NFS client updates for Linux 3.20

Highlights include:
 
 - Fix a use-after-free in decode_cb_sequence_args()
 - Fix a compile error when #undef CONFIG_PROC_FS
 - NFSv4.1 backchannel spinlocking issue
 - Cleanups in the NFS unstable write code requested by Linus
 - NFSv4.1 fix issues when the server denies our backchannel request
 - Cleanups in create_session and bind_conn_to_session
 -----BEGIN PGP SIGNATURE-----
 Version: GnuPG v1
 
 iQIcBAABAgAGBQJU51zBAAoJEGcL54qWCgDyyvwQALp7vxXCTwC3/OSoPkhujd2C
 HGMoI2nVSl8CY0LFYxoT6auhefZCgQChmQXkGtlZwO+bkEr9rvJ6ii8lhuDR+JXF
 M0bAwOX+bNxUFkpqYYF3Q4Hi//tMJCIZdqUp2irtyFLL/qlNoN2ktoEYMIjMY5uO
 4L1fxj9KaVuMtFuqt3xeSSe41LaXxitKyefyVJfbqz5qbcPGiXzS7WKYAun4RyvM
 7rCir9kfnxxEX3+hc7xHxWeFJnW0jUMklrjNvnrubnpHE/fX2sUAnjX2hmx5bLg/
 4puxADzhPT4f3LdGDKXVWaULuTy20VksOJnB82TKJ3rLIELJixGZw88svOrFcGvt
 5CMI8BvOihwn8ov+sSj7Xedz4046btwA1YHkUcwPV3LZAlyx/FSq4nasO4Cn27yl
 OPdkcAL1YR5I83mEKA+8BOVXJuFx5vKhkwmMdReJkBmxsWbSwB/qp6caPD9DtuXI
 K0qJYWHMqN+Dv9npi0Q4WR6vmnzxV+Eq7Z2D9WPW0FL8nT3STE0eWIrNipyqOv+7
 bHptPxrrZej+i3c922bZ6hdaE2uAlwG8FPgEGhHFtm+s09RgDPDG7NiaeCutf9cQ
 9ub82Hlk9nLTqq5X3poUPV35RS6THnZcybhngNMy8F1cPSUTs+/H+l91sEW/Zhgw
 odMB/DEa9sRnZGX5rQXE
 =NieR
 -----END PGP SIGNATURE-----

Merge tag 'nfs-for-3.20-2' of git://git.linux-nfs.org/projects/trondmy/linux-nfs

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

   - Fix a use-after-free in decode_cb_sequence_args()
   - Fix a compile error when #undef CONFIG_PROC_FS
   - NFSv4.1 backchannel spinlocking issue
   - Cleanups in the NFS unstable write code requested by Linus
   - NFSv4.1 fix issues when the server denies our backchannel request
   - Cleanups in create_session and bind_conn_to_session"

* tag 'nfs-for-3.20-2' of git://git.linux-nfs.org/projects/trondmy/linux-nfs:
  NFSv4.1: Clean up bind_conn_to_session
  NFSv4.1: Always set up a forward channel when binding the session
  NFSv4.1: Don't set up a backchannel if the server didn't agree to do so
  NFSv4.1: Clean up create_session
  pnfs: Refactor the *_layout_mark_request_commit to use pnfs_layout_mark_request_commit
  NFSv4: Kill unused nfs_inode->delegation_state field
  NFS: struct nfs_commit_info.lock must always point to inode->i_lock
  nfs: Can call nfs_clear_page_commit() instead
  nfs: Provide and use helper functions for marking a page as unstable
  SUNRPC: Always manipulate rpc_rqst::rq_bc_pa_list under xprt->bc_pa_lock
  SUNRPC: Fix a compile error when #undef CONFIG_PROC_FS
  NFSv4.1: Convert open-coded array allocation calls to kmalloc_array()
  NFSv4.1: Fix a kfree() of uninitialised pointers in decode_cb_sequence_args

fs/nfs/callback_proc.c

@@ -427,6 +427,8 @@ __be32 nfs4_callback_sequence(struct cb_sequenceargs *args,
 	if (clp == NULL)
 		goto out;
 
+	if (!(clp->cl_session->flags & SESSION4_BACK_CHAN))
+		goto out;
 	tbl = &clp->cl_session->bc_slot_table;
 
 	spin_lock(&tbl->slot_tbl_lock);

fs/nfs/callback_xdr.c

@@ -313,7 +313,7 @@ __be32 decode_devicenotify_args(struct svc_rqst *rqstp,
 		goto out;
 	}
 
-	args->devs = kmalloc(n * sizeof(*args->devs), GFP_KERNEL);
+	args->devs = kmalloc_array(n, sizeof(*args->devs), GFP_KERNEL);
 	if (!args->devs) {
 		status = htonl(NFS4ERR_DELAY);
 		goto out;
@@ -415,7 +415,7 @@ static __be32 decode_rc_list(struct xdr_stream *xdr,
 			     rc_list->rcl_nrefcalls * 2 * sizeof(uint32_t));
 		if (unlikely(p == NULL))
 			goto out;
-		rc_list->rcl_refcalls = kmalloc(rc_list->rcl_nrefcalls *
+		rc_list->rcl_refcalls = kmalloc_array(rc_list->rcl_nrefcalls,
 						sizeof(*rc_list->rcl_refcalls),
 						GFP_KERNEL);
 		if (unlikely(rc_list->rcl_refcalls == NULL))
@@ -464,8 +464,10 @@ static __be32 decode_cb_sequence_args(struct svc_rqst *rqstp,
 
 		for (i = 0; i < args->csa_nrclists; i++) {
 			status = decode_rc_list(xdr, &args->csa_rclists[i]);
-			if (status)
+			if (status) {
+				args->csa_nrclists = i;
 				goto out_free;
+			}
 		}
 	}
 	status = 0;

fs/nfs/delegation.c

@@ -180,7 +180,6 @@ void nfs_inode_reclaim_delegation(struct inode *inode, struct rpc_cred *cred,
 			delegation->cred = get_rpccred(cred);
 			clear_bit(NFS_DELEGATION_NEED_RECLAIM,
 				  &delegation->flags);
-			NFS_I(inode)->delegation_state = delegation->type;
 			spin_unlock(&delegation->lock);
 			put_rpccred(oldcred);
 			rcu_read_unlock();
@@ -275,7 +274,6 @@ nfs_detach_delegation_locked(struct nfs_inode *nfsi,
 	set_bit(NFS_DELEGATION_RETURNING, &delegation->flags);
 	list_del_rcu(&delegation->super_list);
 	delegation->inode = NULL;
-	nfsi->delegation_state = 0;
 	rcu_assign_pointer(nfsi->delegation, NULL);
 	spin_unlock(&delegation->lock);
 	return delegation;
@@ -355,7 +353,6 @@ int nfs_inode_set_delegation(struct inode *inode, struct rpc_cred *cred, struct
 					&delegation->stateid)) {
 			nfs_update_inplace_delegation(old_delegation,
 					delegation);
-			nfsi->delegation_state = old_delegation->type;
 			goto out;
 		}
 		/*
@@ -379,7 +376,6 @@ int nfs_inode_set_delegation(struct inode *inode, struct rpc_cred *cred, struct
 			goto out;
 	}
 	list_add_rcu(&delegation->super_list, &server->delegations);
-	nfsi->delegation_state = delegation->type;
 	rcu_assign_pointer(nfsi->delegation, delegation);
 	delegation = NULL;
 

fs/nfs/direct.c

@@ -283,7 +283,7 @@ static void nfs_direct_release_pages(struct page **pages, unsigned int npages)
 void nfs_init_cinfo_from_dreq(struct nfs_commit_info *cinfo,
 			      struct nfs_direct_req *dreq)
 {
-	cinfo->lock = &dreq->lock;
+	cinfo->lock = &dreq->inode->i_lock;
 	cinfo->mds = &dreq->mds_cinfo;
 	cinfo->ds = &dreq->ds_cinfo;
 	cinfo->dreq = dreq;

fs/nfs/filelayout/filelayout.c

@@ -960,52 +960,19 @@ filelayout_mark_request_commit(struct nfs_page *req,
 {
 	struct nfs4_filelayout_segment *fl = FILELAYOUT_LSEG(lseg);
 	u32 i, j;
-	struct list_head *list;
-	struct pnfs_commit_bucket *buckets;
 
 	if (fl->commit_through_mds) {
-		list = &cinfo->mds->list;
+		nfs_request_add_commit_list(req, &cinfo->mds->list, cinfo);
-		spin_lock(cinfo->lock);
+	} else {
-		goto mds_commit;
+		/* Note that we are calling nfs4_fl_calc_j_index on each page
-	}
+		 * that ends up being committed to a data server.  An attractive
-
+		 * alternative is to add a field to nfs_write_data and nfs_page
-	/* Note that we are calling nfs4_fl_calc_j_index on each page
+		 * to store the value calculated in filelayout_write_pagelist
-	 * that ends up being committed to a data server.  An attractive
+		 * and just use that here.
-	 * alternative is to add a field to nfs_write_data and nfs_page
-	 * to store the value calculated in filelayout_write_pagelist
-	 * and just use that here.
-	 */
-	j = nfs4_fl_calc_j_index(lseg, req_offset(req));
-	i = select_bucket_index(fl, j);
-	spin_lock(cinfo->lock);
-	buckets = cinfo->ds->buckets;
-	list = &buckets[i].written;
-	if (list_empty(list)) {
-		/* Non-empty buckets hold a reference on the lseg.  That ref
-		 * is normally transferred to the COMMIT call and released
-		 * there.  It could also be released if the last req is pulled
-		 * off due to a rewrite, in which case it will be done in
-		 * pnfs_generic_clear_request_commit
 		 */
-		buckets[i].wlseg = pnfs_get_lseg(lseg);
+		j = nfs4_fl_calc_j_index(lseg, req_offset(req));
-	}
+		i = select_bucket_index(fl, j);
-	set_bit(PG_COMMIT_TO_DS, &req->wb_flags);
+		pnfs_layout_mark_request_commit(req, lseg, cinfo, i);
-	cinfo->ds->nwritten++;
-
-mds_commit:
-	/* nfs_request_add_commit_list(). We need to add req to list without
-	 * dropping cinfo lock.
-	 */
-	set_bit(PG_CLEAN, &(req)->wb_flags);
-	nfs_list_add_request(req, list);
-	cinfo->mds->ncommit++;
-	spin_unlock(cinfo->lock);
-	if (!cinfo->dreq) {
-		inc_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
-		inc_bdi_stat(inode_to_bdi(page_file_mapping(req->wb_page)->host),
-			     BDI_RECLAIMABLE);
-		__mark_inode_dirty(req->wb_context->dentry->d_inode,
-				   I_DIRTY_DATASYNC);
 	}
 }
 

fs/nfs/flexfilelayout/flexfilelayout.c

@@ -1332,47 +1332,6 @@ ff_layout_write_pagelist(struct nfs_pgio_header *hdr, int sync)
 	return PNFS_ATTEMPTED;
 }
 
-static void
-ff_layout_mark_request_commit(struct nfs_page *req,
-			      struct pnfs_layout_segment *lseg,
-			      struct nfs_commit_info *cinfo,
-			      u32 ds_commit_idx)
-{
-	struct list_head *list;
-	struct pnfs_commit_bucket *buckets;
-
-	spin_lock(cinfo->lock);
-	buckets = cinfo->ds->buckets;
-	list = &buckets[ds_commit_idx].written;
-	if (list_empty(list)) {
-		/* Non-empty buckets hold a reference on the lseg.  That ref
-		 * is normally transferred to the COMMIT call and released
-		 * there.  It could also be released if the last req is pulled
-		 * off due to a rewrite, in which case it will be done in
-		 * pnfs_common_clear_request_commit
-		 */
-		WARN_ON_ONCE(buckets[ds_commit_idx].wlseg != NULL);
-		buckets[ds_commit_idx].wlseg = pnfs_get_lseg(lseg);
-	}
-	set_bit(PG_COMMIT_TO_DS, &req->wb_flags);
-	cinfo->ds->nwritten++;
-
-	/* nfs_request_add_commit_list(). We need to add req to list without
-	 * dropping cinfo lock.
-	 */
-	set_bit(PG_CLEAN, &(req)->wb_flags);
-	nfs_list_add_request(req, list);
-	cinfo->mds->ncommit++;
-	spin_unlock(cinfo->lock);
-	if (!cinfo->dreq) {
-		inc_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
-		inc_bdi_stat(inode_to_bdi(page_file_mapping(req->wb_page)->host),
-			     BDI_RECLAIMABLE);
-		__mark_inode_dirty(req->wb_context->dentry->d_inode,
-				   I_DIRTY_DATASYNC);
-	}
-}
-
 static u32 calc_ds_index_from_commit(struct pnfs_layout_segment *lseg, u32 i)
 {
 	return i;
@@ -1540,7 +1499,7 @@ static struct pnfs_layoutdriver_type flexfilelayout_type = {
 	.pg_write_ops		= &ff_layout_pg_write_ops,
 	.get_ds_info		= ff_layout_get_ds_info,
 	.free_deviceid_node	= ff_layout_free_deveiceid_node,
-	.mark_request_commit	= ff_layout_mark_request_commit,
+	.mark_request_commit	= pnfs_layout_mark_request_commit,
 	.clear_request_commit	= pnfs_generic_clear_request_commit,
 	.scan_commit_lists	= pnfs_generic_scan_commit_lists,
 	.recover_commit_reqs	= pnfs_generic_recover_commit_reqs,

fs/nfs/inode.c

@@ -1775,7 +1775,6 @@ static inline void nfs4_init_once(struct nfs_inode *nfsi)
 #if IS_ENABLED(CONFIG_NFS_V4)
 	INIT_LIST_HEAD(&nfsi->open_states);
 	nfsi->delegation = NULL;
-	nfsi->delegation_state = 0;
 	init_rwsem(&nfsi->rwsem);
 	nfsi->layout = NULL;
 #endif

fs/nfs/internal.h

@@ -597,6 +597,19 @@ void nfs_super_set_maxbytes(struct super_block *sb, __u64 maxfilesize)
 		sb->s_maxbytes = MAX_LFS_FILESIZE;
 }
 
+/*
+ * Record the page as unstable and mark its inode as dirty.
+ */
+static inline
+void nfs_mark_page_unstable(struct page *page)
+{
+	struct inode *inode = page_file_mapping(page)->host;
+
+	inc_zone_page_state(page, NR_UNSTABLE_NFS);
+	inc_bdi_stat(inode_to_bdi(inode), BDI_RECLAIMABLE);
+	 __mark_inode_dirty(inode, I_DIRTY_DATASYNC);
+}
+
 /*
  * Determine the number of bytes of data the page contains
  */

fs/nfs/nfs4proc.c

@@ -6648,47 +6648,47 @@ nfs41_same_server_scope(struct nfs41_server_scope *a,
 int nfs4_proc_bind_conn_to_session(struct nfs_client *clp, struct rpc_cred *cred)
 {
 	int status;
+	struct nfs41_bind_conn_to_session_args args = {
+		.client = clp,
+		.dir = NFS4_CDFC4_FORE_OR_BOTH,
+	};
 	struct nfs41_bind_conn_to_session_res res;
 	struct rpc_message msg = {
 		.rpc_proc =
 			&nfs4_procedures[NFSPROC4_CLNT_BIND_CONN_TO_SESSION],
-		.rpc_argp = clp,
+		.rpc_argp = &args,
 		.rpc_resp = &res,
 		.rpc_cred = cred,
 	};
 
 	dprintk("--> %s\n", __func__);
 
-	res.session = kzalloc(sizeof(struct nfs4_session), GFP_NOFS);
+	nfs4_copy_sessionid(&args.sessionid, &clp->cl_session->sess_id);
-	if (unlikely(res.session == NULL)) {
+	if (!(clp->cl_session->flags & SESSION4_BACK_CHAN))
-		status = -ENOMEM;
+		args.dir = NFS4_CDFC4_FORE;
-		goto out;
-	}
 
 	status = rpc_call_sync(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
 	trace_nfs4_bind_conn_to_session(clp, status);
 	if (status == 0) {
-		if (memcmp(res.session->sess_id.data,
+		if (memcmp(res.sessionid.data,
 		    clp->cl_session->sess_id.data, NFS4_MAX_SESSIONID_LEN)) {
 			dprintk("NFS: %s: Session ID mismatch\n", __func__);
 			status = -EIO;
-			goto out_session;
+			goto out;
 		}
-		if (res.dir != NFS4_CDFS4_BOTH) {
+		if ((res.dir & args.dir) != res.dir || res.dir == 0) {
 			dprintk("NFS: %s: Unexpected direction from server\n",
 				__func__);
 			status = -EIO;
-			goto out_session;
+			goto out;
 		}
-		if (res.use_conn_in_rdma_mode) {
+		if (res.use_conn_in_rdma_mode != args.use_conn_in_rdma_mode) {
 			dprintk("NFS: %s: Server returned RDMA mode = true\n",
 				__func__);
 			status = -EIO;
-			goto out_session;
+			goto out;
 		}
 	}
-out_session:
-	kfree(res.session);
 out:
 	dprintk("<-- %s status= %d\n", __func__, status);
 	return status;
@@ -7166,10 +7166,11 @@ static void nfs4_init_channel_attrs(struct nfs41_create_session_args *args)
 		args->bc_attrs.max_reqs);
 }
 
-static int nfs4_verify_fore_channel_attrs(struct nfs41_create_session_args *args, struct nfs4_session *session)
+static int nfs4_verify_fore_channel_attrs(struct nfs41_create_session_args *args,
+		struct nfs41_create_session_res *res)
 {
 	struct nfs4_channel_attrs *sent = &args->fc_attrs;
-	struct nfs4_channel_attrs *rcvd = &session->fc_attrs;
+	struct nfs4_channel_attrs *rcvd = &res->fc_attrs;
 
 	if (rcvd->max_resp_sz > sent->max_resp_sz)
 		return -EINVAL;
@@ -7188,11 +7189,14 @@ static int nfs4_verify_fore_channel_attrs(struct nfs41_create_session_args *args
 	return 0;
 }
 
-static int nfs4_verify_back_channel_attrs(struct nfs41_create_session_args *args, struct nfs4_session *session)
+static int nfs4_verify_back_channel_attrs(struct nfs41_create_session_args *args,
+		struct nfs41_create_session_res *res)
 {
 	struct nfs4_channel_attrs *sent = &args->bc_attrs;
-	struct nfs4_channel_attrs *rcvd = &session->bc_attrs;
+	struct nfs4_channel_attrs *rcvd = &res->bc_attrs;
 
+	if (!(res->flags & SESSION4_BACK_CHAN))
+		goto out;
 	if (rcvd->max_rqst_sz > sent->max_rqst_sz)
 		return -EINVAL;
 	if (rcvd->max_resp_sz < sent->max_resp_sz)
@@ -7204,18 +7208,30 @@ static int nfs4_verify_back_channel_attrs(struct nfs41_create_session_args *args
 		return -EINVAL;
 	if (rcvd->max_reqs != sent->max_reqs)
 		return -EINVAL;
+out:
 	return 0;
 }
 
 static int nfs4_verify_channel_attrs(struct nfs41_create_session_args *args,
-				     struct nfs4_session *session)
+				     struct nfs41_create_session_res *res)
 {
 	int ret;
 
-	ret = nfs4_verify_fore_channel_attrs(args, session);
+	ret = nfs4_verify_fore_channel_attrs(args, res);
 	if (ret)
 		return ret;
-	return nfs4_verify_back_channel_attrs(args, session);
+	return nfs4_verify_back_channel_attrs(args, res);
+}
+
+static void nfs4_update_session(struct nfs4_session *session,
+		struct nfs41_create_session_res *res)
+{
+	nfs4_copy_sessionid(&session->sess_id, &res->sessionid);
+	session->flags = res->flags;
+	memcpy(&session->fc_attrs, &res->fc_attrs, sizeof(session->fc_attrs));
+	if (res->flags & SESSION4_BACK_CHAN)
+		memcpy(&session->bc_attrs, &res->bc_attrs,
+				sizeof(session->bc_attrs));
 }
 
 static int _nfs4_proc_create_session(struct nfs_client *clp,
@@ -7224,11 +7240,12 @@ static int _nfs4_proc_create_session(struct nfs_client *clp,
 	struct nfs4_session *session = clp->cl_session;
 	struct nfs41_create_session_args args = {
 		.client = clp,
+		.clientid = clp->cl_clientid,
+		.seqid = clp->cl_seqid,
 		.cb_program = NFS4_CALLBACK,
 	};
-	struct nfs41_create_session_res res = {
+	struct nfs41_create_session_res res;
-		.client = clp,
+
-	};
 	struct rpc_message msg = {
 		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CREATE_SESSION],
 		.rpc_argp = &args,
@@ -7245,11 +7262,15 @@ static int _nfs4_proc_create_session(struct nfs_client *clp,
 
 	if (!status) {
 		/* Verify the session's negotiated channel_attrs values */
-		status = nfs4_verify_channel_attrs(&args, session);
+		status = nfs4_verify_channel_attrs(&args, &res);
 		/* Increment the clientid slot sequence id */
-		clp->cl_seqid++;
+		if (clp->cl_seqid == res.seqid)
+			clp->cl_seqid++;
+		if (status)
+			goto out;
+		nfs4_update_session(session, &res);
 	}
-
+out:
 	return status;
 }
 

fs/nfs/nfs4session.c

@@ -450,7 +450,7 @@ int nfs4_setup_session_slot_tables(struct nfs4_session *ses)
 	tbl = &ses->fc_slot_table;
 	tbl->session = ses;
 	status = nfs4_realloc_slot_table(tbl, ses->fc_attrs.max_reqs, 1);
-	if (status) /* -ENOMEM */
+	if (status || !(ses->flags & SESSION4_BACK_CHAN)) /* -ENOMEM */
 		return status;
 	/* Back channel */
 	tbl = &ses->bc_slot_table;

fs/nfs/nfs4session.h

@@ -118,6 +118,12 @@ static inline int nfs4_has_persistent_session(const struct nfs_client *clp)
 	return 0;
 }
 
+static inline void nfs4_copy_sessionid(struct nfs4_sessionid *dst,
+		const struct nfs4_sessionid *src)
+{
+	memcpy(dst->data, src->data, NFS4_MAX_SESSIONID_LEN);
+}
+
 #ifdef CONFIG_CRC32
 /*
  * nfs_session_id_hash - calculate the crc32 hash for the session id

fs/nfs/nfs4xdr.c

@@ -1715,17 +1715,17 @@ static void encode_secinfo(struct xdr_stream *xdr, const struct qstr *name, stru
 #if defined(CONFIG_NFS_V4_1)
 /* NFSv4.1 operations */
 static void encode_bind_conn_to_session(struct xdr_stream *xdr,
-				   struct nfs4_session *session,
+				   struct nfs41_bind_conn_to_session_args *args,
 				   struct compound_hdr *hdr)
 {
 	__be32 *p;
 
 	encode_op_hdr(xdr, OP_BIND_CONN_TO_SESSION,
 		decode_bind_conn_to_session_maxsz, hdr);
-	encode_opaque_fixed(xdr, session->sess_id.data, NFS4_MAX_SESSIONID_LEN);
+	encode_opaque_fixed(xdr, args->sessionid.data, NFS4_MAX_SESSIONID_LEN);
 	p = xdr_reserve_space(xdr, 8);
-	*p++ = cpu_to_be32(NFS4_CDFC4_BACK_OR_BOTH);
+	*p++ = cpu_to_be32(args->dir);
-	*p = 0;	/* use_conn_in_rdma_mode = False */
+	*p = (args->use_conn_in_rdma_mode) ? cpu_to_be32(1) : cpu_to_be32(0);
 }
 
 static void encode_op_map(struct xdr_stream *xdr, struct nfs4_op_map *op_map)
@@ -1806,8 +1806,8 @@ static void encode_create_session(struct xdr_stream *xdr,
 
 	encode_op_hdr(xdr, OP_CREATE_SESSION, decode_create_session_maxsz, hdr);
 	p = reserve_space(xdr, 16 + 2*28 + 20 + clnt->cl_nodelen + 12);
-	p = xdr_encode_hyper(p, clp->cl_clientid);
+	p = xdr_encode_hyper(p, args->clientid);
-	*p++ = cpu_to_be32(clp->cl_seqid);			/*Sequence id */
+	*p++ = cpu_to_be32(args->seqid);			/*Sequence id */
 	*p++ = cpu_to_be32(args->flags);			/*flags */
 
 	/* Fore Channel */
@@ -2734,14 +2734,14 @@ static void nfs4_xdr_enc_fsid_present(struct rpc_rqst *req,
  */
 static void nfs4_xdr_enc_bind_conn_to_session(struct rpc_rqst *req,
 				struct xdr_stream *xdr,
-				struct nfs_client *clp)
+				struct nfs41_bind_conn_to_session_args *args)
 {
 	struct compound_hdr hdr = {
-		.minorversion = clp->cl_mvops->minor_version,
+		.minorversion = args->client->cl_mvops->minor_version,
 	};
 
 	encode_compound_hdr(xdr, req, &hdr);
-	encode_bind_conn_to_session(xdr, clp->cl_session, &hdr);
+	encode_bind_conn_to_session(xdr, args, &hdr);
 	encode_nops(&hdr);
 }
 
@@ -5613,7 +5613,7 @@ static int decode_bind_conn_to_session(struct xdr_stream *xdr,
 
 	status = decode_op_hdr(xdr, OP_BIND_CONN_TO_SESSION);
 	if (!status)
-		status = decode_sessionid(xdr, &res->session->sess_id);
+		status = decode_sessionid(xdr, &res->sessionid);
 	if (unlikely(status))
 		return status;
 
@@ -5641,12 +5641,10 @@ static int decode_create_session(struct xdr_stream *xdr,
 {
 	__be32 *p;
 	int status;
-	struct nfs_client *clp = res->client;
-	struct nfs4_session *session = clp->cl_session;
 
 	status = decode_op_hdr(xdr, OP_CREATE_SESSION);
 	if (!status)
-		status = decode_sessionid(xdr, &session->sess_id);
+		status = decode_sessionid(xdr, &res->sessionid);
 	if (unlikely(status))
 		return status;
 
@@ -5654,13 +5652,13 @@ static int decode_create_session(struct xdr_stream *xdr,
 	p = xdr_inline_decode(xdr, 8);
 	if (unlikely(!p))
 		goto out_overflow;
-	clp->cl_seqid = be32_to_cpup(p++);
+	res->seqid = be32_to_cpup(p++);
-	session->flags = be32_to_cpup(p);
+	res->flags = be32_to_cpup(p);
 
 	/* Channel attributes */
-	status = decode_chan_attrs(xdr, &session->fc_attrs);
+	status = decode_chan_attrs(xdr, &res->fc_attrs);
 	if (!status)
-		status = decode_chan_attrs(xdr, &session->bc_attrs);
+		status = decode_chan_attrs(xdr, &res->bc_attrs);
 	return status;
 out_overflow:
 	print_overflow_msg(__func__, xdr);

fs/nfs/pnfs.h

@@ -344,6 +344,10 @@ void nfs4_pnfs_ds_connect(struct nfs_server *mds_srv, struct nfs4_pnfs_ds *ds,
 struct nfs4_pnfs_ds_addr *nfs4_decode_mp_ds_addr(struct net *net,
 						 struct xdr_stream *xdr,
 						 gfp_t gfp_flags);
+void pnfs_layout_mark_request_commit(struct nfs_page *req,
+				     struct pnfs_layout_segment *lseg,
+				     struct nfs_commit_info *cinfo,
+				     u32 ds_commit_idx);
 
 static inline bool nfs_have_layout(struct inode *inode)
 {

fs/nfs/pnfs_nfs.c

@@ -838,3 +838,33 @@ out_err:
 	return NULL;
 }
 EXPORT_SYMBOL_GPL(nfs4_decode_mp_ds_addr);
+
+void
+pnfs_layout_mark_request_commit(struct nfs_page *req,
+				struct pnfs_layout_segment *lseg,
+				struct nfs_commit_info *cinfo,
+				u32 ds_commit_idx)
+{
+	struct list_head *list;
+	struct pnfs_commit_bucket *buckets;
+
+	spin_lock(cinfo->lock);
+	buckets = cinfo->ds->buckets;
+	list = &buckets[ds_commit_idx].written;
+	if (list_empty(list)) {
+		/* Non-empty buckets hold a reference on the lseg.  That ref
+		 * is normally transferred to the COMMIT call and released
+		 * there.  It could also be released if the last req is pulled
+		 * off due to a rewrite, in which case it will be done in
+		 * pnfs_common_clear_request_commit
+		 */
+		WARN_ON_ONCE(buckets[ds_commit_idx].wlseg != NULL);
+		buckets[ds_commit_idx].wlseg = pnfs_get_lseg(lseg);
+	}
+	set_bit(PG_COMMIT_TO_DS, &req->wb_flags);
+	cinfo->ds->nwritten++;
+	spin_unlock(cinfo->lock);
+
+	nfs_request_add_commit_list(req, list, cinfo);
+}
+EXPORT_SYMBOL_GPL(pnfs_layout_mark_request_commit);

fs/nfs/write.c

@@ -789,13 +789,8 @@ nfs_request_add_commit_list(struct nfs_page *req, struct list_head *dst,
 	nfs_list_add_request(req, dst);
 	cinfo->mds->ncommit++;
 	spin_unlock(cinfo->lock);
-	if (!cinfo->dreq) {
+	if (!cinfo->dreq)
-		inc_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
+		nfs_mark_page_unstable(req->wb_page);
-		inc_bdi_stat(inode_to_bdi(page_file_mapping(req->wb_page)->host),
-			     BDI_RECLAIMABLE);
-		__mark_inode_dirty(req->wb_context->dentry->d_inode,
-				   I_DIRTY_DATASYNC);
-	}
 }
 EXPORT_SYMBOL_GPL(nfs_request_add_commit_list);
 
@@ -1605,11 +1600,8 @@ void nfs_retry_commit(struct list_head *page_list,
 		req = nfs_list_entry(page_list->next);
 		nfs_list_remove_request(req);
 		nfs_mark_request_commit(req, lseg, cinfo, ds_commit_idx);
-		if (!cinfo->dreq) {
+		if (!cinfo->dreq)
-			dec_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
+			nfs_clear_page_commit(req->wb_page);
-			dec_bdi_stat(inode_to_bdi(page_file_mapping(req->wb_page)->host),
-				     BDI_RECLAIMABLE);
-		}
 		nfs_unlock_and_release_request(req);
 	}
 }

include/linux/nfs_fs.h

@@ -180,7 +180,6 @@ struct nfs_inode {
         /* NFSv4 state */
 	struct list_head	open_states;
 	struct nfs_delegation __rcu *delegation;
-	fmode_t			 delegation_state;
 	struct rw_semaphore	rwsem;
 
 	/* pNFS layout information */

include/linux/nfs_xdr.h

@@ -1167,8 +1167,15 @@ struct nfs41_impl_id {
 	struct nfstime4			date;
 };
 
+struct nfs41_bind_conn_to_session_args {
+	struct nfs_client		*client;
+	struct nfs4_sessionid		sessionid;
+	u32				dir;
+	bool				use_conn_in_rdma_mode;
+};
+
 struct nfs41_bind_conn_to_session_res {
-	struct nfs4_session		*session;
+	struct nfs4_sessionid		sessionid;
 	u32				dir;
 	bool				use_conn_in_rdma_mode;
 };
@@ -1185,6 +1192,8 @@ struct nfs41_exchange_id_res {
 
 struct nfs41_create_session_args {
 	struct nfs_client	       *client;
+	u64				clientid;
+	uint32_t			seqid;
 	uint32_t			flags;
 	uint32_t			cb_program;
 	struct nfs4_channel_attrs	fc_attrs;	/* Fore Channel */
@@ -1192,7 +1201,11 @@ struct nfs41_create_session_args {
 };
 
 struct nfs41_create_session_res {
-	struct nfs_client	       *client;
+	struct nfs4_sessionid		sessionid;
+	uint32_t			seqid;
+	uint32_t			flags;
+	struct nfs4_channel_attrs	fc_attrs;	/* Fore Channel */
+	struct nfs4_channel_attrs	bc_attrs;	/* Back Channel */
 };
 
 struct nfs41_reclaim_complete_args {
@@ -1351,7 +1364,7 @@ struct nfs_commit_completion_ops {
 };
 
 struct nfs_commit_info {
-	spinlock_t			*lock;
+	spinlock_t			*lock;	/* inode->i_lock */
 	struct nfs_mds_commit_info	*mds;
 	struct pnfs_ds_commit_info	*ds;
 	struct nfs_direct_req		*dreq;	/* O_DIRECT request */

include/linux/sunrpc/metrics.h

@@ -89,8 +89,11 @@ void			rpc_free_iostats(struct rpc_iostats *);
 static inline struct rpc_iostats *rpc_alloc_iostats(struct rpc_clnt *clnt) { return NULL; }
 static inline void rpc_count_iostats(const struct rpc_task *task,
 				     struct rpc_iostats *stats) {}
-static inline void rpc_count_iostats_metrics(const struct rpc_task *,
+static inline void rpc_count_iostats_metrics(const struct rpc_task *task,
-					     struct rpc_iostats *) {}
+					     struct rpc_iostats *stats)
+{
+}
+
 static inline void rpc_print_iostats(struct seq_file *seq, struct rpc_clnt *clnt) {}
 static inline void rpc_free_iostats(struct rpc_iostats *stats) {}
 

net/sunrpc/backchannel_rqst.c

@@ -309,12 +309,15 @@ void xprt_complete_bc_request(struct rpc_rqst *req, uint32_t copied)
 	struct rpc_xprt *xprt = req->rq_xprt;
 	struct svc_serv *bc_serv = xprt->bc_serv;
 
+	spin_lock(&xprt->bc_pa_lock);
+	list_del(&req->rq_bc_pa_list);
+	spin_unlock(&xprt->bc_pa_lock);
+
 	req->rq_private_buf.len = copied;
 	set_bit(RPC_BC_PA_IN_USE, &req->rq_bc_pa_state);
 
 	dprintk("RPC:       add callback request to list\n");
 	spin_lock(&bc_serv->sv_cb_lock);
-	list_del(&req->rq_bc_pa_list);
 	list_add(&req->rq_bc_list, &bc_serv->sv_cb_list);
 	wake_up(&bc_serv->sv_cb_waitq);
 	spin_unlock(&bc_serv->sv_cb_lock);