1582,8 @@ struct nfs_rpc_ops {
 	int	(*lookup)  (struct inode *, const struct qstr *,
 			    struct nfs_fh *, struct nfs_fattr *,
 			    struct nfs4_label *);
+	int	(*lookupp) (struct inode *, struct nfs_fh *,
+			    struct nfs_fattr *, struct nfs4_label *);
 	int	(*access)  (struct inode *, struct nfs_access_entry *);
 	int	(*readlink)(struct inode *, struct page *, unsigned int,
 			    unsigned int);
@@ -1585,7 +1602,7 @@ struct nfs_rpc_ops {
 	int	(*mkdir)   (struct inode *, struct dentry *, struct iattr *);
 	int	(*rmdir)   (struct inode *, const struct qstr *);
 	int	(*readdir) (struct dentry *, struct rpc_cred *,
-			    u64, struct page **, unsigned int, int);
+			    u64, struct page **, unsigned int, bool);
 	int	(*mknod)   (struct inode *, struct dentry *, struct iattr *,
 			    dev_t);
 	int	(*statfs)  (struct nfs_server *, struct nfs_fh *,
@@ -1595,7 +1612,7 @@ struct nfs_rpc_ops {
 	int	(*pathconf) (struct nfs_server *, struct nfs_fh *,
 			     struct nfs_pathconf *);
 	int	(*set_capabilities)(struct nfs_server *, struct nfs_fh *);
-	int	(*decode_dirent)(struct xdr_stream *, struct nfs_entry *, int);
+	int	(*decode_dirent)(struct xdr_stream *, struct nfs_entry *, bool);
 	int	(*pgio_rpc_prepare)(struct rpc_task *,
 				    struct nfs_pgio_header *);
 	void	(*read_setup)(struct nfs_pgio_header *, struct rpc_message *);

net/sunrpc/auth_gss/gss_krb5_crypto.c

@@ -34,6 +34,7 @@
  * WARRANTIES OF MERCHANTIBILITY AND FITNESS FOR A PARTICULAR PURPOSE.
  */
 
+#include <crypto/algapi.h>
 #include <crypto/hash.h>
 #include <crypto/skcipher.h>
 #include <linux/err.h>
@@ -927,7 +928,7 @@ gss_krb5_aes_decrypt(struct krb5_ctx *kctx, u32 offset, struct xdr_buf *buf,
 	if (ret)
 		goto out_err;
 
-	if (memcmp(pkt_hmac, our_hmac, kctx->gk5e->cksumlength) != 0) {
+	if (crypto_memneq(pkt_hmac, our_hmac, kctx->gk5e->cksumlength) != 0) {
 		ret = GSS_S_BAD_SIG;
 		goto out_err;
 	}

net/sunrpc/xprt.c

@@ -1047,13 +1047,15 @@ out:
 	return ret;
 }
 
-static struct rpc_rqst *xprt_dynamic_alloc_slot(struct rpc_xprt *xprt, gfp_t gfp_flags)
+static struct rpc_rqst *xprt_dynamic_alloc_slot(struct rpc_xprt *xprt)
 {
 	struct rpc_rqst *req = ERR_PTR(-EAGAIN);
 
 	if (!atomic_add_unless(&xprt->num_reqs, 1, xprt->max_reqs))
 		goto out;
-	req = kzalloc(sizeof(struct rpc_rqst), gfp_flags);
+	spin_unlock(&xprt->reserve_lock);
+	req = kzalloc(sizeof(struct rpc_rqst), GFP_NOFS);
+	spin_lock(&xprt->reserve_lock);
 	if (req != NULL)
 		goto out;
 	atomic_dec(&xprt->num_reqs);
@@ -1081,7 +1083,7 @@ void xprt_alloc_slot(struct rpc_xprt *xprt, struct rpc_task *task)
 		list_del(&req->rq_list);
 		goto out_init_req;
 	}
-	req = xprt_dynamic_alloc_slot(xprt, GFP_NOWAIT|__GFP_NOWARN);
+	req = xprt_dynamic_alloc_slot(xprt);
 	if (!IS_ERR(req))
 		goto out_init_req;
 	switch (PTR_ERR(req)) {

net/sunrpc/xprtrdma/fmr_ops.c

@@ -91,7 +91,7 @@ __fmr_unmap(struct rpcrdma_mw *mw)
 
 	list_add(&mw->fmr.fm_mr->list, &l);
 	rc = ib_unmap_fmr(&l);
-	list_del_init(&mw->fmr.fm_mr->list);
+	list_del(&mw->fmr.fm_mr->list);
 	return rc;
 }
 
@@ -213,13 +213,11 @@ fmr_op_map(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr_seg *seg,
 		    offset_in_page((seg-1)->mr_offset + (seg-1)->mr_len))
 			break;
 	}
-	mw->mw_nents = i;
 	mw->mw_dir = rpcrdma_data_dir(writing);
-	if (i == 0)
-		goto out_dmamap_err;
 
-	if (!ib_dma_map_sg(r_xprt->rx_ia.ri_device,
+	mw->mw_nents = ib_dma_map_sg(r_xprt->rx_ia.ri_device,
-			   mw->mw_sg, mw->mw_nents, mw->mw_dir))
+				     mw->mw_sg, i, mw->mw_dir);
+	if (!mw->mw_nents)
 		goto out_dmamap_err;
 
 	for (i = 0, dma_pages = mw->fmr.fm_physaddrs; i < mw->mw_nents; i++)
@@ -237,16 +235,18 @@ fmr_op_map(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr_seg *seg,
 	return mw->mw_nents;
 
 out_dmamap_err:
-	pr_err("rpcrdma: failed to dma map sg %p sg_nents %u\n",
+	pr_err("rpcrdma: failed to DMA map sg %p sg_nents %d\n",
-	       mw->mw_sg, mw->mw_nents);
+	       mw->mw_sg, i);
-	rpcrdma_defer_mr_recovery(mw);
+	rpcrdma_put_mw(r_xprt, mw);
 	return -EIO;
 
 out_maperr:
 	pr_err("rpcrdma: ib_map_phys_fmr %u@0x%llx+%i (%d) status %i\n",
 	       len, (unsigned long long)dma_pages[0],
 	       pageoff, mw->mw_nents, rc);
-	rpcrdma_defer_mr_recovery(mw);
+	ib_dma_unmap_sg(r_xprt->rx_ia.ri_device,
+			mw->mw_sg, mw->mw_nents, mw->mw_dir);
+	rpcrdma_put_mw(r_xprt, mw);
 	return -EIO;
 }
 
@@ -255,24 +255,26 @@ out_maperr:
  * Sleeps until it is safe for the host CPU to access the
  * previously mapped memory regions.
  *
- * Caller ensures that req->rl_registered is not empty.
+ * Caller ensures that @mws is not empty before the call. This
+ * function empties the list.
  */
 static void
-fmr_op_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req)
+fmr_op_unmap_sync(struct rpcrdma_xprt *r_xprt, struct list_head *mws)
 {
-	struct rpcrdma_mw *mw, *tmp;
+	struct rpcrdma_mw *mw;
 	LIST_HEAD(unmap_list);
 	int rc;
 
-	dprintk("RPC:       %s: req %p\n", __func__, req);
-
 	/* ORDER: Invalidate all of the req's MRs first
 	 *
 	 * ib_unmap_fmr() is slow, so use a single call instead
 	 * of one call per mapped FMR.
 	 */
-	list_for_each_entry(mw, &req->rl_registered, mw_list)
+	list_for_each_entry(mw, mws, mw_list) {
+		dprintk("RPC:       %s: unmapping fmr %p\n",
+			__func__, &mw->fmr);
 		list_add_tail(&mw->fmr.fm_mr->list, &unmap_list);
+	}
 	r_xprt->rx_stats.local_inv_needed++;
 	rc = ib_unmap_fmr(&unmap_list);
 	if (rc)
@@ -281,9 +283,11 @@ fmr_op_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req)
 	/* ORDER: Now DMA unmap all of the req's MRs, and return
 	 * them to the free MW list.
 	 */
-	list_for_each_entry_safe(mw, tmp, &req->rl_registered, mw_list) {
+	while (!list_empty(mws)) {
-		list_del_init(&mw->mw_list);
+		mw = rpcrdma_pop_mw(mws);
-		list_del_init(&mw->fmr.fm_mr->list);
+		dprintk("RPC:       %s: DMA unmapping fmr %p\n",
+			__func__, &mw->fmr);
+		list_del(&mw->fmr.fm_mr->list);
 		ib_dma_unmap_sg(r_xprt->rx_ia.ri_device,
 				mw->mw_sg, mw->mw_nents, mw->mw_dir);
 		rpcrdma_put_mw(r_xprt, mw);
@@ -294,8 +298,9 @@ fmr_op_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req)
 out_reset:
 	pr_err("rpcrdma: ib_unmap_fmr failed (%i)\n", rc);
 
-	list_for_each_entry_safe(mw, tmp, &req->rl_registered, mw_list) {
+	while (!list_empty(mws)) {
-		list_del_init(&mw->fmr.fm_mr->list);
+		mw = rpcrdma_pop_mw(mws);
+		list_del(&mw->fmr.fm_mr->list);
 		fmr_op_recover_mr(mw);
 	}
 }

net/sunrpc/xprtrdma/frwr_ops.c

@@ -277,7 +277,7 @@ __frwr_sendcompletion_flush(struct ib_wc *wc, const char *wr)
 }
 
 /**
- * frwr_wc_fastreg - Invoked by RDMA provider for each polled FastReg WC
+ * frwr_wc_fastreg - Invoked by RDMA provider for a flushed FastReg WC
  * @cq:	completion queue (ignored)
  * @wc:	completed WR
  *
@@ -298,7 +298,7 @@ frwr_wc_fastreg(struct ib_cq *cq, struct ib_wc *wc)
 }
 
 /**
- * frwr_wc_localinv - Invoked by RDMA provider for each polled LocalInv WC
+ * frwr_wc_localinv - Invoked by RDMA provider for a flushed LocalInv WC
  * @cq:	completion queue (ignored)
  * @wc:	completed WR
  *
@@ -319,7 +319,7 @@ frwr_wc_localinv(struct ib_cq *cq, struct ib_wc *wc)
 }
 
 /**
- * frwr_wc_localinv - Invoked by RDMA provider for each polled LocalInv WC
+ * frwr_wc_localinv_wake - Invoked by RDMA provider for a signaled LocalInv WC
  * @cq:	completion queue (ignored)
  * @wc:	completed WR
  *
@@ -355,7 +355,7 @@ frwr_op_map(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr_seg *seg,
 	struct ib_mr *mr;
 	struct ib_reg_wr *reg_wr;
 	struct ib_send_wr *bad_wr;
-	int rc, i, n, dma_nents;
+	int rc, i, n;
 	u8 key;
 
 	mw = NULL;
@@ -391,14 +391,10 @@ frwr_op_map(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr_seg *seg,
 		    offset_in_page((seg-1)->mr_offset + (seg-1)->mr_len))
 			break;
 	}
-	mw->mw_nents = i;
 	mw->mw_dir = rpcrdma_data_dir(writing);
-	if (i == 0)
-		goto out_dmamap_err;
 
-	dma_nents = ib_dma_map_sg(ia->ri_device,
+	mw->mw_nents = ib_dma_map_sg(ia->ri_device, mw->mw_sg, i, mw->mw_dir);
-				  mw->mw_sg, mw->mw_nents, mw->mw_dir);
+	if (!mw->mw_nents)
-	if (!dma_nents)
 		goto out_dmamap_err;
 
 	n = ib_map_mr_sg(mr, mw->mw_sg, mw->mw_nents, NULL, PAGE_SIZE);
@@ -436,13 +432,14 @@ frwr_op_map(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr_seg *seg,
 	return mw->mw_nents;
 
 out_dmamap_err:
-	pr_err("rpcrdma: failed to dma map sg %p sg_nents %u\n",
+	pr_err("rpcrdma: failed to DMA map sg %p sg_nents %d\n",
-	       mw->mw_sg, mw->mw_nents);
+	       mw->mw_sg, i);
-	rpcrdma_defer_mr_recovery(mw);
+	frmr->fr_state = FRMR_IS_INVALID;
+	rpcrdma_put_mw(r_xprt, mw);
 	return -EIO;
 
 out_mapmr_err:
-	pr_err("rpcrdma: failed to map mr %p (%u/%u)\n",
+	pr_err("rpcrdma: failed to map mr %p (%d/%d)\n",
 	       frmr->fr_mr, n, mw->mw_nents);
 	rpcrdma_defer_mr_recovery(mw);
 	return -EIO;
@@ -458,21 +455,19 @@ out_senderr:
  * Sleeps until it is safe for the host CPU to access the
  * previously mapped memory regions.
  *
- * Caller ensures that req->rl_registered is not empty.
+ * Caller ensures that @mws is not empty before the call. This
+ * function empties the list.
  */
 static void
-frwr_op_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req)
+frwr_op_unmap_sync(struct rpcrdma_xprt *r_xprt, struct list_head *mws)
 {
 	struct ib_send_wr *first, **prev, *last, *bad_wr;
-	struct rpcrdma_rep *rep = req->rl_reply;
 	struct rpcrdma_ia *ia = &r_xprt->rx_ia;
 	struct rpcrdma_frmr *f;
 	struct rpcrdma_mw *mw;
 	int count, rc;
 
-	dprintk("RPC:       %s: req %p\n", __func__, req);
+	/* ORDER: Invalidate all of the MRs first
-
-	/* ORDER: Invalidate all of the req's MRs first
 	 *
 	 * Chain the LOCAL_INV Work Requests and post them with
 	 * a single ib_post_send() call.
@@ -480,11 +475,10 @@ frwr_op_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req)
 	f = NULL;
 	count = 0;
 	prev = &first;
-	list_for_each_entry(mw, &req->rl_registered, mw_list) {
+	list_for_each_entry(mw, mws, mw_list) {
 		mw->frmr.fr_state = FRMR_IS_INVALID;
 
-		if ((rep->rr_wc_flags & IB_WC_WITH_INVALIDATE) &&
+		if (mw->mw_flags & RPCRDMA_MW_F_RI)
-		    (mw->mw_handle == rep->rr_inv_rkey))
 			continue;
 
 		f = &mw->frmr;
@@ -524,18 +518,19 @@ frwr_op_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req)
 	 * unless ri_id->qp is a valid pointer.
 	 */
 	r_xprt->rx_stats.local_inv_needed++;
+	bad_wr = NULL;
 	rc = ib_post_send(ia->ri_id->qp, first, &bad_wr);
+	if (bad_wr != first)
+		wait_for_completion(&f->fr_linv_done);
 	if (rc)
 		goto reset_mrs;
 
-	wait_for_completion(&f->fr_linv_done);
+	/* ORDER: Now DMA unmap all of the MRs, and return
-
-	/* ORDER: Now DMA unmap all of the req's MRs, and return
 	 * them to the free MW list.
 	 */
 unmap:
-	while (!list_empty(&req->rl_registered)) {
+	while (!list_empty(mws)) {
-		mw = rpcrdma_pop_mw(&req->rl_registered);
+		mw = rpcrdma_pop_mw(mws);
 		dprintk("RPC:       %s: DMA unmapping frmr %p\n",
 			__func__, &mw->frmr);
 		ib_dma_unmap_sg(ia->ri_device,
@@ -546,17 +541,19 @@ unmap:
 
 reset_mrs:
 	pr_err("rpcrdma: FRMR invalidate ib_post_send returned %i\n", rc);
-	rdma_disconnect(ia->ri_id);
 
 	/* Find and reset the MRs in the LOCAL_INV WRs that did not
-	 * get posted. This is synchronous, and slow.
+	 * get posted.
 	 */
-	list_for_each_entry(mw, &req->rl_registered, mw_list) {
+	rpcrdma_init_cqcount(&r_xprt->rx_ep, -count);
-		f = &mw->frmr;
+	while (bad_wr) {
-		if (mw->mw_handle == bad_wr->ex.invalidate_rkey) {
+		f = container_of(bad_wr, struct rpcrdma_frmr,
-			__frwr_reset_mr(ia, mw);
+				 fr_invwr);
-			bad_wr = bad_wr->next;
+		mw = container_of(f, struct rpcrdma_mw, frmr);
-		}
+
+		__frwr_reset_mr(ia, mw);
+
+		bad_wr = bad_wr->next;
 	}
 	goto unmap;
 }

net/sunrpc/xprtrdma/rpc_rdma.c

@@ -141,7 +141,7 @@ static bool rpcrdma_args_inline(struct rpcrdma_xprt *r_xprt,
 
 	if (xdr->page_len) {
 		remaining = xdr->page_len;
-		offset = xdr->page_base & ~PAGE_MASK;
+		offset = offset_in_page(xdr->page_base);
 		count = 0;
 		while (remaining) {
 			remaining -= min_t(unsigned int,
@@ -222,7 +222,7 @@ rpcrdma_convert_iovs(struct rpcrdma_xprt *r_xprt, struct xdr_buf *xdrbuf,
 
 	len = xdrbuf->page_len;
 	ppages = xdrbuf->pages + (xdrbuf->page_base >> PAGE_SHIFT);
-	page_base = xdrbuf->page_base & ~PAGE_MASK;
+	page_base = offset_in_page(xdrbuf->page_base);
 	p = 0;
 	while (len && n < RPCRDMA_MAX_SEGS) {
 		if (!ppages[p]) {
@@ -540,7 +540,7 @@ rpcrdma_prepare_msg_sges(struct rpcrdma_ia *ia, struct rpcrdma_req *req,
 			goto out;
 
 		page = virt_to_page(xdr->tail[0].iov_base);
-		page_base = (unsigned long)xdr->tail[0].iov_base & ~PAGE_MASK;
+		page_base = offset_in_page(xdr->tail[0].iov_base);
 
 		/* If the content in the page list is an odd length,
 		 * xdr_write_pages() has added a pad at the beginning
@@ -557,7 +557,7 @@ rpcrdma_prepare_msg_sges(struct rpcrdma_ia *ia, struct rpcrdma_req *req,
 	 */
 	if (xdr->page_len) {
 		ppages = xdr->pages + (xdr->page_base >> PAGE_SHIFT);
-		page_base = xdr->page_base & ~PAGE_MASK;
+		page_base = offset_in_page(xdr->page_base);
 		remaining = xdr->page_len;
 		while (remaining) {
 			sge_no++;
@@ -587,7 +587,7 @@ rpcrdma_prepare_msg_sges(struct rpcrdma_ia *ia, struct rpcrdma_req *req,
 	 */
 	if (xdr->tail[0].iov_len) {
 		page = virt_to_page(xdr->tail[0].iov_base);
-		page_base = (unsigned long)xdr->tail[0].iov_base & ~PAGE_MASK;
+		page_base = offset_in_page(xdr->tail[0].iov_base);
 		len = xdr->tail[0].iov_len;
 
 map_tail:
@@ -734,6 +734,9 @@ rpcrdma_marshal_req(struct rpc_rqst *rqst)
 		rpclen = 0;
 	}
 
+	req->rl_xid = rqst->rq_xid;
+	rpcrdma_insert_req(&r_xprt->rx_buf, req);
+
 	/* This implementation supports the following combinations
 	 * of chunk lists in one RPC-over-RDMA Call message:
 	 *
@@ -875,9 +878,9 @@ rpcrdma_inline_fixup(struct rpc_rqst *rqst, char *srcp, int copy_len, int pad)
 	srcp += curlen;
 	copy_len -= curlen;
 
-	page_base = rqst->rq_rcv_buf.page_base;
+	ppages = rqst->rq_rcv_buf.pages +
-	ppages = rqst->rq_rcv_buf.pages + (page_base >> PAGE_SHIFT);
+		(rqst->rq_rcv_buf.page_base >> PAGE_SHIFT);
-	page_base &= ~PAGE_MASK;
+	page_base = offset_in_page(rqst->rq_rcv_buf.page_base);
 	fixup_copy_count = 0;
 	if (copy_len && rqst->rq_rcv_buf.page_len) {
 		int pagelist_len;
@@ -928,6 +931,24 @@ rpcrdma_inline_fixup(struct rpc_rqst *rqst, char *srcp, int copy_len, int pad)
 	return fixup_copy_count;
 }
 
+/* Caller must guarantee @rep remains stable during this call.
+ */
+static void
+rpcrdma_mark_remote_invalidation(struct list_head *mws,
+				 struct rpcrdma_rep *rep)
+{
+	struct rpcrdma_mw *mw;
+
+	if (!(rep->rr_wc_flags & IB_WC_WITH_INVALIDATE))
+		return;
+
+	list_for_each_entry(mw, mws, mw_list)
+		if (mw->mw_handle == rep->rr_inv_rkey) {
+			mw->mw_flags = RPCRDMA_MW_F_RI;
+			break; /* only one invalidated MR per RPC */
+		}
+}
+
 #if defined(CONFIG_SUNRPC_BACKCHANNEL)
 /* By convention, backchannel calls arrive via rdma_msg type
  * messages, and never populate the chunk lists. This makes
@@ -969,14 +990,16 @@ rpcrdma_reply_handler(struct work_struct *work)
 {
 	struct rpcrdma_rep *rep =
 			container_of(work, struct rpcrdma_rep, rr_work);
+	struct rpcrdma_xprt *r_xprt = rep->rr_rxprt;
+	struct rpcrdma_buffer *buf = &r_xprt->rx_buf;
+	struct rpc_xprt *xprt = &r_xprt->rx_xprt;
 	struct rpcrdma_msg *headerp;
 	struct rpcrdma_req *req;
 	struct rpc_rqst *rqst;
-	struct rpcrdma_xprt *r_xprt = rep->rr_rxprt;
-	struct rpc_xprt *xprt = &r_xprt->rx_xprt;
 	__be32 *iptr;
 	int rdmalen, status, rmerr;
 	unsigned long cwnd;
+	struct list_head mws;
 
 	dprintk("RPC:       %s: incoming rep %p\n", __func__, rep);
 
@@ -994,27 +1017,45 @@ rpcrdma_reply_handler(struct work_struct *work)
 	/* Match incoming rpcrdma_rep to an rpcrdma_req to
 	 * get context for handling any incoming chunks.
 	 */
-	spin_lock_bh(&xprt->transport_lock);
+	spin_lock(&buf->rb_lock);
-	rqst = xprt_lookup_rqst(xprt, headerp->rm_xid);
+	req = rpcrdma_lookup_req_locked(&r_xprt->rx_buf,
-	if (!rqst)
+					headerp->rm_xid);
+	if (!req)
 		goto out_nomatch;
-
-	req = rpcr_to_rdmar(rqst);
 	if (req->rl_reply)
 		goto out_duplicate;
 
-	/* Sanity checking has passed. We are now committed
+	list_replace_init(&req->rl_registered, &mws);
-	 * to complete this transaction.
+	rpcrdma_mark_remote_invalidation(&mws, rep);
+
+	/* Avoid races with signals and duplicate replies
+	 * by marking this req as matched.
 	 */
-	list_del_init(&rqst->rq_list);
+	req->rl_reply = rep;
-	spin_unlock_bh(&xprt->transport_lock);
+	spin_unlock(&buf->rb_lock);
+
 	dprintk("RPC:       %s: reply %p completes request %p (xid 0x%08x)\n",
 		__func__, rep, req, be32_to_cpu(headerp->rm_xid));
 
-	/* from here on, the reply is no longer an orphan */
+	/* Invalidate and unmap the data payloads before waking the
-	req->rl_reply = rep;
+	 * waiting application. This guarantees the memory regions
-	xprt->reestablish_timeout = 0;
+	 * are properly fenced from the server before the application
+	 * accesses the data. It also ensures proper send flow control:
+	 * waking the next RPC waits until this RPC has relinquished
+	 * all its Send Queue entries.
+	 */
+	if (!list_empty(&mws))
+		r_xprt->rx_ia.ri_ops->ro_unmap_sync(r_xprt, &mws);
 
+	/* Perform XID lookup, reconstruction of the RPC reply, and
+	 * RPC completion while holding the transport lock to ensure
+	 * the rep, rqst, and rq_task pointers remain stable.
+	 */
+	spin_lock_bh(&xprt->transport_lock);
+	rqst = xprt_lookup_rqst(xprt, headerp->rm_xid);
+	if (!rqst)
+		goto out_norqst;
+	xprt->reestablish_timeout = 0;
 	if (headerp->rm_vers != rpcrdma_version)
 		goto out_badversion;
 
@@ -1024,12 +1065,9 @@ rpcrdma_reply_handler(struct work_struct *work)
 	case rdma_msg:
 		/* never expect read chunks */
 		/* never expect reply chunks (two ways to check) */
-		/* never expect write chunks without having offered RDMA */
 		if (headerp->rm_body.rm_chunks[0] != xdr_zero ||
 		    (headerp->rm_body.rm_chunks[1] == xdr_zero &&
-		     headerp->rm_body.rm_chunks[2] != xdr_zero) ||
+		     headerp->rm_body.rm_chunks[2] != xdr_zero))
-		    (headerp->rm_body.rm_chunks[1] != xdr_zero &&
-		     list_empty(&req->rl_registered)))
 			goto badheader;
 		if (headerp->rm_body.rm_chunks[1] != xdr_zero) {
 			/* count any expected write chunks in read reply */
@@ -1066,8 +1104,7 @@ rpcrdma_reply_handler(struct work_struct *work)
 		/* never expect read or write chunks, always reply chunks */
 		if (headerp->rm_body.rm_chunks[0] != xdr_zero ||
 		    headerp->rm_body.rm_chunks[1] != xdr_zero ||
-		    headerp->rm_body.rm_chunks[2] != xdr_one ||
+		    headerp->rm_body.rm_chunks[2] != xdr_one)
-		    list_empty(&req->rl_registered))
 			goto badheader;
 		iptr = (__be32 *)((unsigned char *)headerp +
 							RPCRDMA_HDRLEN_MIN);
@@ -1093,17 +1130,6 @@ badheader:
 	}
 
 out:
-	/* Invalidate and flush the data payloads before waking the
-	 * waiting application. This guarantees the memory region is
-	 * properly fenced from the server before the application
-	 * accesses the data. It also ensures proper send flow
-	 * control: waking the next RPC waits until this RPC has
-	 * relinquished all its Send Queue entries.
-	 */
-	if (!list_empty(&req->rl_registered))
-		r_xprt->rx_ia.ri_ops->ro_unmap_sync(r_xprt, req);
-
-	spin_lock_bh(&xprt->transport_lock);
 	cwnd = xprt->cwnd;
 	xprt->cwnd = atomic_read(&r_xprt->rx_buf.rb_credits) << RPC_CWNDSHIFT;
 	if (xprt->cwnd > cwnd)
@@ -1112,7 +1138,7 @@ out:
 	xprt_complete_rqst(rqst->rq_task, status);
 	spin_unlock_bh(&xprt->transport_lock);
 	dprintk("RPC:       %s: xprt_complete_rqst(0x%p, 0x%p, %d)\n",
-			__func__, xprt, rqst, status);
+		__func__, xprt, rqst, status);
 	return;
 
 out_badstatus:
@@ -1161,26 +1187,37 @@ out_rdmaerr:
 	r_xprt->rx_stats.bad_reply_count++;
 	goto out;
 
-/* If no pending RPC transaction was matched, post a replacement
+/* The req was still available, but by the time the transport_lock
- * receive buffer before returning.
+ * was acquired, the rqst and task had been released. Thus the RPC
+ * has already been terminated.
  */
+out_norqst:
+	spin_unlock_bh(&xprt->transport_lock);
+	rpcrdma_buffer_put(req);
+	dprintk("RPC:       %s: race, no rqst left for req %p\n",
+		__func__, req);
+	return;
+
 out_shortreply:
 	dprintk("RPC:       %s: short/invalid reply\n", __func__);
 	goto repost;
 
 out_nomatch:
-	spin_unlock_bh(&xprt->transport_lock);
+	spin_unlock(&buf->rb_lock);
 	dprintk("RPC:       %s: no match for incoming xid 0x%08x len %d\n",
 		__func__, be32_to_cpu(headerp->rm_xid),
 		rep->rr_len);
 	goto repost;
 
 out_duplicate:
-	spin_unlock_bh(&xprt->transport_lock);
+	spin_unlock(&buf->rb_lock);
 	dprintk("RPC:       %s: "
 		"duplicate reply %p to RPC request %p: xid 0x%08x\n",
 		__func__, rep, req, be32_to_cpu(headerp->rm_xid));
 
+/* If no pending RPC transaction was matched, post a replacement
+ * receive buffer before returning.
+ */
 repost:
 	r_xprt->rx_stats.bad_reply_count++;
 	if (rpcrdma_ep_post_recv(&r_xprt->rx_ia, rep))

net/sunrpc/xprtrdma/transport.c

@@ -684,7 +684,8 @@ xprt_rdma_free(struct rpc_task *task)
 
 	dprintk("RPC:       %s: called on 0x%p\n", __func__, req->rl_reply);
 
-	if (unlikely(!list_empty(&req->rl_registered)))
+	rpcrdma_remove_req(&r_xprt->rx_buf, req);
+	if (!list_empty(&req->rl_registered))
 		ia->ri_ops->ro_unmap_safe(r_xprt, req, !RPC_IS_ASYNC(task));
 	rpcrdma_unmap_sges(ia, req);
 	rpcrdma_buffer_put(req);

net/sunrpc/xprtrdma/verbs.c

@@ -243,8 +243,6 @@ rpcrdma_conn_upcall(struct rdma_cm_id *id, struct rdma_cm_event *event)
 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
 	struct sockaddr *sap = (struct sockaddr *)&ep->rep_remote_addr;
 #endif
-	struct ib_qp_attr *attr = &ia->ri_qp_attr;
-	struct ib_qp_init_attr *iattr = &ia->ri_qp_init_attr;
 	int connstate = 0;
 
 	switch (event->event) {
@@ -267,7 +265,8 @@ rpcrdma_conn_upcall(struct rdma_cm_id *id, struct rdma_cm_event *event)
 		break;
 	case RDMA_CM_EVENT_DEVICE_REMOVAL:
 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
-		pr_info("rpcrdma: removing device for %pIS:%u\n",
+		pr_info("rpcrdma: removing device %s for %pIS:%u\n",
+			ia->ri_device->name,
 			sap, rpc_get_port(sap));
 #endif
 		set_bit(RPCRDMA_IAF_REMOVING, &ia->ri_flags);
@@ -282,13 +281,6 @@ rpcrdma_conn_upcall(struct rdma_cm_id *id, struct rdma_cm_event *event)
 		return 1;
 	case RDMA_CM_EVENT_ESTABLISHED:
 		connstate = 1;
-		ib_query_qp(ia->ri_id->qp, attr,
-			    IB_QP_MAX_QP_RD_ATOMIC | IB_QP_MAX_DEST_RD_ATOMIC,
-			    iattr);
-		dprintk("RPC:       %s: %d responder resources"
-			" (%d initiator)\n",
-			__func__, attr->max_dest_rd_atomic,
-			attr->max_rd_atomic);
 		rpcrdma_update_connect_private(xprt, &event->param.conn);
 		goto connected;
 	case RDMA_CM_EVENT_CONNECT_ERROR:
@@ -298,11 +290,9 @@ rpcrdma_conn_upcall(struct rdma_cm_id *id, struct rdma_cm_event *event)
 		connstate = -ENETDOWN;
 		goto connected;
 	case RDMA_CM_EVENT_REJECTED:
-#if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
+		dprintk("rpcrdma: connection to %pIS:%u rejected: %s\n",
-		pr_info("rpcrdma: connection to %pIS:%u on %s rejected: %s\n",
+			sap, rpc_get_port(sap),
-			sap, rpc_get_port(sap), ia->ri_device->name,
 			rdma_reject_msg(id, event->status));
-#endif
 		connstate = -ECONNREFUSED;
 		if (event->status == IB_CM_REJ_STALE_CONN)
 			connstate = -EAGAIN;
@@ -310,37 +300,19 @@ rpcrdma_conn_upcall(struct rdma_cm_id *id, struct rdma_cm_event *event)
 	case RDMA_CM_EVENT_DISCONNECTED:
 		connstate = -ECONNABORTED;
 connected:
-		dprintk("RPC:       %s: %sconnected\n",
-					__func__, connstate > 0 ? "" : "dis");
 		atomic_set(&xprt->rx_buf.rb_credits, 1);
 		ep->rep_connected = connstate;
 		rpcrdma_conn_func(ep);
 		wake_up_all(&ep->rep_connect_wait);
 		/*FALLTHROUGH*/
 	default:
-		dprintk("RPC:       %s: %pIS:%u (ep 0x%p): %s\n",
+		dprintk("RPC:       %s: %pIS:%u on %s/%s (ep 0x%p): %s\n",
-			__func__, sap, rpc_get_port(sap), ep,
+			__func__, sap, rpc_get_port(sap),
-			rdma_event_msg(event->event));
+			ia->ri_device->name, ia->ri_ops->ro_displayname,
+			ep, rdma_event_msg(event->event));
 		break;
 	}
 
-#if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
-	if (connstate == 1) {
-		int ird = attr->max_dest_rd_atomic;
-		int tird = ep->rep_remote_cma.responder_resources;
-
-		pr_info("rpcrdma: connection to %pIS:%u on %s, memreg '%s', %d credits, %d responders%s\n",
-			sap, rpc_get_port(sap),
-			ia->ri_device->name,
-			ia->ri_ops->ro_displayname,
-			xprt->rx_buf.rb_max_requests,
-			ird, ird < 4 && ird < tird / 2 ? " (low!)" : "");
-	} else if (connstate < 0) {
-		pr_info("rpcrdma: connection to %pIS:%u closed (%d)\n",
-			sap, rpc_get_port(sap), connstate);
-	}
-#endif
-
 	return 0;
 }
 
@@ -971,7 +943,6 @@ rpcrdma_create_req(struct rpcrdma_xprt *r_xprt)
 	if (req == NULL)
 		return ERR_PTR(-ENOMEM);
 
-	INIT_LIST_HEAD(&req->rl_free);
 	spin_lock(&buffer->rb_reqslock);
 	list_add(&req->rl_all, &buffer->rb_allreqs);
 	spin_unlock(&buffer->rb_reqslock);
@@ -1033,6 +1004,7 @@ rpcrdma_buffer_create(struct rpcrdma_xprt *r_xprt)
 	spin_lock_init(&buf->rb_recovery_lock);
 	INIT_LIST_HEAD(&buf->rb_mws);
 	INIT_LIST_HEAD(&buf->rb_all);
+	INIT_LIST_HEAD(&buf->rb_pending);
 	INIT_LIST_HEAD(&buf->rb_stale_mrs);
 	INIT_DELAYED_WORK(&buf->rb_refresh_worker,
 			  rpcrdma_mr_refresh_worker);
@@ -1055,7 +1027,7 @@ rpcrdma_buffer_create(struct rpcrdma_xprt *r_xprt)
 			goto out;
 		}
 		req->rl_backchannel = false;
-		list_add(&req->rl_free, &buf->rb_send_bufs);
+		list_add(&req->rl_list, &buf->rb_send_bufs);
 	}
 
 	INIT_LIST_HEAD(&buf->rb_recv_bufs);
@@ -1084,8 +1056,8 @@ rpcrdma_buffer_get_req_locked(struct rpcrdma_buffer *buf)
 	struct rpcrdma_req *req;
 
 	req = list_first_entry(&buf->rb_send_bufs,
-			       struct rpcrdma_req, rl_free);
+			       struct rpcrdma_req, rl_list);
-	list_del(&req->rl_free);
+	list_del_init(&req->rl_list);
 	return req;
 }
 
@@ -1187,6 +1159,7 @@ rpcrdma_get_mw(struct rpcrdma_xprt *r_xprt)
 
 	if (!mw)
 		goto out_nomws;
+	mw->mw_flags = 0;
 	return mw;
 
 out_nomws:
@@ -1267,7 +1240,7 @@ rpcrdma_buffer_put(struct rpcrdma_req *req)
 
 	spin_lock(&buffers->rb_lock);
 	buffers->rb_send_count--;
-	list_add_tail(&req->rl_free, &buffers->rb_send_bufs);
+	list_add_tail(&req->rl_list, &buffers->rb_send_bufs);
 	if (rep) {
 		buffers->rb_recv_count--;
 		list_add_tail(&rep->rr_list, &buffers->rb_recv_bufs);

net/sunrpc/xprtrdma/xprt_rdma.h

@@ -271,6 +271,7 @@ struct rpcrdma_mw {
 	struct scatterlist	*mw_sg;
 	int			mw_nents;
 	enum dma_data_direction	mw_dir;
+	unsigned long		mw_flags;
 	union {
 		struct rpcrdma_fmr	fmr;
 		struct rpcrdma_frmr	frmr;
@@ -282,6 +283,11 @@ struct rpcrdma_mw {
 	struct list_head	mw_all;
 };
 
+/* mw_flags */
+enum {
+	RPCRDMA_MW_F_RI		= 1,
+};
+
 /*
  * struct rpcrdma_req -- structure central to the request/reply sequence.
  *
@@ -334,7 +340,8 @@ enum {
 
 struct rpcrdma_buffer;
 struct rpcrdma_req {
-	struct list_head	rl_free;
+	struct list_head	rl_list;
+	__be32			rl_xid;
 	unsigned int		rl_mapped_sges;
 	unsigned int		rl_connect_cookie;
 	struct rpcrdma_buffer	*rl_buffer;
@@ -396,6 +403,7 @@ struct rpcrdma_buffer {
 	int			rb_send_count, rb_recv_count;
 	struct list_head	rb_send_bufs;
 	struct list_head	rb_recv_bufs;
+	struct list_head	rb_pending;
 	u32			rb_max_requests;
 	atomic_t		rb_credits;	/* most recent credit grant */
 
@@ -461,7 +469,7 @@ struct rpcrdma_memreg_ops {
 				  struct rpcrdma_mr_seg *, int, bool,
 				  struct rpcrdma_mw **);
 	void		(*ro_unmap_sync)(struct rpcrdma_xprt *,
-					 struct rpcrdma_req *);
+					 struct list_head *);
 	void		(*ro_unmap_safe)(struct rpcrdma_xprt *,
 					 struct rpcrdma_req *, bool);
 	void		(*ro_recover_mr)(struct rpcrdma_mw *);
@@ -544,6 +552,34 @@ void rpcrdma_destroy_req(struct rpcrdma_req *);
 int rpcrdma_buffer_create(struct rpcrdma_xprt *);
 void rpcrdma_buffer_destroy(struct rpcrdma_buffer *);
 
+static inline void
+rpcrdma_insert_req(struct rpcrdma_buffer *buffers, struct rpcrdma_req *req)
+{
+	spin_lock(&buffers->rb_lock);
+	if (list_empty(&req->rl_list))
+		list_add_tail(&req->rl_list, &buffers->rb_pending);
+	spin_unlock(&buffers->rb_lock);
+}
+
+static inline struct rpcrdma_req *
+rpcrdma_lookup_req_locked(struct rpcrdma_buffer *buffers, __be32 xid)
+{
+	struct rpcrdma_req *pos;
+
+	list_for_each_entry(pos, &buffers->rb_pending, rl_list)
+		if (pos->rl_xid == xid)
+			return pos;
+	return NULL;
+}
+
+static inline void
+rpcrdma_remove_req(struct rpcrdma_buffer *buffers, struct rpcrdma_req *req)
+{
+	spin_lock(&buffers->rb_lock);
+	list_del(&req->rl_list);
+	spin_unlock(&buffers->rb_lock);
+}
+
 struct rpcrdma_mw *rpcrdma_get_mw(struct rpcrdma_xprt *);
 void rpcrdma_put_mw(struct rpcrdma_xprt *, struct rpcrdma_mw *);
 struct rpcrdma_req *rpcrdma_buffer_get(struct rpcrdma_buffer *);