Merge tag 'nfs-for-3.17-1' of git://git.linux-nfs.org/projects/trondmy/linux-nfs

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

   - stable fix for a bug in nfs3_list_one_acl()
   - speed up NFS path walks by supporting LOOKUP_RCU
   - more read/write code cleanups
   - pNFS fixes for layout return on close
   - fixes for the RCU handling in the rpcsec_gss code
   - more NFS/RDMA fixes"

* tag 'nfs-for-3.17-1' of git://git.linux-nfs.org/projects/trondmy/linux-nfs: (79 commits)
  nfs: reject changes to resvport and sharecache during remount
  NFS: Avoid infinite loop when RELEASE_LOCKOWNER getting expired error
  SUNRPC: remove all refcounting of groupinfo from rpcauth_lookupcred
  NFS: fix two problems in lookup_revalidate in RCU-walk
  NFS: allow lockless access to access_cache
  NFS: teach nfs_lookup_verify_inode to handle LOOKUP_RCU
  NFS: teach nfs_neg_need_reval to understand LOOKUP_RCU
  NFS: support RCU_WALK in nfs_permission()
  sunrpc/auth: allow lockless (rcu) lookup of credential cache.
  NFS: prepare for RCU-walk support but pushing tests later in code.
  NFS: nfs4_lookup_revalidate: only evaluate parent if it will be used.
  NFS: add checks for returned value of try_module_get()
  nfs: clear_request_commit while holding i_lock
  pnfs: add pnfs_put_lseg_async
  pnfs: find swapped pages on pnfs commit lists too
  nfs: fix comment and add warn_on for PG_INODE_REF
  nfs: check wait_on_bit_lock err in page_group_lock
  sunrpc: remove "ec" argument from encrypt_v2 operation
  sunrpc: clean up sparse endianness warnings in gss_krb5_wrap.c
  sunrpc: clean up sparse endianness warnings in gss_krb5_seal.c
  ...

fs/nfs/blocklayout/blocklayout.c

@@ -210,8 +210,7 @@ static void bl_end_io_read(struct bio *bio, int err)
 			SetPageUptodate(bvec->bv_page);
 
 	if (err) {
-		struct nfs_pgio_data *rdata = par->data;
-		struct nfs_pgio_header *header = rdata->header;
+		struct nfs_pgio_header *header = par->data;
 
 		if (!header->pnfs_error)
 			header->pnfs_error = -EIO;
@@ -224,43 +223,44 @@ static void bl_end_io_read(struct bio *bio, int err)
 static void bl_read_cleanup(struct work_struct *work)
 {
 	struct rpc_task *task;
-	struct nfs_pgio_data *rdata;
+	struct nfs_pgio_header *hdr;
 	dprintk("%s enter\n", __func__);
 	task = container_of(work, struct rpc_task, u.tk_work);
-	rdata = container_of(task, struct nfs_pgio_data, task);
-	pnfs_ld_read_done(rdata);
+	hdr = container_of(task, struct nfs_pgio_header, task);
+	pnfs_ld_read_done(hdr);
 }
 
 static void
 bl_end_par_io_read(void *data, int unused)
 {
-	struct nfs_pgio_data *rdata = data;
+	struct nfs_pgio_header *hdr = data;
 
-	rdata->task.tk_status = rdata->header->pnfs_error;
-	INIT_WORK(&rdata->task.u.tk_work, bl_read_cleanup);
-	schedule_work(&rdata->task.u.tk_work);
+	hdr->task.tk_status = hdr->pnfs_error;
+	INIT_WORK(&hdr->task.u.tk_work, bl_read_cleanup);
+	schedule_work(&hdr->task.u.tk_work);
 }
 
 static enum pnfs_try_status
-bl_read_pagelist(struct nfs_pgio_data *rdata)
+bl_read_pagelist(struct nfs_pgio_header *hdr)
 {
-	struct nfs_pgio_header *header = rdata->header;
+	struct nfs_pgio_header *header = hdr;
 	int i, hole;
 	struct bio *bio = NULL;
 	struct pnfs_block_extent *be = NULL, *cow_read = NULL;
 	sector_t isect, extent_length = 0;
 	struct parallel_io *par;
-	loff_t f_offset = rdata->args.offset;
-	size_t bytes_left = rdata->args.count;
+	loff_t f_offset = hdr->args.offset;
+	size_t bytes_left = hdr->args.count;
 	unsigned int pg_offset, pg_len;
-	struct page **pages = rdata->args.pages;
-	int pg_index = rdata->args.pgbase >> PAGE_CACHE_SHIFT;
+	struct page **pages = hdr->args.pages;
+	int pg_index = hdr->args.pgbase >> PAGE_CACHE_SHIFT;
 	const bool is_dio = (header->dreq != NULL);
 
 	dprintk("%s enter nr_pages %u offset %lld count %u\n", __func__,
-	       rdata->pages.npages, f_offset, (unsigned int)rdata->args.count);
+		hdr->page_array.npages, f_offset,
+		(unsigned int)hdr->args.count);
 
-	par = alloc_parallel(rdata);
+	par = alloc_parallel(hdr);
 	if (!par)
 		goto use_mds;
 	par->pnfs_callback = bl_end_par_io_read;
@@ -268,7 +268,7 @@ bl_read_pagelist(struct nfs_pgio_data *rdata)
 
 	isect = (sector_t) (f_offset >> SECTOR_SHIFT);
 	/* Code assumes extents are page-aligned */
-	for (i = pg_index; i < rdata->pages.npages; i++) {
+	for (i = pg_index; i < hdr->page_array.npages; i++) {
 		if (!extent_length) {
 			/* We've used up the previous extent */
 			bl_put_extent(be);
@@ -317,7 +317,8 @@ bl_read_pagelist(struct nfs_pgio_data *rdata)
 			struct pnfs_block_extent *be_read;
 
 			be_read = (hole && cow_read) ? cow_read : be;
-			bio = do_add_page_to_bio(bio, rdata->pages.npages - i,
+			bio = do_add_page_to_bio(bio,
+						 hdr->page_array.npages - i,
 						 READ,
 						 isect, pages[i], be_read,
 						 bl_end_io_read, par,
@@ -332,10 +333,10 @@ bl_read_pagelist(struct nfs_pgio_data *rdata)
 		extent_length -= PAGE_CACHE_SECTORS;
 	}
 	if ((isect << SECTOR_SHIFT) >= header->inode->i_size) {
-		rdata->res.eof = 1;
-		rdata->res.count = header->inode->i_size - rdata->args.offset;
+		hdr->res.eof = 1;
+		hdr->res.count = header->inode->i_size - hdr->args.offset;
 	} else {
-		rdata->res.count = (isect << SECTOR_SHIFT) - rdata->args.offset;
+		hdr->res.count = (isect << SECTOR_SHIFT) - hdr->args.offset;
 	}
 out:
 	bl_put_extent(be);
@@ -390,8 +391,7 @@ static void bl_end_io_write_zero(struct bio *bio, int err)
 	}
 
 	if (unlikely(err)) {
-		struct nfs_pgio_data *data = par->data;
-		struct nfs_pgio_header *header = data->header;
+		struct nfs_pgio_header *header = par->data;
 
 		if (!header->pnfs_error)
 			header->pnfs_error = -EIO;
@@ -405,8 +405,7 @@ static void bl_end_io_write(struct bio *bio, int err)
 {
 	struct parallel_io *par = bio->bi_private;
 	const int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
-	struct nfs_pgio_data *data = par->data;
-	struct nfs_pgio_header *header = data->header;
+	struct nfs_pgio_header *header = par->data;
 
 	if (!uptodate) {
 		if (!header->pnfs_error)
@@ -423,32 +422,32 @@ static void bl_end_io_write(struct bio *bio, int err)
 static void bl_write_cleanup(struct work_struct *work)
 {
 	struct rpc_task *task;
-	struct nfs_pgio_data *wdata;
+	struct nfs_pgio_header *hdr;
 	dprintk("%s enter\n", __func__);
 	task = container_of(work, struct rpc_task, u.tk_work);
-	wdata = container_of(task, struct nfs_pgio_data, task);
-	if (likely(!wdata->header->pnfs_error)) {
+	hdr = container_of(task, struct nfs_pgio_header, task);
+	if (likely(!hdr->pnfs_error)) {
 		/* Marks for LAYOUTCOMMIT */
-		mark_extents_written(BLK_LSEG2EXT(wdata->header->lseg),
-				     wdata->args.offset, wdata->args.count);
+		mark_extents_written(BLK_LSEG2EXT(hdr->lseg),
+				     hdr->args.offset, hdr->args.count);
 	}
-	pnfs_ld_write_done(wdata);
+	pnfs_ld_write_done(hdr);
 }
 
 /* Called when last of bios associated with a bl_write_pagelist call finishes */
 static void bl_end_par_io_write(void *data, int num_se)
 {
-	struct nfs_pgio_data *wdata = data;
+	struct nfs_pgio_header *hdr = data;
 
-	if (unlikely(wdata->header->pnfs_error)) {
-		bl_free_short_extents(&BLK_LSEG2EXT(wdata->header->lseg)->bl_inval,
+	if (unlikely(hdr->pnfs_error)) {
+		bl_free_short_extents(&BLK_LSEG2EXT(hdr->lseg)->bl_inval,
 					num_se);
 	}
 
-	wdata->task.tk_status = wdata->header->pnfs_error;
-	wdata->verf.committed = NFS_FILE_SYNC;
-	INIT_WORK(&wdata->task.u.tk_work, bl_write_cleanup);
-	schedule_work(&wdata->task.u.tk_work);
+	hdr->task.tk_status = hdr->pnfs_error;
+	hdr->verf.committed = NFS_FILE_SYNC;
+	INIT_WORK(&hdr->task.u.tk_work, bl_write_cleanup);
+	schedule_work(&hdr->task.u.tk_work);
 }
 
 /* FIXME STUB - mark intersection of layout and page as bad, so is not
@@ -673,18 +672,17 @@ check_page:
 }
 
 static enum pnfs_try_status
-bl_write_pagelist(struct nfs_pgio_data *wdata, int sync)
+bl_write_pagelist(struct nfs_pgio_header *header, int sync)
 {
-	struct nfs_pgio_header *header = wdata->header;
 	int i, ret, npg_zero, pg_index, last = 0;
 	struct bio *bio = NULL;
 	struct pnfs_block_extent *be = NULL, *cow_read = NULL;
 	sector_t isect, last_isect = 0, extent_length = 0;
 	struct parallel_io *par = NULL;
-	loff_t offset = wdata->args.offset;
-	size_t count = wdata->args.count;
+	loff_t offset = header->args.offset;
+	size_t count = header->args.count;
 	unsigned int pg_offset, pg_len, saved_len;
-	struct page **pages = wdata->args.pages;
+	struct page **pages = header->args.pages;
 	struct page *page;
 	pgoff_t index;
 	u64 temp;
@@ -699,11 +697,11 @@ bl_write_pagelist(struct nfs_pgio_data *wdata, int sync)
 		dprintk("pnfsblock nonblock aligned DIO writes. Resend MDS\n");
 		goto out_mds;
 	}
-	/* At this point, wdata->pages is a (sequential) list of nfs_pages.
+	/* At this point, header->page_aray is a (sequential) list of nfs_pages.
 	 * We want to write each, and if there is an error set pnfs_error
 	 * to have it redone using nfs.
 	 */
-	par = alloc_parallel(wdata);
+	par = alloc_parallel(header);
 	if (!par)
 		goto out_mds;
 	par->pnfs_callback = bl_end_par_io_write;
@@ -790,8 +788,8 @@ next_page:
 	bio = bl_submit_bio(WRITE, bio);
 
 	/* Middle pages */
-	pg_index = wdata->args.pgbase >> PAGE_CACHE_SHIFT;
-	for (i = pg_index; i < wdata->pages.npages; i++) {
+	pg_index = header->args.pgbase >> PAGE_CACHE_SHIFT;
+	for (i = pg_index; i < header->page_array.npages; i++) {
 		if (!extent_length) {
 			/* We've used up the previous extent */
 			bl_put_extent(be);
@@ -862,7 +860,8 @@ next_page:
 		}
 
 
-		bio = do_add_page_to_bio(bio, wdata->pages.npages - i, WRITE,
+		bio = do_add_page_to_bio(bio, header->page_array.npages - i,
+					 WRITE,
 					 isect, pages[i], be,
 					 bl_end_io_write, par,
 					 pg_offset, pg_len);
@@ -890,7 +889,7 @@ next_page:
 	}
 
 write_done:
-	wdata->res.count = wdata->args.count;
+	header->res.count = header->args.count;
 out:
 	bl_put_extent(be);
 	bl_put_extent(cow_read);
@@ -1063,7 +1062,7 @@ nfs4_blk_get_deviceinfo(struct nfs_server *server, const struct nfs_fh *fh,
 		return ERR_PTR(-ENOMEM);
 	}
 
-	pages = kzalloc(max_pages * sizeof(struct page *), GFP_NOFS);
+	pages = kcalloc(max_pages, sizeof(struct page *), GFP_NOFS);
 	if (pages == NULL) {
 		kfree(dev);
 		return ERR_PTR(-ENOMEM);

fs/nfs/callback.c

@@ -428,6 +428,18 @@ check_gss_callback_principal(struct nfs_client *clp, struct svc_rqst *rqstp)
 	if (p == NULL)
 		return 0;
 
+	/*
+	 * Did we get the acceptor from userland during the SETCLIENID
+	 * negotiation?
+	 */
+	if (clp->cl_acceptor)
+		return !strcmp(p, clp->cl_acceptor);
+
+	/*
+	 * Otherwise try to verify it using the cl_hostname. Note that this
+	 * doesn't work if a non-canonical hostname was used in the devname.
+	 */
+
 	/* Expect a GSS_C_NT_HOSTBASED_NAME like "nfs@serverhostname" */
 
 	if (memcmp(p, "nfs@", 4) != 0)

fs/nfs/client.c

@@ -110,8 +110,8 @@ struct nfs_subversion *get_nfs_version(unsigned int version)
 		mutex_unlock(&nfs_version_mutex);
 	}
 
-	if (!IS_ERR(nfs))
-		try_module_get(nfs->owner);
+	if (!IS_ERR(nfs) && !try_module_get(nfs->owner))
+		return ERR_PTR(-EAGAIN);
 	return nfs;
 }
 
@@ -158,7 +158,8 @@ struct nfs_client *nfs_alloc_client(const struct nfs_client_initdata *cl_init)
 		goto error_0;
 
 	clp->cl_nfs_mod = cl_init->nfs_mod;
-	try_module_get(clp->cl_nfs_mod->owner);
+	if (!try_module_get(clp->cl_nfs_mod->owner))
+		goto error_dealloc;
 
 	clp->rpc_ops = clp->cl_nfs_mod->rpc_ops;
 
@@ -190,6 +191,7 @@ struct nfs_client *nfs_alloc_client(const struct nfs_client_initdata *cl_init)
 
 error_cleanup:
 	put_nfs_version(clp->cl_nfs_mod);
+error_dealloc:
 	kfree(clp);
 error_0:
 	return ERR_PTR(err);
@@ -252,6 +254,7 @@ void nfs_free_client(struct nfs_client *clp)
 	put_net(clp->cl_net);
 	put_nfs_version(clp->cl_nfs_mod);
 	kfree(clp->cl_hostname);
+	kfree(clp->cl_acceptor);
 	kfree(clp);
 
 	dprintk("<-- nfs_free_client()\n");
@@ -482,8 +485,13 @@ nfs_get_client(const struct nfs_client_initdata *cl_init,
 	struct nfs_net *nn = net_generic(cl_init->net, nfs_net_id);
 	const struct nfs_rpc_ops *rpc_ops = cl_init->nfs_mod->rpc_ops;
 
+	if (cl_init->hostname == NULL) {
+		WARN_ON(1);
+		return NULL;
+	}
+
 	dprintk("--> nfs_get_client(%s,v%u)\n",
-		cl_init->hostname ?: "", rpc_ops->version);
+		cl_init->hostname, rpc_ops->version);
 
 	/* see if the client already exists */
 	do {
@@ -510,7 +518,7 @@ nfs_get_client(const struct nfs_client_initdata *cl_init,
 	} while (!IS_ERR(new));
 
 	dprintk("<-- nfs_get_client() Failed to find %s (%ld)\n",
-		cl_init->hostname ?: "", PTR_ERR(new));
+		cl_init->hostname, PTR_ERR(new));
 	return new;
 }
 EXPORT_SYMBOL_GPL(nfs_get_client);

fs/nfs/delegation.c

@@ -41,14 +41,8 @@ void nfs_mark_delegation_referenced(struct nfs_delegation *delegation)
 	set_bit(NFS_DELEGATION_REFERENCED, &delegation->flags);
 }
 
-/**
- * nfs_have_delegation - check if inode has a delegation
- * @inode: inode to check
- * @flags: delegation types to check for
- *
- * Returns one if inode has the indicated delegation, otherwise zero.
- */
-int nfs4_have_delegation(struct inode *inode, fmode_t flags)
+static int
+nfs4_do_check_delegation(struct inode *inode, fmode_t flags, bool mark)
 {
 	struct nfs_delegation *delegation;
 	int ret = 0;
@@ -58,12 +52,34 @@ int nfs4_have_delegation(struct inode *inode, fmode_t flags)
 	delegation = rcu_dereference(NFS_I(inode)->delegation);
 	if (delegation != NULL && (delegation->type & flags) == flags &&
 	    !test_bit(NFS_DELEGATION_RETURNING, &delegation->flags)) {
-		nfs_mark_delegation_referenced(delegation);
+		if (mark)
+			nfs_mark_delegation_referenced(delegation);
 		ret = 1;
 	}
 	rcu_read_unlock();
 	return ret;
 }
+/**
+ * nfs_have_delegation - check if inode has a delegation, mark it
+ * NFS_DELEGATION_REFERENCED if there is one.
+ * @inode: inode to check
+ * @flags: delegation types to check for
+ *
+ * Returns one if inode has the indicated delegation, otherwise zero.
+ */
+int nfs4_have_delegation(struct inode *inode, fmode_t flags)
+{
+	return nfs4_do_check_delegation(inode, flags, true);
+}
+
+/*
+ * nfs4_check_delegation - check if inode has a delegation, do not mark
+ * NFS_DELEGATION_REFERENCED if it has one.
+ */
+int nfs4_check_delegation(struct inode *inode, fmode_t flags)
+{
+	return nfs4_do_check_delegation(inode, flags, false);
+}
 
 static int nfs_delegation_claim_locks(struct nfs_open_context *ctx, struct nfs4_state *state, const nfs4_stateid *stateid)
 {

fs/nfs/delegation.h

@@ -59,6 +59,7 @@ bool nfs4_copy_delegation_stateid(nfs4_stateid *dst, struct inode *inode, fmode_
 
 void nfs_mark_delegation_referenced(struct nfs_delegation *delegation);
 int nfs4_have_delegation(struct inode *inode, fmode_t flags);
+int nfs4_check_delegation(struct inode *inode, fmode_t flags);
 
 #endif
 

fs/nfs/dir.c

@@ -988,9 +988,13 @@ EXPORT_SYMBOL_GPL(nfs_force_lookup_revalidate);
  * A check for whether or not the parent directory has changed.
  * In the case it has, we assume that the dentries are untrustworthy
  * and may need to be looked up again.
+ * If rcu_walk prevents us from performing a full check, return 0.
  */
-static int nfs_check_verifier(struct inode *dir, struct dentry *dentry)
+static int nfs_check_verifier(struct inode *dir, struct dentry *dentry,
+			      int rcu_walk)
 {
+	int ret;
+
 	if (IS_ROOT(dentry))
 		return 1;
 	if (NFS_SERVER(dir)->flags & NFS_MOUNT_LOOKUP_CACHE_NONE)
@@ -998,7 +1002,11 @@ static int nfs_check_verifier(struct inode *dir, struct dentry *dentry)
 	if (!nfs_verify_change_attribute(dir, dentry->d_time))
 		return 0;
 	/* Revalidate nfsi->cache_change_attribute before we declare a match */
-	if (nfs_revalidate_inode(NFS_SERVER(dir), dir) < 0)
+	if (rcu_walk)
+		ret = nfs_revalidate_inode_rcu(NFS_SERVER(dir), dir);
+	else
+		ret = nfs_revalidate_inode(NFS_SERVER(dir), dir);
+	if (ret < 0)
 		return 0;
 	if (!nfs_verify_change_attribute(dir, dentry->d_time))
 		return 0;
@@ -1042,6 +1050,8 @@ int nfs_lookup_verify_inode(struct inode *inode, unsigned int flags)
 out:
 	return (inode->i_nlink == 0) ? -ENOENT : 0;
 out_force:
+	if (flags & LOOKUP_RCU)
+		return -ECHILD;
 	ret = __nfs_revalidate_inode(server, inode);
 	if (ret != 0)
 		return ret;
@@ -1054,6 +1064,9 @@ out_force:
  *
  * If parent mtime has changed, we revalidate, else we wait for a
  * period corresponding to the parent's attribute cache timeout value.
+ *
+ * If LOOKUP_RCU prevents us from performing a full check, return 1
+ * suggesting a reval is needed.
  */
 static inline
 int nfs_neg_need_reval(struct inode *dir, struct dentry *dentry,
@@ -1064,7 +1077,7 @@ int nfs_neg_need_reval(struct inode *dir, struct dentry *dentry,
 		return 0;
 	if (NFS_SERVER(dir)->flags & NFS_MOUNT_LOOKUP_CACHE_NONEG)
 		return 1;
-	return !nfs_check_verifier(dir, dentry);
+	return !nfs_check_verifier(dir, dentry, flags & LOOKUP_RCU);
 }
 
 /*
@@ -1088,21 +1101,30 @@ static int nfs_lookup_revalidate(struct dentry *dentry, unsigned int flags)
 	struct nfs4_label *label = NULL;
 	int error;
 
-	if (flags & LOOKUP_RCU)
-		return -ECHILD;
-
-	parent = dget_parent(dentry);
-	dir = parent->d_inode;
+	if (flags & LOOKUP_RCU) {
+		parent = ACCESS_ONCE(dentry->d_parent);
+		dir = ACCESS_ONCE(parent->d_inode);
+		if (!dir)
+			return -ECHILD;
+	} else {
+		parent = dget_parent(dentry);
+		dir = parent->d_inode;
+	}
 	nfs_inc_stats(dir, NFSIOS_DENTRYREVALIDATE);
 	inode = dentry->d_inode;
 
 	if (!inode) {
-		if (nfs_neg_need_reval(dir, dentry, flags))
+		if (nfs_neg_need_reval(dir, dentry, flags)) {
+			if (flags & LOOKUP_RCU)
+				return -ECHILD;
 			goto out_bad;
+		}
 		goto out_valid_noent;
 	}
 
 	if (is_bad_inode(inode)) {
+		if (flags & LOOKUP_RCU)
+			return -ECHILD;
 		dfprintk(LOOKUPCACHE, "%s: %pd2 has dud inode\n",
 				__func__, dentry);
 		goto out_bad;
@@ -1112,12 +1134,20 @@ static int nfs_lookup_revalidate(struct dentry *dentry, unsigned int flags)
 		goto out_set_verifier;
 
 	/* Force a full look up iff the parent directory has changed */
-	if (!nfs_is_exclusive_create(dir, flags) && nfs_check_verifier(dir, dentry)) {
-		if (nfs_lookup_verify_inode(inode, flags))
+	if (!nfs_is_exclusive_create(dir, flags) &&
+	    nfs_check_verifier(dir, dentry, flags & LOOKUP_RCU)) {
+
+		if (nfs_lookup_verify_inode(inode, flags)) {
+			if (flags & LOOKUP_RCU)
+				return -ECHILD;
 			goto out_zap_parent;
+		}
 		goto out_valid;
 	}
 
+	if (flags & LOOKUP_RCU)
+		return -ECHILD;
+
 	if (NFS_STALE(inode))
 		goto out_bad;
 
@@ -1153,13 +1183,18 @@ out_set_verifier:
 	/* Success: notify readdir to use READDIRPLUS */
 	nfs_advise_use_readdirplus(dir);
  out_valid_noent:
-	dput(parent);
+	if (flags & LOOKUP_RCU) {
+		if (parent != ACCESS_ONCE(dentry->d_parent))
+			return -ECHILD;
+	} else
+		dput(parent);
 	dfprintk(LOOKUPCACHE, "NFS: %s(%pd2) is valid\n",
 			__func__, dentry);
 	return 1;
 out_zap_parent:
 	nfs_zap_caches(dir);
  out_bad:
+	WARN_ON(flags & LOOKUP_RCU);
 	nfs_free_fattr(fattr);
 	nfs_free_fhandle(fhandle);
 	nfs4_label_free(label);
@@ -1185,6 +1220,7 @@ out_zap_parent:
 			__func__, dentry);
 	return 0;
 out_error:
+	WARN_ON(flags & LOOKUP_RCU);
 	nfs_free_fattr(fattr);
 	nfs_free_fhandle(fhandle);
 	nfs4_label_free(label);
@@ -1529,14 +1565,9 @@ EXPORT_SYMBOL_GPL(nfs_atomic_open);
 
 static int nfs4_lookup_revalidate(struct dentry *dentry, unsigned int flags)
 {
-	struct dentry *parent = NULL;
 	struct inode *inode;
-	struct inode *dir;
 	int ret = 0;
 
-	if (flags & LOOKUP_RCU)
-		return -ECHILD;
-
 	if (!(flags & LOOKUP_OPEN) || (flags & LOOKUP_DIRECTORY))
 		goto no_open;
 	if (d_mountpoint(dentry))
@@ -1545,34 +1576,47 @@ static int nfs4_lookup_revalidate(struct dentry *dentry, unsigned int flags)
 		goto no_open;
 
 	inode = dentry->d_inode;
-	parent = dget_parent(dentry);
-	dir = parent->d_inode;
 
 	/* We can't create new files in nfs_open_revalidate(), so we
 	 * optimize away revalidation of negative dentries.
 	 */
 	if (inode == NULL) {
+		struct dentry *parent;
+		struct inode *dir;
+
+		if (flags & LOOKUP_RCU) {
+			parent = ACCESS_ONCE(dentry->d_parent);
+			dir = ACCESS_ONCE(parent->d_inode);
+			if (!dir)
+				return -ECHILD;
+		} else {
+			parent = dget_parent(dentry);
+			dir = parent->d_inode;
+		}
 		if (!nfs_neg_need_reval(dir, dentry, flags))
 			ret = 1;
+		else if (flags & LOOKUP_RCU)
+			ret = -ECHILD;
+		if (!(flags & LOOKUP_RCU))
+			dput(parent);
+		else if (parent != ACCESS_ONCE(dentry->d_parent))
+			return -ECHILD;
 		goto out;
 	}
 
 	/* NFS only supports OPEN on regular files */
 	if (!S_ISREG(inode->i_mode))
-		goto no_open_dput;
+		goto no_open;
 	/* We cannot do exclusive creation on a positive dentry */
 	if (flags & LOOKUP_EXCL)
-		goto no_open_dput;
+		goto no_open;
 
 	/* Let f_op->open() actually open (and revalidate) the file */
 	ret = 1;
 
 out:
-	dput(parent);
 	return ret;
 
-no_open_dput:
-	dput(parent);
 no_open:
 	return nfs_lookup_revalidate(dentry, flags);
 }
@@ -2028,10 +2072,14 @@ static DEFINE_SPINLOCK(nfs_access_lru_lock);
 static LIST_HEAD(nfs_access_lru_list);
 static atomic_long_t nfs_access_nr_entries;
 
+static unsigned long nfs_access_max_cachesize = ULONG_MAX;
+module_param(nfs_access_max_cachesize, ulong, 0644);
+MODULE_PARM_DESC(nfs_access_max_cachesize, "NFS access maximum total cache length");
+
 static void nfs_access_free_entry(struct nfs_access_entry *entry)
 {
 	put_rpccred(entry->cred);
-	kfree(entry);
+	kfree_rcu(entry, rcu_head);
 	smp_mb__before_atomic();
 	atomic_long_dec(&nfs_access_nr_entries);
 	smp_mb__after_atomic();
@@ -2048,19 +2096,14 @@ static void nfs_access_free_list(struct list_head *head)
 	}
 }
 
-unsigned long
-nfs_access_cache_scan(struct shrinker *shrink, struct shrink_control *sc)
+static unsigned long
+nfs_do_access_cache_scan(unsigned int nr_to_scan)
 {
 	LIST_HEAD(head);
 	struct nfs_inode *nfsi, *next;
 	struct nfs_access_entry *cache;
-	int nr_to_scan = sc->nr_to_scan;
-	gfp_t gfp_mask = sc->gfp_mask;
 	long freed = 0;
 
-	if ((gfp_mask & GFP_KERNEL) != GFP_KERNEL)
-		return SHRINK_STOP;
-
 	spin_lock(&nfs_access_lru_lock);
 	list_for_each_entry_safe(nfsi, next, &nfs_access_lru_list, access_cache_inode_lru) {
 		struct inode *inode;
@@ -2093,12 +2136,40 @@ remove_lru_entry:
 	return freed;
 }
 
+unsigned long
+nfs_access_cache_scan(struct shrinker *shrink, struct shrink_control *sc)
+{
+	int nr_to_scan = sc->nr_to_scan;
+	gfp_t gfp_mask = sc->gfp_mask;
+
+	if ((gfp_mask & GFP_KERNEL) != GFP_KERNEL)
+		return SHRINK_STOP;
+	return nfs_do_access_cache_scan(nr_to_scan);
+}
+
+
 unsigned long
 nfs_access_cache_count(struct shrinker *shrink, struct shrink_control *sc)
 {
 	return vfs_pressure_ratio(atomic_long_read(&nfs_access_nr_entries));
 }
 
+static void
+nfs_access_cache_enforce_limit(void)
+{
+	long nr_entries = atomic_long_read(&nfs_access_nr_entries);
+	unsigned long diff;
+	unsigned int nr_to_scan;
+
+	if (nr_entries < 0 || nr_entries <= nfs_access_max_cachesize)
+		return;
+	nr_to_scan = 100;
+	diff = nr_entries - nfs_access_max_cachesize;
+	if (diff < nr_to_scan)
+		nr_to_scan = diff;
+	nfs_do_access_cache_scan(nr_to_scan);
+}
+
 static void __nfs_access_zap_cache(struct nfs_inode *nfsi, struct list_head *head)
 {
 	struct rb_root *root_node = &nfsi->access_cache;
@@ -2186,6 +2257,38 @@ out_zap:
 	return -ENOENT;
 }
 
+static int nfs_access_get_cached_rcu(struct inode *inode, struct rpc_cred *cred, struct nfs_access_entry *res)
+{
+	/* Only check the most recently returned cache entry,
+	 * but do it without locking.
+	 */
+	struct nfs_inode *nfsi = NFS_I(inode);
+	struct nfs_access_entry *cache;
+	int err = -ECHILD;
+	struct list_head *lh;
+
+	rcu_read_lock();
+	if (nfsi->cache_validity & NFS_INO_INVALID_ACCESS)
+		goto out;
+	lh = rcu_dereference(nfsi->access_cache_entry_lru.prev);
+	cache = list_entry(lh, struct nfs_access_entry, lru);
+	if (lh == &nfsi->access_cache_entry_lru ||
+	    cred != cache->cred)
+		cache = NULL;
+	if (cache == NULL)
+		goto out;
+	if (!nfs_have_delegated_attributes(inode) &&
+	    !time_in_range_open(jiffies, cache->jiffies, cache->jiffies + nfsi->attrtimeo))
+		goto out;
+	res->jiffies = cache->jiffies;
+	res->cred = cache->cred;
+	res->mask = cache->mask;
+	err = 0;
+out:
+	rcu_read_unlock();
+	return err;
+}
+
 static void nfs_access_add_rbtree(struct inode *inode, struct nfs_access_entry *set)
 {
 	struct nfs_inode *nfsi = NFS_I(inode);
@@ -2229,6 +2332,11 @@ void nfs_access_add_cache(struct inode *inode, struct nfs_access_entry *set)
 	cache->cred = get_rpccred(set->cred);
 	cache->mask = set->mask;
 
+	/* The above field assignments must be visible
+	 * before this item appears on the lru.  We cannot easily
+	 * use rcu_assign_pointer, so just force the memory barrier.
+	 */
+	smp_wmb();
 	nfs_access_add_rbtree(inode, cache);
 
 	/* Update accounting */
@@ -2244,6 +2352,7 @@ void nfs_access_add_cache(struct inode *inode, struct nfs_access_entry *set)
 					&nfs_access_lru_list);
 		spin_unlock(&nfs_access_lru_lock);
 	}
+	nfs_access_cache_enforce_limit();
 }
 EXPORT_SYMBOL_GPL(nfs_access_add_cache);
 
@@ -2267,10 +2376,16 @@ static int nfs_do_access(struct inode *inode, struct rpc_cred *cred, int mask)
 
 	trace_nfs_access_enter(inode);
 
-	status = nfs_access_get_cached(inode, cred, &cache);
+	status = nfs_access_get_cached_rcu(inode, cred, &cache);
+	if (status != 0)
+		status = nfs_access_get_cached(inode, cred, &cache);
 	if (status == 0)
 		goto out_cached;
 
+	status = -ECHILD;
+	if (mask & MAY_NOT_BLOCK)
+		goto out;
+
 	/* Be clever: ask server to check for all possible rights */
 	cache.mask = MAY_EXEC | MAY_WRITE | MAY_READ;
 	cache.cred = cred;
@@ -2321,9 +2436,6 @@ int nfs_permission(struct inode *inode, int mask)
 	struct rpc_cred *cred;
 	int res = 0;
 
-	if (mask & MAY_NOT_BLOCK)
-		return -ECHILD;
-
 	nfs_inc_stats(inode, NFSIOS_VFSACCESS);
 
 	if ((mask & (MAY_READ | MAY_WRITE | MAY_EXEC)) == 0)
@@ -2350,12 +2462,23 @@ force_lookup:
 	if (!NFS_PROTO(inode)->access)
 		goto out_notsup;
 
-	cred = rpc_lookup_cred();
-	if (!IS_ERR(cred)) {
-		res = nfs_do_access(inode, cred, mask);
-		put_rpccred(cred);
-	} else
+	/* Always try fast lookups first */
+	rcu_read_lock();
+	cred = rpc_lookup_cred_nonblock();
+	if (!IS_ERR(cred))
+		res = nfs_do_access(inode, cred, mask|MAY_NOT_BLOCK);
+	else
 		res = PTR_ERR(cred);
+	rcu_read_unlock();
+	if (res == -ECHILD && !(mask & MAY_NOT_BLOCK)) {
+		/* Fast lookup failed, try the slow way */
+		cred = rpc_lookup_cred();
+		if (!IS_ERR(cred)) {
+			res = nfs_do_access(inode, cred, mask);
+			put_rpccred(cred);
+		} else
+			res = PTR_ERR(cred);
+	}
 out:
 	if (!res && (mask & MAY_EXEC) && !execute_ok(inode))
 		res = -EACCES;
@@ -2364,6 +2487,9 @@ out:
 		inode->i_sb->s_id, inode->i_ino, mask, res);
 	return res;
 out_notsup:
+	if (mask & MAY_NOT_BLOCK)
+		return -ECHILD;
+
 	res = nfs_revalidate_inode(NFS_SERVER(inode), inode);
 	if (res == 0)
 		res = generic_permission(inode, mask);

fs/nfs/direct.c

@@ -148,8 +148,8 @@ static void nfs_direct_set_hdr_verf(struct nfs_direct_req *dreq,
 {
 	struct nfs_writeverf *verfp;
 
-	verfp = nfs_direct_select_verf(dreq, hdr->data->ds_clp,
-				      hdr->data->ds_idx);
+	verfp = nfs_direct_select_verf(dreq, hdr->ds_clp,
+				      hdr->ds_idx);
 	WARN_ON_ONCE(verfp->committed >= 0);
 	memcpy(verfp, &hdr->verf, sizeof(struct nfs_writeverf));
 	WARN_ON_ONCE(verfp->committed < 0);
@@ -169,8 +169,8 @@ static int nfs_direct_set_or_cmp_hdr_verf(struct nfs_direct_req *dreq,
 {
 	struct nfs_writeverf *verfp;
 
-	verfp = nfs_direct_select_verf(dreq, hdr->data->ds_clp,
-					 hdr->data->ds_idx);
+	verfp = nfs_direct_select_verf(dreq, hdr->ds_clp,
+					 hdr->ds_idx);
 	if (verfp->committed < 0) {
 		nfs_direct_set_hdr_verf(dreq, hdr);
 		return 0;
@@ -715,7 +715,7 @@ static void nfs_direct_write_completion(struct nfs_pgio_header *hdr)
 {
 	struct nfs_direct_req *dreq = hdr->dreq;
 	struct nfs_commit_info cinfo;
-	int bit = -1;
+	bool request_commit = false;
 	struct nfs_page *req = nfs_list_entry(hdr->pages.next);
 
 	if (test_bit(NFS_IOHDR_REDO, &hdr->flags))
@@ -729,27 +729,20 @@ static void nfs_direct_write_completion(struct nfs_pgio_header *hdr)
 		dreq->flags = 0;
 		dreq->error = hdr->error;
 	}
-	if (dreq->error != 0)
-		bit = NFS_IOHDR_ERROR;
-	else {
+	if (dreq->error == 0) {
 		dreq->count += hdr->good_bytes;
-		if (test_bit(NFS_IOHDR_NEED_RESCHED, &hdr->flags)) {
-			dreq->flags = NFS_ODIRECT_RESCHED_WRITES;
-			bit = NFS_IOHDR_NEED_RESCHED;
-		} else if (test_bit(NFS_IOHDR_NEED_COMMIT, &hdr->flags)) {
+		if (nfs_write_need_commit(hdr)) {
 			if (dreq->flags == NFS_ODIRECT_RESCHED_WRITES)
-				bit = NFS_IOHDR_NEED_RESCHED;
+				request_commit = true;
 			else if (dreq->flags == 0) {
 				nfs_direct_set_hdr_verf(dreq, hdr);
-				bit = NFS_IOHDR_NEED_COMMIT;
+				request_commit = true;
 				dreq->flags = NFS_ODIRECT_DO_COMMIT;
 			} else if (dreq->flags == NFS_ODIRECT_DO_COMMIT) {
-				if (nfs_direct_set_or_cmp_hdr_verf(dreq, hdr)) {
+				request_commit = true;
+				if (nfs_direct_set_or_cmp_hdr_verf(dreq, hdr))
 					dreq->flags =
 						NFS_ODIRECT_RESCHED_WRITES;
-					bit = NFS_IOHDR_NEED_RESCHED;
-				} else
-					bit = NFS_IOHDR_NEED_COMMIT;
 			}
 		}
 	}
@@ -759,9 +752,7 @@ static void nfs_direct_write_completion(struct nfs_pgio_header *hdr)
 
 		req = nfs_list_entry(hdr->pages.next);
 		nfs_list_remove_request(req);
-		switch (bit) {
-		case NFS_IOHDR_NEED_RESCHED:
-		case NFS_IOHDR_NEED_COMMIT:
+		if (request_commit) {
 			kref_get(&req->wb_kref);
 			nfs_mark_request_commit(req, hdr->lseg, &cinfo);
 		}

fs/nfs/filelayout/filelayout.c

@@ -84,45 +84,37 @@ filelayout_get_dserver_offset(struct pnfs_layout_segment *lseg, loff_t offset)
 	BUG();
 }
 
-static void filelayout_reset_write(struct nfs_pgio_data *data)
+static void filelayout_reset_write(struct nfs_pgio_header *hdr)
 {
-	struct nfs_pgio_header *hdr = data->header;
-	struct rpc_task *task = &data->task;
+	struct rpc_task *task = &hdr->task;
 
 	if (!test_and_set_bit(NFS_IOHDR_REDO, &hdr->flags)) {
 		dprintk("%s Reset task %5u for i/o through MDS "
 			"(req %s/%llu, %u bytes @ offset %llu)\n", __func__,
-			data->task.tk_pid,
+			hdr->task.tk_pid,
 			hdr->inode->i_sb->s_id,
 			(unsigned long long)NFS_FILEID(hdr->inode),
-			data->args.count,
-			(unsigned long long)data->args.offset);
+			hdr->args.count,
+			(unsigned long long)hdr->args.offset);
 
-		task->tk_status = pnfs_write_done_resend_to_mds(hdr->inode,
-							&hdr->pages,
-							hdr->completion_ops,
-							hdr->dreq);
+		task->tk_status = pnfs_write_done_resend_to_mds(hdr);
 	}
 }
 
-static void filelayout_reset_read(struct nfs_pgio_data *data)
+static void filelayout_reset_read(struct nfs_pgio_header *hdr)
 {
-	struct nfs_pgio_header *hdr = data->header;
-	struct rpc_task *task = &data->task;
+	struct rpc_task *task = &hdr->task;
 
 	if (!test_and_set_bit(NFS_IOHDR_REDO, &hdr->flags)) {
 		dprintk("%s Reset task %5u for i/o through MDS "
 			"(req %s/%llu, %u bytes @ offset %llu)\n", __func__,
-			data->task.tk_pid,
+			hdr->task.tk_pid,
 			hdr->inode->i_sb->s_id,
 			(unsigned long long)NFS_FILEID(hdr->inode),
-			data->args.count,
-			(unsigned long long)data->args.offset);
+			hdr->args.count,
+			(unsigned long long)hdr->args.offset);
 
-		task->tk_status = pnfs_read_done_resend_to_mds(hdr->inode,
-							&hdr->pages,
-							hdr->completion_ops,
-							hdr->dreq);
+		task->tk_status = pnfs_read_done_resend_to_mds(hdr);
 	}
 }
 
@@ -243,18 +235,17 @@ wait_on_recovery:
 /* NFS_PROTO call done callback routines */
 
 static int filelayout_read_done_cb(struct rpc_task *task,
-				struct nfs_pgio_data *data)
+				struct nfs_pgio_header *hdr)
 {
-	struct nfs_pgio_header *hdr = data->header;
 	int err;
 
-	trace_nfs4_pnfs_read(data, task->tk_status);
-	err = filelayout_async_handle_error(task, data->args.context->state,
-					    data->ds_clp, hdr->lseg);
+	trace_nfs4_pnfs_read(hdr, task->tk_status);
+	err = filelayout_async_handle_error(task, hdr->args.context->state,
+					    hdr->ds_clp, hdr->lseg);
 
 	switch (err) {
 	case -NFS4ERR_RESET_TO_MDS:
-		filelayout_reset_read(data);
+		filelayout_reset_read(hdr);
 		return task->tk_status;
 	case -EAGAIN:
 		rpc_restart_call_prepare(task);
@@ -270,15 +261,14 @@ static int filelayout_read_done_cb(struct rpc_task *task,
  * rfc5661 is not clear about which credential should be used.
  */
 static void
-filelayout_set_layoutcommit(struct nfs_pgio_data *wdata)
+filelayout_set_layoutcommit(struct nfs_pgio_header *hdr)
 {
-	struct nfs_pgio_header *hdr = wdata->header;
 
 	if (FILELAYOUT_LSEG(hdr->lseg)->commit_through_mds ||
-	    wdata->res.verf->committed == NFS_FILE_SYNC)
+	    hdr->res.verf->committed == NFS_FILE_SYNC)
 		return;
 
-	pnfs_set_layoutcommit(wdata);
+	pnfs_set_layoutcommit(hdr);
 	dprintk("%s inode %lu pls_end_pos %lu\n", __func__, hdr->inode->i_ino,
 		(unsigned long) NFS_I(hdr->inode)->layout->plh_lwb);
 }
@@ -305,83 +295,82 @@ filelayout_reset_to_mds(struct pnfs_layout_segment *lseg)
  */
 static void filelayout_read_prepare(struct rpc_task *task, void *data)
 {
-	struct nfs_pgio_data *rdata = data;
+	struct nfs_pgio_header *hdr = data;
 
-	if (unlikely(test_bit(NFS_CONTEXT_BAD, &rdata->args.context->flags))) {
+	if (unlikely(test_bit(NFS_CONTEXT_BAD, &hdr->args.context->flags))) {
 		rpc_exit(task, -EIO);
 		return;
 	}
-	if (filelayout_reset_to_mds(rdata->header->lseg)) {
+	if (filelayout_reset_to_mds(hdr->lseg)) {
 		dprintk("%s task %u reset io to MDS\n", __func__, task->tk_pid);
-		filelayout_reset_read(rdata);
+		filelayout_reset_read(hdr);
 		rpc_exit(task, 0);
 		return;
 	}
-	rdata->pgio_done_cb = filelayout_read_done_cb;
+	hdr->pgio_done_cb = filelayout_read_done_cb;
 
-	if (nfs41_setup_sequence(rdata->ds_clp->cl_session,
-			&rdata->args.seq_args,
-			&rdata->res.seq_res,
+	if (nfs41_setup_sequence(hdr->ds_clp->cl_session,
+			&hdr->args.seq_args,
+			&hdr->res.seq_res,
 			task))
 		return;
-	if (nfs4_set_rw_stateid(&rdata->args.stateid, rdata->args.context,
-			rdata->args.lock_context, FMODE_READ) == -EIO)
+	if (nfs4_set_rw_stateid(&hdr->args.stateid, hdr->args.context,
+			hdr->args.lock_context, FMODE_READ) == -EIO)
 		rpc_exit(task, -EIO); /* lost lock, terminate I/O */
 }
 
 static void filelayout_read_call_done(struct rpc_task *task, void *data)
 {
-	struct nfs_pgio_data *rdata = data;
+	struct nfs_pgio_header *hdr = data;
 
 	dprintk("--> %s task->tk_status %d\n", __func__, task->tk_status);
 
-	if (test_bit(NFS_IOHDR_REDO, &rdata->header->flags) &&
+	if (test_bit(NFS_IOHDR_REDO, &hdr->flags) &&
 	    task->tk_status == 0) {
-		nfs41_sequence_done(task, &rdata->res.seq_res);
+		nfs41_sequence_done(task, &hdr->res.seq_res);
 		return;
 	}
 
 	/* Note this may cause RPC to be resent */
-	rdata->header->mds_ops->rpc_call_done(task, data);
+	hdr->mds_ops->rpc_call_done(task, data);
 }
 
 static void filelayout_read_count_stats(struct rpc_task *task, void *data)
 {
-	struct nfs_pgio_data *rdata = data;
+	struct nfs_pgio_header *hdr = data;
 
-	rpc_count_iostats(task, NFS_SERVER(rdata->header->inode)->client->cl_metrics);
+	rpc_count_iostats(task, NFS_SERVER(hdr->inode)->client->cl_metrics);
 }
 
 static void filelayout_read_release(void *data)
 {
-	struct nfs_pgio_data *rdata = data;
-	struct pnfs_layout_hdr *lo = rdata->header->lseg->pls_layout;
+	struct nfs_pgio_header *hdr = data;
+	struct pnfs_layout_hdr *lo = hdr->lseg->pls_layout;
 
 	filelayout_fenceme(lo->plh_inode, lo);
-	nfs_put_client(rdata->ds_clp);
-	rdata->header->mds_ops->rpc_release(data);
+	nfs_put_client(hdr->ds_clp);
+	hdr->mds_ops->rpc_release(data);
 }
 
 static int filelayout_write_done_cb(struct rpc_task *task,
-				struct nfs_pgio_data *data)
+				struct nfs_pgio_header *hdr)
 {
-	struct nfs_pgio_header *hdr = data->header;
 	int err;
 
-	trace_nfs4_pnfs_write(data, task->tk_status);
-	err = filelayout_async_handle_error(task, data->args.context->state,
-					    data->ds_clp, hdr->lseg);
+	trace_nfs4_pnfs_write(hdr, task->tk_status);
+	err = filelayout_async_handle_error(task, hdr->args.context->state,
+					    hdr->ds_clp, hdr->lseg);
 
 	switch (err) {
 	case -NFS4ERR_RESET_TO_MDS:
-		filelayout_reset_write(data);
+		filelayout_reset_write(hdr);
 		return task->tk_status;
 	case -EAGAIN:
 		rpc_restart_call_prepare(task);
 		return -EAGAIN;
 	}
 
-	filelayout_set_layoutcommit(data);
+	filelayout_set_layoutcommit(hdr);
 	return 0;
 }
 
@@ -419,57 +408,57 @@ static int filelayout_commit_done_cb(struct rpc_task *task,
 
 static void filelayout_write_prepare(struct rpc_task *task, void *data)
 {
-	struct nfs_pgio_data *wdata = data;
+	struct nfs_pgio_header *hdr = data;
 
-	if (unlikely(test_bit(NFS_CONTEXT_BAD, &wdata->args.context->flags))) {
+	if (unlikely(test_bit(NFS_CONTEXT_BAD, &hdr->args.context->flags))) {
 		rpc_exit(task, -EIO);
 		return;
 	}
-	if (filelayout_reset_to_mds(wdata->header->lseg)) {
+	if (filelayout_reset_to_mds(hdr->lseg)) {
 		dprintk("%s task %u reset io to MDS\n", __func__, task->tk_pid);
-		filelayout_reset_write(wdata);
+		filelayout_reset_write(hdr);
 		rpc_exit(task, 0);
 		return;
 	}
-	if (nfs41_setup_sequence(wdata->ds_clp->cl_session,
-			&wdata->args.seq_args,
-			&wdata->res.seq_res,
+	if (nfs41_setup_sequence(hdr->ds_clp->cl_session,
+			&hdr->args.seq_args,
+			&hdr->res.seq_res,
 			task))
 		return;
-	if (nfs4_set_rw_stateid(&wdata->args.stateid, wdata->args.context,
-			wdata->args.lock_context, FMODE_WRITE) == -EIO)
+	if (nfs4_set_rw_stateid(&hdr->args.stateid, hdr->args.context,
+			hdr->args.lock_context, FMODE_WRITE) == -EIO)
 		rpc_exit(task, -EIO); /* lost lock, terminate I/O */
 }
 
 static void filelayout_write_call_done(struct rpc_task *task, void *data)
 {
-	struct nfs_pgio_data *wdata = data;
+	struct nfs_pgio_header *hdr = data;
 
-	if (test_bit(NFS_IOHDR_REDO, &wdata->header->flags) &&
+	if (test_bit(NFS_IOHDR_REDO, &hdr->flags) &&
 	    task->tk_status == 0) {
-		nfs41_sequence_done(task, &wdata->res.seq_res);
+		nfs41_sequence_done(task, &hdr->res.seq_res);
 		return;
 	}
 
 	/* Note this may cause RPC to be resent */
-	wdata->header->mds_ops->rpc_call_done(task, data);
+	hdr->mds_ops->rpc_call_done(task, data);
 }
 
 static void filelayout_write_count_stats(struct rpc_task *task, void *data)
 {
-	struct nfs_pgio_data *wdata = data;
+	struct nfs_pgio_header *hdr = data;
 
-	rpc_count_iostats(task, NFS_SERVER(wdata->header->inode)->client->cl_metrics);
+	rpc_count_iostats(task, NFS_SERVER(hdr->inode)->client->cl_metrics);
 }
 
 static void filelayout_write_release(void *data)
 {
-	struct nfs_pgio_data *wdata = data;
-	struct pnfs_layout_hdr *lo = wdata->header->lseg->pls_layout;
+	struct nfs_pgio_header *hdr = data;
+	struct pnfs_layout_hdr *lo = hdr->lseg->pls_layout;
 
 	filelayout_fenceme(lo->plh_inode, lo);
-	nfs_put_client(wdata->ds_clp);
-	wdata->header->mds_ops->rpc_release(data);
+	nfs_put_client(hdr->ds_clp);
+	hdr->mds_ops->rpc_release(data);
 }
 
 static void filelayout_commit_prepare(struct rpc_task *task, void *data)
@@ -529,19 +518,18 @@ static const struct rpc_call_ops filelayout_commit_call_ops = {
 };
 
 static enum pnfs_try_status
-filelayout_read_pagelist(struct nfs_pgio_data *data)
+filelayout_read_pagelist(struct nfs_pgio_header *hdr)
 {
-	struct nfs_pgio_header *hdr = data->header;
 	struct pnfs_layout_segment *lseg = hdr->lseg;
 	struct nfs4_pnfs_ds *ds;
 	struct rpc_clnt *ds_clnt;
-	loff_t offset = data->args.offset;
+	loff_t offset = hdr->args.offset;
 	u32 j, idx;
 	struct nfs_fh *fh;
 
 	dprintk("--> %s ino %lu pgbase %u req %Zu@%llu\n",
 		__func__, hdr->inode->i_ino,
-		data->args.pgbase, (size_t)data->args.count, offset);
+		hdr->args.pgbase, (size_t)hdr->args.count, offset);
 
 	/* Retrieve the correct rpc_client for the byte range */
 	j = nfs4_fl_calc_j_index(lseg, offset);
@@ -559,30 +547,29 @@ filelayout_read_pagelist(struct nfs_pgio_data *data)
 
 	/* No multipath support. Use first DS */
 	atomic_inc(&ds->ds_clp->cl_count);
-	data->ds_clp = ds->ds_clp;
-	data->ds_idx = idx;
+	hdr->ds_clp = ds->ds_clp;
+	hdr->ds_idx = idx;
 	fh = nfs4_fl_select_ds_fh(lseg, j);
 	if (fh)
-		data->args.fh = fh;
+		hdr->args.fh = fh;
 
-	data->args.offset = filelayout_get_dserver_offset(lseg, offset);
-	data->mds_offset = offset;
+	hdr->args.offset = filelayout_get_dserver_offset(lseg, offset);
+	hdr->mds_offset = offset;
 
 	/* Perform an asynchronous read to ds */
-	nfs_initiate_pgio(ds_clnt, data,
+	nfs_initiate_pgio(ds_clnt, hdr,
 			    &filelayout_read_call_ops, 0, RPC_TASK_SOFTCONN);
 	return PNFS_ATTEMPTED;
 }
 
 /* Perform async writes. */
 static enum pnfs_try_status
-filelayout_write_pagelist(struct nfs_pgio_data *data, int sync)
+filelayout_write_pagelist(struct nfs_pgio_header *hdr, int sync)
 {
-	struct nfs_pgio_header *hdr = data->header;
 	struct pnfs_layout_segment *lseg = hdr->lseg;
 	struct nfs4_pnfs_ds *ds;
 	struct rpc_clnt *ds_clnt;
-	loff_t offset = data->args.offset;
+	loff_t offset = hdr->args.offset;
 	u32 j, idx;
 	struct nfs_fh *fh;
 
@@ -598,21 +585,20 @@ filelayout_write_pagelist(struct nfs_pgio_data *data, int sync)
 		return PNFS_NOT_ATTEMPTED;
 
 	dprintk("%s ino %lu sync %d req %Zu@%llu DS: %s cl_count %d\n",
-		__func__, hdr->inode->i_ino, sync, (size_t) data->args.count,
+		__func__, hdr->inode->i_ino, sync, (size_t) hdr->args.count,
 		offset, ds->ds_remotestr, atomic_read(&ds->ds_clp->cl_count));
 
-	data->pgio_done_cb = filelayout_write_done_cb;
+	hdr->pgio_done_cb = filelayout_write_done_cb;
 	atomic_inc(&ds->ds_clp->cl_count);
-	data->ds_clp = ds->ds_clp;
-	data->ds_idx = idx;
+	hdr->ds_clp = ds->ds_clp;
+	hdr->ds_idx = idx;
 	fh = nfs4_fl_select_ds_fh(lseg, j);
 	if (fh)
-		data->args.fh = fh;
-
-	data->args.offset = filelayout_get_dserver_offset(lseg, offset);
+		hdr->args.fh = fh;
+	hdr->args.offset = filelayout_get_dserver_offset(lseg, offset);
 
 	/* Perform an asynchronous write */
-	nfs_initiate_pgio(ds_clnt, data,
+	nfs_initiate_pgio(ds_clnt, hdr,
 				    &filelayout_write_call_ops, sync,
 				    RPC_TASK_SOFTCONN);
 	return PNFS_ATTEMPTED;
@@ -1023,6 +1009,7 @@ static u32 select_bucket_index(struct nfs4_filelayout_segment *fl, u32 j)
 
 /* The generic layer is about to remove the req from the commit list.
  * If this will make the bucket empty, it will need to put the lseg reference.
+ * Note this is must be called holding the inode (/cinfo) lock
  */
 static void
 filelayout_clear_request_commit(struct nfs_page *req,
@@ -1030,7 +1017,6 @@ filelayout_clear_request_commit(struct nfs_page *req,
 {
 	struct pnfs_layout_segment *freeme = NULL;
 
-	spin_lock(cinfo->lock);
 	if (!test_and_clear_bit(PG_COMMIT_TO_DS, &req->wb_flags))
 		goto out;
 	cinfo->ds->nwritten--;
@@ -1045,22 +1031,25 @@ filelayout_clear_request_commit(struct nfs_page *req,
 	}
 out:
 	nfs_request_remove_commit_list(req, cinfo);
-	spin_unlock(cinfo->lock);
-	pnfs_put_lseg(freeme);
+	pnfs_put_lseg_async(freeme);
 }
 
-static struct list_head *
-filelayout_choose_commit_list(struct nfs_page *req,
-			      struct pnfs_layout_segment *lseg,
-			      struct nfs_commit_info *cinfo)
+static void
+filelayout_mark_request_commit(struct nfs_page *req,
+			       struct pnfs_layout_segment *lseg,
+			       struct nfs_commit_info *cinfo)
+
 {
 	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)
-		return &cinfo->mds->list;
+	if (fl->commit_through_mds) {
+		list = &cinfo->mds->list;
+		spin_lock(cinfo->lock);
+		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
@@ -1084,19 +1073,22 @@ filelayout_choose_commit_list(struct nfs_page *req,
 	}
 	set_bit(PG_COMMIT_TO_DS, &req->wb_flags);
 	cinfo->ds->nwritten++;
-	spin_unlock(cinfo->lock);
-	return list;
-}
 
-static void
-filelayout_mark_request_commit(struct nfs_page *req,
-			       struct pnfs_layout_segment *lseg,
-			       struct nfs_commit_info *cinfo)
-{
-	struct list_head *list;
-
-	list = filelayout_choose_commit_list(req, lseg, cinfo);
-	nfs_request_add_commit_list(req, list, cinfo);
+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(page_file_mapping(req->wb_page)->backing_dev_info,
+			     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)
@@ -1244,15 +1236,63 @@ restart:
 	spin_unlock(cinfo->lock);
 }
 
+/* filelayout_search_commit_reqs - Search lists in @cinfo for the head reqest
+ *				   for @page
+ * @cinfo - commit info for current inode
+ * @page - page to search for matching head request
+ *
+ * Returns a the head request if one is found, otherwise returns NULL.
+ */
+static struct nfs_page *
+filelayout_search_commit_reqs(struct nfs_commit_info *cinfo, struct page *page)
+{
+	struct nfs_page *freq, *t;
+	struct pnfs_commit_bucket *b;
+	int i;
+
+	/* Linearly search the commit lists for each bucket until a matching
+	 * request is found */
+	for (i = 0, b = cinfo->ds->buckets; i < cinfo->ds->nbuckets; i++, b++) {
+		list_for_each_entry_safe(freq, t, &b->written, wb_list) {
+			if (freq->wb_page == page)
+				return freq->wb_head;
+		}
+		list_for_each_entry_safe(freq, t, &b->committing, wb_list) {
+			if (freq->wb_page == page)
+				return freq->wb_head;
+		}
+	}
+
+	return NULL;
+}
+
+static void filelayout_retry_commit(struct nfs_commit_info *cinfo, int idx)
+{
+	struct pnfs_ds_commit_info *fl_cinfo = cinfo->ds;
+	struct pnfs_commit_bucket *bucket = fl_cinfo->buckets;
+	struct pnfs_layout_segment *freeme;
+	int i;
+
+	for (i = idx; i < fl_cinfo->nbuckets; i++, bucket++) {
+		if (list_empty(&bucket->committing))
+			continue;
+		nfs_retry_commit(&bucket->committing, bucket->clseg, cinfo);
+		spin_lock(cinfo->lock);
+		freeme = bucket->clseg;
+		bucket->clseg = NULL;
+		spin_unlock(cinfo->lock);
+		pnfs_put_lseg(freeme);
+	}
+}
+
 static unsigned int
 alloc_ds_commits(struct nfs_commit_info *cinfo, struct list_head *list)
 {
 	struct pnfs_ds_commit_info *fl_cinfo;
 	struct pnfs_commit_bucket *bucket;
 	struct nfs_commit_data *data;
-	int i, j;
+	int i;
 	unsigned int nreq = 0;
-	struct pnfs_layout_segment *freeme;
 
 	fl_cinfo = cinfo->ds;
 	bucket = fl_cinfo->buckets;
@@ -1272,16 +1312,7 @@ alloc_ds_commits(struct nfs_commit_info *cinfo, struct list_head *list)
 	}
 
 	/* Clean up on error */
-	for (j = i; j < fl_cinfo->nbuckets; j++, bucket++) {
-		if (list_empty(&bucket->committing))
-			continue;
-		nfs_retry_commit(&bucket->committing, bucket->clseg, cinfo);
-		spin_lock(cinfo->lock);
-		freeme = bucket->clseg;
-		bucket->clseg = NULL;
-		spin_unlock(cinfo->lock);
-		pnfs_put_lseg(freeme);
-	}
+	filelayout_retry_commit(cinfo, i);
 	/* Caller will clean up entries put on list */
 	return nreq;
 }
@@ -1301,8 +1332,12 @@ filelayout_commit_pagelist(struct inode *inode, struct list_head *mds_pages,
 			data->lseg = NULL;
 			list_add(&data->pages, &list);
 			nreq++;
-		} else
+		} else {
 			nfs_retry_commit(mds_pages, NULL, cinfo);
+			filelayout_retry_commit(cinfo, 0);
+			cinfo->completion_ops->error_cleanup(NFS_I(inode));
+			return -ENOMEM;
+		}
 	}
 
 	nreq += alloc_ds_commits(cinfo, &list);
@@ -1380,6 +1415,7 @@ static struct pnfs_layoutdriver_type filelayout_type = {
 	.clear_request_commit	= filelayout_clear_request_commit,
 	.scan_commit_lists	= filelayout_scan_commit_lists,
 	.recover_commit_reqs	= filelayout_recover_commit_reqs,
+	.search_commit_reqs	= filelayout_search_commit_reqs,
 	.commit_pagelist	= filelayout_commit_pagelist,
 	.read_pagelist		= filelayout_read_pagelist,
 	.write_pagelist		= filelayout_write_pagelist,

fs/nfs/filelayout/filelayoutdev.c

@@ -695,7 +695,7 @@ filelayout_get_device_info(struct inode *inode,
 	if (pdev == NULL)
 		return NULL;
 
-	pages = kzalloc(max_pages * sizeof(struct page *), gfp_flags);
+	pages = kcalloc(max_pages, sizeof(struct page *), gfp_flags);
 	if (pages == NULL) {
 		kfree(pdev);
 		return NULL;

fs/nfs/inode.c

@@ -1002,6 +1002,15 @@ int nfs_revalidate_inode(struct nfs_server *server, struct inode *inode)
 }
 EXPORT_SYMBOL_GPL(nfs_revalidate_inode);
 
+int nfs_revalidate_inode_rcu(struct nfs_server *server, struct inode *inode)
+{
+	if (!(NFS_I(inode)->cache_validity &
+			(NFS_INO_INVALID_ATTR|NFS_INO_INVALID_LABEL))
+			&& !nfs_attribute_cache_expired(inode))
+		return NFS_STALE(inode) ? -ESTALE : 0;
+	return -ECHILD;
+}
+
 static int nfs_invalidate_mapping(struct inode *inode, struct address_space *mapping)
 {
 	struct nfs_inode *nfsi = NFS_I(inode);

fs/nfs/internal.h

@@ -247,11 +247,11 @@ void nfs_set_pgio_error(struct nfs_pgio_header *hdr, int error, loff_t pos);
 int nfs_iocounter_wait(struct nfs_io_counter *c);
 
 extern const struct nfs_pageio_ops nfs_pgio_rw_ops;
-struct nfs_rw_header *nfs_rw_header_alloc(const struct nfs_rw_ops *);
-void nfs_rw_header_free(struct nfs_pgio_header *);
-void nfs_pgio_data_release(struct nfs_pgio_data *);
+struct nfs_pgio_header *nfs_pgio_header_alloc(const struct nfs_rw_ops *);
+void nfs_pgio_header_free(struct nfs_pgio_header *);
+void nfs_pgio_data_destroy(struct nfs_pgio_header *);
 int nfs_generic_pgio(struct nfs_pageio_descriptor *, struct nfs_pgio_header *);
-int nfs_initiate_pgio(struct rpc_clnt *, struct nfs_pgio_data *,
+int nfs_initiate_pgio(struct rpc_clnt *, struct nfs_pgio_header *,
 		      const struct rpc_call_ops *, int, int);
 void nfs_free_request(struct nfs_page *req);
 
@@ -451,6 +451,7 @@ int nfs_scan_commit(struct inode *inode, struct list_head *dst,
 void nfs_mark_request_commit(struct nfs_page *req,
 			     struct pnfs_layout_segment *lseg,
 			     struct nfs_commit_info *cinfo);
+int nfs_write_need_commit(struct nfs_pgio_header *);
 int nfs_generic_commit_list(struct inode *inode, struct list_head *head,
 			    int how, struct nfs_commit_info *cinfo);
 void nfs_retry_commit(struct list_head *page_list,
@@ -491,7 +492,7 @@ static inline void nfs_inode_dio_wait(struct inode *inode)
 extern ssize_t nfs_dreq_bytes_left(struct nfs_direct_req *dreq);
 
 /* nfs4proc.c */
-extern void __nfs4_read_done_cb(struct nfs_pgio_data *);
+extern void __nfs4_read_done_cb(struct nfs_pgio_header *);
 extern struct nfs_client *nfs4_init_client(struct nfs_client *clp,
 			    const struct rpc_timeout *timeparms,
 			    const char *ip_addr);

fs/nfs/nfs3acl.c

@@ -256,7 +256,7 @@ nfs3_list_one_acl(struct inode *inode, int type, const char *name, void *data,
 	char *p = data + *result;
 
 	acl = get_acl(inode, type);
-	if (!acl)
+	if (IS_ERR_OR_NULL(acl))
 		return 0;
 
 	posix_acl_release(acl);

fs/nfs/nfs3proc.c

@@ -795,41 +795,44 @@ nfs3_proc_pathconf(struct nfs_server *server, struct nfs_fh *fhandle,
 	return status;
 }
 
-static int nfs3_read_done(struct rpc_task *task, struct nfs_pgio_data *data)
+static int nfs3_read_done(struct rpc_task *task, struct nfs_pgio_header *hdr)
 {
-	struct inode *inode = data->header->inode;
+	struct inode *inode = hdr->inode;
 
 	if (nfs3_async_handle_jukebox(task, inode))
 		return -EAGAIN;
 
 	nfs_invalidate_atime(inode);
-	nfs_refresh_inode(inode, &data->fattr);
+	nfs_refresh_inode(inode, &hdr->fattr);
 	return 0;
 }
 
-static void nfs3_proc_read_setup(struct nfs_pgio_data *data, struct rpc_message *msg)
+static void nfs3_proc_read_setup(struct nfs_pgio_header *hdr,
+				 struct rpc_message *msg)
 {
 	msg->rpc_proc = &nfs3_procedures[NFS3PROC_READ];
 }
 
-static int nfs3_proc_pgio_rpc_prepare(struct rpc_task *task, struct nfs_pgio_data *data)
+static int nfs3_proc_pgio_rpc_prepare(struct rpc_task *task,
+				      struct nfs_pgio_header *hdr)
 {
 	rpc_call_start(task);
 	return 0;
 }
 
-static int nfs3_write_done(struct rpc_task *task, struct nfs_pgio_data *data)
+static int nfs3_write_done(struct rpc_task *task, struct nfs_pgio_header *hdr)
 {
-	struct inode *inode = data->header->inode;
+	struct inode *inode = hdr->inode;
 
 	if (nfs3_async_handle_jukebox(task, inode))
 		return -EAGAIN;
 	if (task->tk_status >= 0)
-		nfs_post_op_update_inode_force_wcc(inode, data->res.fattr);
+		nfs_post_op_update_inode_force_wcc(inode, hdr->res.fattr);
 	return 0;
 }
 
-static void nfs3_proc_write_setup(struct nfs_pgio_data *data, struct rpc_message *msg)
+static void nfs3_proc_write_setup(struct nfs_pgio_header *hdr,
+				  struct rpc_message *msg)
 {
 	msg->rpc_proc = &nfs3_procedures[NFS3PROC_WRITE];
 }

fs/nfs/nfs4_fs.h

@@ -54,7 +54,7 @@ struct nfs4_minor_version_ops {
 			const nfs4_stateid *);
 	int	(*find_root_sec)(struct nfs_server *, struct nfs_fh *,
 			struct nfs_fsinfo *);
-	int	(*free_lock_state)(struct nfs_server *,
+	void	(*free_lock_state)(struct nfs_server *,
 			struct nfs4_lock_state *);
 	const struct rpc_call_ops *call_sync_ops;
 	const struct nfs4_state_recovery_ops *reboot_recovery_ops;
@@ -129,27 +129,17 @@ enum {
  * LOCK: one nfs4_state (LOCK) to hold the lock stateid nfs4_state(OPEN)
  */
 
-struct nfs4_lock_owner {
-	unsigned int lo_type;
-#define NFS4_ANY_LOCK_TYPE	(0U)
-#define NFS4_FLOCK_LOCK_TYPE	(1U << 0)
-#define NFS4_POSIX_LOCK_TYPE	(1U << 1)
-	union {
-		fl_owner_t posix_owner;
-		pid_t flock_owner;
-	} lo_u;
-};
-
 struct nfs4_lock_state {
-	struct list_head	ls_locks;	/* Other lock stateids */
-	struct nfs4_state *	ls_state;	/* Pointer to open state */
+	struct list_head		ls_locks;   /* Other lock stateids */
+	struct nfs4_state *		ls_state;   /* Pointer to open state */
 #define NFS_LOCK_INITIALIZED 0
 #define NFS_LOCK_LOST        1
-	unsigned long		ls_flags;
+	unsigned long			ls_flags;
 	struct nfs_seqid_counter	ls_seqid;
-	nfs4_stateid		ls_stateid;
-	atomic_t		ls_count;
-	struct nfs4_lock_owner	ls_owner;
+	nfs4_stateid			ls_stateid;
+	atomic_t			ls_count;
+	fl_owner_t			ls_owner;
+	struct work_struct		ls_release;
 };
 
 /* bits for nfs4_state->flags */
@@ -337,11 +327,11 @@ nfs4_state_protect(struct nfs_client *clp, unsigned long sp4_mode,
  */
 static inline void
 nfs4_state_protect_write(struct nfs_client *clp, struct rpc_clnt **clntp,
-			 struct rpc_message *msg, struct nfs_pgio_data *wdata)
+			 struct rpc_message *msg, struct nfs_pgio_header *hdr)
 {
 	if (_nfs4_state_protect(clp, NFS_SP4_MACH_CRED_WRITE, clntp, msg) &&
 	    !test_bit(NFS_SP4_MACH_CRED_COMMIT, &clp->cl_sp4_flags))
-		wdata->args.stable = NFS_FILE_SYNC;
+		hdr->args.stable = NFS_FILE_SYNC;
 }
 #else /* CONFIG_NFS_v4_1 */
 static inline struct nfs4_session *nfs4_get_session(const struct nfs_server *server)
@@ -369,7 +359,7 @@ nfs4_state_protect(struct nfs_client *clp, unsigned long sp4_flags,
 
 static inline void
 nfs4_state_protect_write(struct nfs_client *clp, struct rpc_clnt **clntp,
-			 struct rpc_message *msg, struct nfs_pgio_data *wdata)
+			 struct rpc_message *msg, struct nfs_pgio_header *hdr)
 {
 }
 #endif /* CONFIG_NFS_V4_1 */

fs/nfs/nfs4client.c

@@ -855,6 +855,11 @@ struct nfs_client *nfs4_set_ds_client(struct nfs_client* mds_clp,
 	};
 	struct rpc_timeout ds_timeout;
 	struct nfs_client *clp;
+	char buf[INET6_ADDRSTRLEN + 1];
+
+	if (rpc_ntop(ds_addr, buf, sizeof(buf)) <= 0)
+		return ERR_PTR(-EINVAL);
+	cl_init.hostname = buf;
 
 	/*
 	 * Set an authflavor equual to the MDS value. Use the MDS nfs_client

fs/nfs/nfs4proc.c

@@ -1952,6 +1952,14 @@ static int _nfs4_recover_proc_open(struct nfs4_opendata *data)
 	return status;
 }
 
+/*
+ * Additional permission checks in order to distinguish between an
+ * open for read, and an open for execute. This works around the
+ * fact that NFSv4 OPEN treats read and execute permissions as being
+ * the same.
+ * Note that in the non-execute case, we want to turn off permission
+ * checking if we just created a new file (POSIX open() semantics).
+ */
 static int nfs4_opendata_access(struct rpc_cred *cred,
 				struct nfs4_opendata *opendata,
 				struct nfs4_state *state, fmode_t fmode,
@@ -1966,14 +1974,14 @@ static int nfs4_opendata_access(struct rpc_cred *cred,
 		return 0;
 
 	mask = 0;
-	/* don't check MAY_WRITE - a newly created file may not have
-	 * write mode bits, but POSIX allows the creating process to write.
-	 * use openflags to check for exec, because fmode won't
-	 * always have FMODE_EXEC set when file open for exec. */
+	/*
+	 * Use openflags to check for exec, because fmode won't
+	 * always have FMODE_EXEC set when file open for exec.
+	 */
 	if (openflags & __FMODE_EXEC) {
 		/* ONLY check for exec rights */
 		mask = MAY_EXEC;
-	} else if (fmode & FMODE_READ)
+	} else if ((fmode & FMODE_READ) && !opendata->file_created)
 		mask = MAY_READ;
 
 	cache.cred = cred;
@@ -2216,8 +2224,15 @@ static int _nfs4_open_and_get_state(struct nfs4_opendata *opendata,
 	seq = raw_seqcount_begin(&sp->so_reclaim_seqcount);
 
 	ret = _nfs4_proc_open(opendata);
-	if (ret != 0)
+	if (ret != 0) {
+		if (ret == -ENOENT) {
+			d_drop(opendata->dentry);
+			d_add(opendata->dentry, NULL);
+			nfs_set_verifier(opendata->dentry,
+					 nfs_save_change_attribute(opendata->dir->d_inode));
+		}
 		goto out;
+	}
 
 	state = nfs4_opendata_to_nfs4_state(opendata);
 	ret = PTR_ERR(state);
@@ -2647,6 +2662,48 @@ static const struct rpc_call_ops nfs4_close_ops = {
 	.rpc_release = nfs4_free_closedata,
 };
 
+static bool nfs4_state_has_opener(struct nfs4_state *state)
+{
+	/* first check existing openers */
+	if (test_bit(NFS_O_RDONLY_STATE, &state->flags) != 0 &&
+	    state->n_rdonly != 0)
+		return true;
+
+	if (test_bit(NFS_O_WRONLY_STATE, &state->flags) != 0 &&
+	    state->n_wronly != 0)
+		return true;
+
+	if (test_bit(NFS_O_RDWR_STATE, &state->flags) != 0 &&
+	    state->n_rdwr != 0)
+		return true;
+
+	return false;
+}
+
+static bool nfs4_roc(struct inode *inode)
+{
+	struct nfs_inode *nfsi = NFS_I(inode);
+	struct nfs_open_context *ctx;
+	struct nfs4_state *state;
+
+	spin_lock(&inode->i_lock);
+	list_for_each_entry(ctx, &nfsi->open_files, list) {
+		state = ctx->state;
+		if (state == NULL)
+			continue;
+		if (nfs4_state_has_opener(state)) {
+			spin_unlock(&inode->i_lock);
+			return false;
+		}
+	}
+	spin_unlock(&inode->i_lock);
+
+	if (nfs4_check_delegation(inode, FMODE_READ))
+		return false;
+
+	return pnfs_roc(inode);
+}
+
 /* 
  * It is possible for data to be read/written from a mem-mapped file 
  * after the sys_close call (which hits the vfs layer as a flush).
@@ -2697,7 +2754,7 @@ int nfs4_do_close(struct nfs4_state *state, gfp_t gfp_mask, int wait)
 	calldata->res.fattr = &calldata->fattr;
 	calldata->res.seqid = calldata->arg.seqid;
 	calldata->res.server = server;
-	calldata->roc = pnfs_roc(state->inode);
+	calldata->roc = nfs4_roc(state->inode);
 	nfs_sb_active(calldata->inode->i_sb);
 
 	msg.rpc_argp = &calldata->arg;
@@ -4033,24 +4090,25 @@ static bool nfs4_error_stateid_expired(int err)
 	return false;
 }
 
-void __nfs4_read_done_cb(struct nfs_pgio_data *data)
+void __nfs4_read_done_cb(struct nfs_pgio_header *hdr)
 {
-	nfs_invalidate_atime(data->header->inode);
+	nfs_invalidate_atime(hdr->inode);
 }
 
-static int nfs4_read_done_cb(struct rpc_task *task, struct nfs_pgio_data *data)
+static int nfs4_read_done_cb(struct rpc_task *task, struct nfs_pgio_header *hdr)
 {
-	struct nfs_server *server = NFS_SERVER(data->header->inode);
+	struct nfs_server *server = NFS_SERVER(hdr->inode);
 
-	trace_nfs4_read(data, task->tk_status);
-	if (nfs4_async_handle_error(task, server, data->args.context->state) == -EAGAIN) {
+	trace_nfs4_read(hdr, task->tk_status);
+	if (nfs4_async_handle_error(task, server,
+				    hdr->args.context->state) == -EAGAIN) {
 		rpc_restart_call_prepare(task);
 		return -EAGAIN;
 	}
 
-	__nfs4_read_done_cb(data);
+	__nfs4_read_done_cb(hdr);
 	if (task->tk_status > 0)
-		renew_lease(server, data->timestamp);
+		renew_lease(server, hdr->timestamp);
 	return 0;
 }
 
@@ -4068,54 +4126,59 @@ static bool nfs4_read_stateid_changed(struct rpc_task *task,
 	return true;
 }
 
-static int nfs4_read_done(struct rpc_task *task, struct nfs_pgio_data *data)
+static int nfs4_read_done(struct rpc_task *task, struct nfs_pgio_header *hdr)
 {
 
 	dprintk("--> %s\n", __func__);
 
-	if (!nfs4_sequence_done(task, &data->res.seq_res))
+	if (!nfs4_sequence_done(task, &hdr->res.seq_res))
 		return -EAGAIN;
-	if (nfs4_read_stateid_changed(task, &data->args))
+	if (nfs4_read_stateid_changed(task, &hdr->args))
 		return -EAGAIN;
-	return data->pgio_done_cb ? data->pgio_done_cb(task, data) :
-				    nfs4_read_done_cb(task, data);
+	return hdr->pgio_done_cb ? hdr->pgio_done_cb(task, hdr) :
+				    nfs4_read_done_cb(task, hdr);
 }
 
-static void nfs4_proc_read_setup(struct nfs_pgio_data *data, struct rpc_message *msg)
+static void nfs4_proc_read_setup(struct nfs_pgio_header *hdr,
+				 struct rpc_message *msg)
 {
-	data->timestamp   = jiffies;
-	data->pgio_done_cb = nfs4_read_done_cb;
+	hdr->timestamp   = jiffies;
+	hdr->pgio_done_cb = nfs4_read_done_cb;
 	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_READ];
-	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 0);
+	nfs4_init_sequence(&hdr->args.seq_args, &hdr->res.seq_res, 0);
 }
 
-static int nfs4_proc_pgio_rpc_prepare(struct rpc_task *task, struct nfs_pgio_data *data)
+static int nfs4_proc_pgio_rpc_prepare(struct rpc_task *task,
+				      struct nfs_pgio_header *hdr)
 {
-	if (nfs4_setup_sequence(NFS_SERVER(data->header->inode),
-			&data->args.seq_args,
-			&data->res.seq_res,
+	if (nfs4_setup_sequence(NFS_SERVER(hdr->inode),
+			&hdr->args.seq_args,
+			&hdr->res.seq_res,
 			task))
 		return 0;
-	if (nfs4_set_rw_stateid(&data->args.stateid, data->args.context,
-				data->args.lock_context, data->header->rw_ops->rw_mode) == -EIO)
+	if (nfs4_set_rw_stateid(&hdr->args.stateid, hdr->args.context,
+				hdr->args.lock_context,
+				hdr->rw_ops->rw_mode) == -EIO)
 		return -EIO;
-	if (unlikely(test_bit(NFS_CONTEXT_BAD, &data->args.context->flags)))
+	if (unlikely(test_bit(NFS_CONTEXT_BAD, &hdr->args.context->flags)))
 		return -EIO;
 	return 0;
 }
 
-static int nfs4_write_done_cb(struct rpc_task *task, struct nfs_pgio_data *data)
+static int nfs4_write_done_cb(struct rpc_task *task,
+			      struct nfs_pgio_header *hdr)
 {
-	struct inode *inode = data->header->inode;
+	struct inode *inode = hdr->inode;
 	
-	trace_nfs4_write(data, task->tk_status);
-	if (nfs4_async_handle_error(task, NFS_SERVER(inode), data->args.context->state) == -EAGAIN) {
+	trace_nfs4_write(hdr, task->tk_status);
+	if (nfs4_async_handle_error(task, NFS_SERVER(inode),
+				    hdr->args.context->state) == -EAGAIN) {
 		rpc_restart_call_prepare(task);
 		return -EAGAIN;
 	}
 	if (task->tk_status >= 0) {
-		renew_lease(NFS_SERVER(inode), data->timestamp);
-		nfs_post_op_update_inode_force_wcc(inode, &data->fattr);
+		renew_lease(NFS_SERVER(inode), hdr->timestamp);
+		nfs_post_op_update_inode_force_wcc(inode, &hdr->fattr);
 	}
 	return 0;
 }
@@ -4134,23 +4197,21 @@ static bool nfs4_write_stateid_changed(struct rpc_task *task,
 	return true;
 }
 
-static int nfs4_write_done(struct rpc_task *task, struct nfs_pgio_data *data)
+static int nfs4_write_done(struct rpc_task *task, struct nfs_pgio_header *hdr)
 {
-	if (!nfs4_sequence_done(task, &data->res.seq_res))
+	if (!nfs4_sequence_done(task, &hdr->res.seq_res))
 		return -EAGAIN;
-	if (nfs4_write_stateid_changed(task, &data->args))
+	if (nfs4_write_stateid_changed(task, &hdr->args))
 		return -EAGAIN;
-	return data->pgio_done_cb ? data->pgio_done_cb(task, data) :
-		nfs4_write_done_cb(task, data);
+	return hdr->pgio_done_cb ? hdr->pgio_done_cb(task, hdr) :
+		nfs4_write_done_cb(task, hdr);
 }
 
 static
-bool nfs4_write_need_cache_consistency_data(const struct nfs_pgio_data *data)
+bool nfs4_write_need_cache_consistency_data(struct nfs_pgio_header *hdr)
 {
-	const struct nfs_pgio_header *hdr = data->header;
-
 	/* Don't request attributes for pNFS or O_DIRECT writes */
-	if (data->ds_clp != NULL || hdr->dreq != NULL)
+	if (hdr->ds_clp != NULL || hdr->dreq != NULL)
 		return false;
 	/* Otherwise, request attributes if and only if we don't hold
 	 * a delegation
@@ -4158,23 +4219,24 @@ bool nfs4_write_need_cache_consistency_data(const struct nfs_pgio_data *data)
 	return nfs4_have_delegation(hdr->inode, FMODE_READ) == 0;
 }
 
-static void nfs4_proc_write_setup(struct nfs_pgio_data *data, struct rpc_message *msg)
+static void nfs4_proc_write_setup(struct nfs_pgio_header *hdr,
+				  struct rpc_message *msg)
 {
-	struct nfs_server *server = NFS_SERVER(data->header->inode);
+	struct nfs_server *server = NFS_SERVER(hdr->inode);
 
-	if (!nfs4_write_need_cache_consistency_data(data)) {
-		data->args.bitmask = NULL;
-		data->res.fattr = NULL;
+	if (!nfs4_write_need_cache_consistency_data(hdr)) {
+		hdr->args.bitmask = NULL;
+		hdr->res.fattr = NULL;
 	} else
-		data->args.bitmask = server->cache_consistency_bitmask;
+		hdr->args.bitmask = server->cache_consistency_bitmask;
 
-	if (!data->pgio_done_cb)
-		data->pgio_done_cb = nfs4_write_done_cb;
-	data->res.server = server;
-	data->timestamp   = jiffies;
+	if (!hdr->pgio_done_cb)
+		hdr->pgio_done_cb = nfs4_write_done_cb;
+	hdr->res.server = server;
+	hdr->timestamp   = jiffies;
 
 	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_WRITE];
-	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 1);
+	nfs4_init_sequence(&hdr->args.seq_args, &hdr->res.seq_res, 1);
 }
 
 static void nfs4_proc_commit_rpc_prepare(struct rpc_task *task, struct nfs_commit_data *data)
@@ -4881,6 +4943,18 @@ nfs4_init_callback_netid(const struct nfs_client *clp, char *buf, size_t len)
 		return scnprintf(buf, len, "tcp");
 }
 
+static void nfs4_setclientid_done(struct rpc_task *task, void *calldata)
+{
+	struct nfs4_setclientid *sc = calldata;
+
+	if (task->tk_status == 0)
+		sc->sc_cred = get_rpccred(task->tk_rqstp->rq_cred);
+}
+
+static const struct rpc_call_ops nfs4_setclientid_ops = {
+	.rpc_call_done = nfs4_setclientid_done,
+};
+
 /**
  * nfs4_proc_setclientid - Negotiate client ID
  * @clp: state data structure
@@ -4907,6 +4981,14 @@ int nfs4_proc_setclientid(struct nfs_client *clp, u32 program,
 		.rpc_resp = res,
 		.rpc_cred = cred,
 	};
+	struct rpc_task *task;
+	struct rpc_task_setup task_setup_data = {
+		.rpc_client = clp->cl_rpcclient,
+		.rpc_message = &msg,
+		.callback_ops = &nfs4_setclientid_ops,
+		.callback_data = &setclientid,
+		.flags = RPC_TASK_TIMEOUT,
+	};
 	int status;
 
 	/* nfs_client_id4 */
@@ -4933,7 +5015,18 @@ int nfs4_proc_setclientid(struct nfs_client *clp, u32 program,
 	dprintk("NFS call  setclientid auth=%s, '%.*s'\n",
 		clp->cl_rpcclient->cl_auth->au_ops->au_name,
 		setclientid.sc_name_len, setclientid.sc_name);
-	status = rpc_call_sync(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
+	task = rpc_run_task(&task_setup_data);
+	if (IS_ERR(task)) {
+		status = PTR_ERR(task);
+		goto out;
+	}
+	status = task->tk_status;
+	if (setclientid.sc_cred) {
+		clp->cl_acceptor = rpcauth_stringify_acceptor(setclientid.sc_cred);
+		put_rpccred(setclientid.sc_cred);
+	}
+	rpc_put_task(task);
+out:
 	trace_nfs4_setclientid(clp, status);
 	dprintk("NFS reply setclientid: %d\n", status);
 	return status;
@@ -4975,6 +5068,9 @@ struct nfs4_delegreturndata {
 	unsigned long timestamp;
 	struct nfs_fattr fattr;
 	int rpc_status;
+	struct inode *inode;
+	bool roc;
+	u32 roc_barrier;
 };
 
 static void nfs4_delegreturn_done(struct rpc_task *task, void *calldata)
@@ -4988,7 +5084,6 @@ static void nfs4_delegreturn_done(struct rpc_task *task, void *calldata)
 	switch (task->tk_status) {
 	case 0:
 		renew_lease(data->res.server, data->timestamp);
-		break;
 	case -NFS4ERR_ADMIN_REVOKED:
 	case -NFS4ERR_DELEG_REVOKED:
 	case -NFS4ERR_BAD_STATEID:
@@ -4996,6 +5091,8 @@ static void nfs4_delegreturn_done(struct rpc_task *task, void *calldata)
 	case -NFS4ERR_STALE_STATEID:
 	case -NFS4ERR_EXPIRED:
 		task->tk_status = 0;
+		if (data->roc)
+			pnfs_roc_set_barrier(data->inode, data->roc_barrier);
 		break;
 	default:
 		if (nfs4_async_handle_error(task, data->res.server, NULL) ==
@@ -5009,6 +5106,10 @@ static void nfs4_delegreturn_done(struct rpc_task *task, void *calldata)
 
 static void nfs4_delegreturn_release(void *calldata)
 {
+	struct nfs4_delegreturndata *data = calldata;
+
+	if (data->roc)
+		pnfs_roc_release(data->inode);
 	kfree(calldata);
 }
 
@@ -5018,6 +5119,10 @@ static void nfs4_delegreturn_prepare(struct rpc_task *task, void *data)
 
 	d_data = (struct nfs4_delegreturndata *)data;
 
+	if (d_data->roc &&
+	    pnfs_roc_drain(d_data->inode, &d_data->roc_barrier, task))
+		return;
+
 	nfs4_setup_sequence(d_data->res.server,
 			&d_data->args.seq_args,
 			&d_data->res.seq_res,
@@ -5061,6 +5166,9 @@ static int _nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, co
 	nfs_fattr_init(data->res.fattr);
 	data->timestamp = jiffies;
 	data->rpc_status = 0;
+	data->inode = inode;
+	data->roc = list_empty(&NFS_I(inode)->open_files) ?
+		    pnfs_roc(inode) : false;
 
 	task_setup_data.callback_data = data;
 	msg.rpc_argp = &data->args;
@@ -5834,8 +5942,10 @@ struct nfs_release_lockowner_data {
 static void nfs4_release_lockowner_prepare(struct rpc_task *task, void *calldata)
 {
 	struct nfs_release_lockowner_data *data = calldata;
-	nfs40_setup_sequence(data->server,
-				&data->args.seq_args, &data->res.seq_res, task);
+	struct nfs_server *server = data->server;
+	nfs40_setup_sequence(server, &data->args.seq_args,
+				&data->res.seq_res, task);
+	data->args.lock_owner.clientid = server->nfs_client->cl_clientid;
 	data->timestamp = jiffies;
 }
 
@@ -5852,6 +5962,8 @@ static void nfs4_release_lockowner_done(struct rpc_task *task, void *calldata)
 		break;
 	case -NFS4ERR_STALE_CLIENTID:
 	case -NFS4ERR_EXPIRED:
+		nfs4_schedule_lease_recovery(server->nfs_client);
+		break;
 	case -NFS4ERR_LEASE_MOVED:
 	case -NFS4ERR_DELAY:
 		if (nfs4_async_handle_error(task, server, NULL) == -EAGAIN)
@@ -5872,7 +5984,8 @@ static const struct rpc_call_ops nfs4_release_lockowner_ops = {
 	.rpc_release = nfs4_release_lockowner_release,
 };
 
-static int nfs4_release_lockowner(struct nfs_server *server, struct nfs4_lock_state *lsp)
+static void
+nfs4_release_lockowner(struct nfs_server *server, struct nfs4_lock_state *lsp)
 {
 	struct nfs_release_lockowner_data *data;
 	struct rpc_message msg = {
@@ -5880,11 +5993,11 @@ static int nfs4_release_lockowner(struct nfs_server *server, struct nfs4_lock_st
 	};
 
 	if (server->nfs_client->cl_mvops->minor_version != 0)
-		return -EINVAL;
+		return;
 
 	data = kmalloc(sizeof(*data), GFP_NOFS);
 	if (!data)
-		return -ENOMEM;
+		return;
 	data->lsp = lsp;
 	data->server = server;
 	data->args.lock_owner.clientid = server->nfs_client->cl_clientid;
@@ -5895,7 +6008,6 @@ static int nfs4_release_lockowner(struct nfs_server *server, struct nfs4_lock_st
 	msg.rpc_resp = &data->res;
 	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 0);
 	rpc_call_async(server->client, &msg, 0, &nfs4_release_lockowner_ops, data);
-	return 0;
 }
 
 #define XATTR_NAME_NFSV4_ACL "system.nfs4_acl"
@@ -8182,7 +8294,8 @@ static int nfs41_free_stateid(struct nfs_server *server,
 	return ret;
 }
 
-static int nfs41_free_lock_state(struct nfs_server *server, struct nfs4_lock_state *lsp)
+static void
+nfs41_free_lock_state(struct nfs_server *server, struct nfs4_lock_state *lsp)
 {
 	struct rpc_task *task;
 	struct rpc_cred *cred = lsp->ls_state->owner->so_cred;
@@ -8190,9 +8303,8 @@ static int nfs41_free_lock_state(struct nfs_server *server, struct nfs4_lock_sta
 	task = _nfs41_free_stateid(server, &lsp->ls_stateid, cred, false);
 	nfs4_free_lock_state(server, lsp);
 	if (IS_ERR(task))
-		return PTR_ERR(task);
+		return;
 	rpc_put_task(task);
-	return 0;
 }
 
 static bool nfs41_match_stateid(const nfs4_stateid *s1,

fs/nfs/nfs4state.c

@@ -787,33 +787,36 @@ void nfs4_close_sync(struct nfs4_state *state, fmode_t fmode)
  * that is compatible with current->files
  */
 static struct nfs4_lock_state *
-__nfs4_find_lock_state(struct nfs4_state *state, fl_owner_t fl_owner, pid_t fl_pid, unsigned int type)
+__nfs4_find_lock_state(struct nfs4_state *state, fl_owner_t fl_owner)
 {
 	struct nfs4_lock_state *pos;
 	list_for_each_entry(pos, &state->lock_states, ls_locks) {
-		if (type != NFS4_ANY_LOCK_TYPE && pos->ls_owner.lo_type != type)
+		if (pos->ls_owner != fl_owner)
 			continue;
-		switch (pos->ls_owner.lo_type) {
-		case NFS4_POSIX_LOCK_TYPE:
-			if (pos->ls_owner.lo_u.posix_owner != fl_owner)
-				continue;
-			break;
-		case NFS4_FLOCK_LOCK_TYPE:
-			if (pos->ls_owner.lo_u.flock_owner != fl_pid)
-				continue;
-		}
 		atomic_inc(&pos->ls_count);
 		return pos;
 	}
 	return NULL;
 }
 
+static void
+free_lock_state_work(struct work_struct *work)
+{
+	struct nfs4_lock_state *lsp = container_of(work,
+					struct nfs4_lock_state, ls_release);
+	struct nfs4_state *state = lsp->ls_state;
+	struct nfs_server *server = state->owner->so_server;
+	struct nfs_client *clp = server->nfs_client;
+
+	clp->cl_mvops->free_lock_state(server, lsp);
+}
+
 /*
  * Return a compatible lock_state. If no initialized lock_state structure
  * exists, return an uninitialized one.
  *
  */
-static struct nfs4_lock_state *nfs4_alloc_lock_state(struct nfs4_state *state, fl_owner_t fl_owner, pid_t fl_pid, unsigned int type)
+static struct nfs4_lock_state *nfs4_alloc_lock_state(struct nfs4_state *state, fl_owner_t fl_owner)
 {
 	struct nfs4_lock_state *lsp;
 	struct nfs_server *server = state->owner->so_server;
@@ -824,21 +827,12 @@ static struct nfs4_lock_state *nfs4_alloc_lock_state(struct nfs4_state *state, f
 	nfs4_init_seqid_counter(&lsp->ls_seqid);
 	atomic_set(&lsp->ls_count, 1);
 	lsp->ls_state = state;
-	lsp->ls_owner.lo_type = type;
-	switch (lsp->ls_owner.lo_type) {
-	case NFS4_FLOCK_LOCK_TYPE:
-		lsp->ls_owner.lo_u.flock_owner = fl_pid;
-		break;
-	case NFS4_POSIX_LOCK_TYPE:
-		lsp->ls_owner.lo_u.posix_owner = fl_owner;
-		break;
-	default:
-		goto out_free;
-	}
+	lsp->ls_owner = fl_owner;
 	lsp->ls_seqid.owner_id = ida_simple_get(&server->lockowner_id, 0, 0, GFP_NOFS);
 	if (lsp->ls_seqid.owner_id < 0)
 		goto out_free;
 	INIT_LIST_HEAD(&lsp->ls_locks);
+	INIT_WORK(&lsp->ls_release, free_lock_state_work);
 	return lsp;
 out_free:
 	kfree(lsp);
@@ -857,13 +851,13 @@ void nfs4_free_lock_state(struct nfs_server *server, struct nfs4_lock_state *lsp
  * exists, return an uninitialized one.
  *
  */
-static struct nfs4_lock_state *nfs4_get_lock_state(struct nfs4_state *state, fl_owner_t owner, pid_t pid, unsigned int type)
+static struct nfs4_lock_state *nfs4_get_lock_state(struct nfs4_state *state, fl_owner_t owner)
 {
 	struct nfs4_lock_state *lsp, *new = NULL;
 	
 	for(;;) {
 		spin_lock(&state->state_lock);
-		lsp = __nfs4_find_lock_state(state, owner, pid, type);
+		lsp = __nfs4_find_lock_state(state, owner);
 		if (lsp != NULL)
 			break;
 		if (new != NULL) {
@@ -874,7 +868,7 @@ static struct nfs4_lock_state *nfs4_get_lock_state(struct nfs4_state *state, fl_
 			break;
 		}
 		spin_unlock(&state->state_lock);
-		new = nfs4_alloc_lock_state(state, owner, pid, type);
+		new = nfs4_alloc_lock_state(state, owner);
 		if (new == NULL)
 			return NULL;
 	}
@@ -902,13 +896,12 @@ void nfs4_put_lock_state(struct nfs4_lock_state *lsp)
 	if (list_empty(&state->lock_states))
 		clear_bit(LK_STATE_IN_USE, &state->flags);
 	spin_unlock(&state->state_lock);
-	server = state->owner->so_server;
-	if (test_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags)) {
-		struct nfs_client *clp = server->nfs_client;
-
-		clp->cl_mvops->free_lock_state(server, lsp);
-	} else
+	if (test_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags))
+		queue_work(nfsiod_workqueue, &lsp->ls_release);
+	else {
+		server = state->owner->so_server;
 		nfs4_free_lock_state(server, lsp);
+	}
 }
 
 static void nfs4_fl_copy_lock(struct file_lock *dst, struct file_lock *src)
@@ -935,13 +928,7 @@ int nfs4_set_lock_state(struct nfs4_state *state, struct file_lock *fl)
 
 	if (fl->fl_ops != NULL)
 		return 0;
-	if (fl->fl_flags & FL_POSIX)
-		lsp = nfs4_get_lock_state(state, fl->fl_owner, 0, NFS4_POSIX_LOCK_TYPE);
-	else if (fl->fl_flags & FL_FLOCK)
-		lsp = nfs4_get_lock_state(state, NULL, fl->fl_pid,
-				NFS4_FLOCK_LOCK_TYPE);
-	else
-		return -EINVAL;
+	lsp = nfs4_get_lock_state(state, fl->fl_owner);
 	if (lsp == NULL)
 		return -ENOMEM;
 	fl->fl_u.nfs4_fl.owner = lsp;
@@ -955,7 +942,6 @@ static int nfs4_copy_lock_stateid(nfs4_stateid *dst,
 {
 	struct nfs4_lock_state *lsp;
 	fl_owner_t fl_owner;
-	pid_t fl_pid;
 	int ret = -ENOENT;
 
 
@@ -966,9 +952,8 @@ static int nfs4_copy_lock_stateid(nfs4_stateid *dst,
 		goto out;
 
 	fl_owner = lockowner->l_owner;
-	fl_pid = lockowner->l_pid;
 	spin_lock(&state->state_lock);
-	lsp = __nfs4_find_lock_state(state, fl_owner, fl_pid, NFS4_ANY_LOCK_TYPE);
+	lsp = __nfs4_find_lock_state(state, fl_owner);
 	if (lsp && test_bit(NFS_LOCK_LOST, &lsp->ls_flags))
 		ret = -EIO;
 	else if (lsp != NULL && test_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags) != 0) {

fs/nfs/nfs4trace.h

@@ -932,11 +932,11 @@ DEFINE_NFS4_IDMAP_EVENT(nfs4_map_gid_to_group);
 
 DECLARE_EVENT_CLASS(nfs4_read_event,
 		TP_PROTO(
-			const struct nfs_pgio_data *data,
+			const struct nfs_pgio_header *hdr,
 			int error
 		),
 
-		TP_ARGS(data, error),
+		TP_ARGS(hdr, error),
 
 		TP_STRUCT__entry(
 			__field(dev_t, dev)
@@ -948,12 +948,12 @@ DECLARE_EVENT_CLASS(nfs4_read_event,
 		),
 
 		TP_fast_assign(
-			const struct inode *inode = data->header->inode;
+			const struct inode *inode = hdr->inode;
 			__entry->dev = inode->i_sb->s_dev;
 			__entry->fileid = NFS_FILEID(inode);
 			__entry->fhandle = nfs_fhandle_hash(NFS_FH(inode));
-			__entry->offset = data->args.offset;
-			__entry->count = data->args.count;
+			__entry->offset = hdr->args.offset;
+			__entry->count = hdr->args.count;
 			__entry->error = error;
 		),
 
@@ -972,10 +972,10 @@ DECLARE_EVENT_CLASS(nfs4_read_event,
 #define DEFINE_NFS4_READ_EVENT(name) \
 	DEFINE_EVENT(nfs4_read_event, name, \
 			TP_PROTO( \
-				const struct nfs_pgio_data *data, \
+				const struct nfs_pgio_header *hdr, \
 				int error \
 			), \
-			TP_ARGS(data, error))
+			TP_ARGS(hdr, error))
 DEFINE_NFS4_READ_EVENT(nfs4_read);
 #ifdef CONFIG_NFS_V4_1
 DEFINE_NFS4_READ_EVENT(nfs4_pnfs_read);
@@ -983,11 +983,11 @@ DEFINE_NFS4_READ_EVENT(nfs4_pnfs_read);
 
 DECLARE_EVENT_CLASS(nfs4_write_event,
 		TP_PROTO(
-			const struct nfs_pgio_data *data,
+			const struct nfs_pgio_header *hdr,
 			int error
 		),
 
-		TP_ARGS(data, error),
+		TP_ARGS(hdr, error),
 
 		TP_STRUCT__entry(
 			__field(dev_t, dev)
@@ -999,12 +999,12 @@ DECLARE_EVENT_CLASS(nfs4_write_event,
 		),
 
 		TP_fast_assign(
-			const struct inode *inode = data->header->inode;
+			const struct inode *inode = hdr->inode;
 			__entry->dev = inode->i_sb->s_dev;
 			__entry->fileid = NFS_FILEID(inode);
 			__entry->fhandle = nfs_fhandle_hash(NFS_FH(inode));
-			__entry->offset = data->args.offset;
-			__entry->count = data->args.count;
+			__entry->offset = hdr->args.offset;
+			__entry->count = hdr->args.count;
 			__entry->error = error;
 		),
 
@@ -1024,10 +1024,10 @@ DECLARE_EVENT_CLASS(nfs4_write_event,
 #define DEFINE_NFS4_WRITE_EVENT(name) \
 	DEFINE_EVENT(nfs4_write_event, name, \
 			TP_PROTO( \
-				const struct nfs_pgio_data *data, \
+				const struct nfs_pgio_header *hdr, \
 				int error \
 			), \
-			TP_ARGS(data, error))
+			TP_ARGS(hdr, error))
 DEFINE_NFS4_WRITE_EVENT(nfs4_write);
 #ifdef CONFIG_NFS_V4_1
 DEFINE_NFS4_WRITE_EVENT(nfs4_pnfs_write);

fs/nfs/nfs4xdr.c

@@ -7092,7 +7092,7 @@ static int nfs4_xdr_dec_reclaim_complete(struct rpc_rqst *rqstp,
 	if (!status)
 		status = decode_sequence(xdr, &res->seq_res, rqstp);
 	if (!status)
-		status = decode_reclaim_complete(xdr, (void *)NULL);
+		status = decode_reclaim_complete(xdr, NULL);
 	return status;
 }
 

fs/nfs/objlayout/objio_osd.c

@@ -439,22 +439,21 @@ static void _read_done(struct ore_io_state *ios, void *private)
 	objlayout_read_done(&objios->oir, status, objios->sync);
 }
 
-int objio_read_pagelist(struct nfs_pgio_data *rdata)
+int objio_read_pagelist(struct nfs_pgio_header *hdr)
 {
-	struct nfs_pgio_header *hdr = rdata->header;
 	struct objio_state *objios;
 	int ret;
 
 	ret = objio_alloc_io_state(NFS_I(hdr->inode)->layout, true,
-			hdr->lseg, rdata->args.pages, rdata->args.pgbase,
-			rdata->args.offset, rdata->args.count, rdata,
+			hdr->lseg, hdr->args.pages, hdr->args.pgbase,
+			hdr->args.offset, hdr->args.count, hdr,
 			GFP_KERNEL, &objios);
 	if (unlikely(ret))
 		return ret;
 
 	objios->ios->done = _read_done;
 	dprintk("%s: offset=0x%llx length=0x%x\n", __func__,
-		rdata->args.offset, rdata->args.count);
+		hdr->args.offset, hdr->args.count);
 	ret = ore_read(objios->ios);
 	if (unlikely(ret))
 		objio_free_result(&objios->oir);
@@ -487,11 +486,11 @@ static void _write_done(struct ore_io_state *ios, void *private)
 static struct page *__r4w_get_page(void *priv, u64 offset, bool *uptodate)
 {
 	struct objio_state *objios = priv;
-	struct nfs_pgio_data *wdata = objios->oir.rpcdata;
-	struct address_space *mapping = wdata->header->inode->i_mapping;
+	struct nfs_pgio_header *hdr = objios->oir.rpcdata;
+	struct address_space *mapping = hdr->inode->i_mapping;
 	pgoff_t index = offset / PAGE_SIZE;
 	struct page *page;
-	loff_t i_size = i_size_read(wdata->header->inode);
+	loff_t i_size = i_size_read(hdr->inode);
 
 	if (offset >= i_size) {
 		*uptodate = true;
@@ -531,15 +530,14 @@ static const struct _ore_r4w_op _r4w_op = {
 	.put_page = &__r4w_put_page,
 };
 
-int objio_write_pagelist(struct nfs_pgio_data *wdata, int how)
+int objio_write_pagelist(struct nfs_pgio_header *hdr, int how)
 {
-	struct nfs_pgio_header *hdr = wdata->header;
 	struct objio_state *objios;
 	int ret;
 
 	ret = objio_alloc_io_state(NFS_I(hdr->inode)->layout, false,
-			hdr->lseg, wdata->args.pages, wdata->args.pgbase,
-			wdata->args.offset, wdata->args.count, wdata, GFP_NOFS,
+			hdr->lseg, hdr->args.pages, hdr->args.pgbase,
+			hdr->args.offset, hdr->args.count, hdr, GFP_NOFS,
 			&objios);
 	if (unlikely(ret))
 		return ret;
@@ -551,7 +549,7 @@ int objio_write_pagelist(struct nfs_pgio_data *wdata, int how)
 		objios->ios->done = _write_done;
 
 	dprintk("%s: offset=0x%llx length=0x%x\n", __func__,
-		wdata->args.offset, wdata->args.count);
+		hdr->args.offset, hdr->args.count);
 	ret = ore_write(objios->ios);
 	if (unlikely(ret)) {
 		objio_free_result(&objios->oir);

fs/nfs/objlayout/objlayout.c

@@ -229,36 +229,36 @@ objlayout_io_set_result(struct objlayout_io_res *oir, unsigned index,
 static void _rpc_read_complete(struct work_struct *work)
 {
 	struct rpc_task *task;
-	struct nfs_pgio_data *rdata;
+	struct nfs_pgio_header *hdr;
 
 	dprintk("%s enter\n", __func__);
 	task = container_of(work, struct rpc_task, u.tk_work);
-	rdata = container_of(task, struct nfs_pgio_data, task);
+	hdr = container_of(task, struct nfs_pgio_header, task);
 
-	pnfs_ld_read_done(rdata);
+	pnfs_ld_read_done(hdr);
 }
 
 void
 objlayout_read_done(struct objlayout_io_res *oir, ssize_t status, bool sync)
 {
-	struct nfs_pgio_data *rdata = oir->rpcdata;
+	struct nfs_pgio_header *hdr = oir->rpcdata;
 
-	oir->status = rdata->task.tk_status = status;
+	oir->status = hdr->task.tk_status = status;
 	if (status >= 0)
-		rdata->res.count = status;
+		hdr->res.count = status;
 	else
-		rdata->header->pnfs_error = status;
+		hdr->pnfs_error = status;
 	objlayout_iodone(oir);
 	/* must not use oir after this point */
 
 	dprintk("%s: Return status=%zd eof=%d sync=%d\n", __func__,
-		status, rdata->res.eof, sync);
+		status, hdr->res.eof, sync);
 
 	if (sync)
-		pnfs_ld_read_done(rdata);
+		pnfs_ld_read_done(hdr);
 	else {
-		INIT_WORK(&rdata->task.u.tk_work, _rpc_read_complete);
-		schedule_work(&rdata->task.u.tk_work);
+		INIT_WORK(&hdr->task.u.tk_work, _rpc_read_complete);
+		schedule_work(&hdr->task.u.tk_work);
 	}
 }
 
@@ -266,12 +266,11 @@ objlayout_read_done(struct objlayout_io_res *oir, ssize_t status, bool sync)
  * Perform sync or async reads.
  */
 enum pnfs_try_status
-objlayout_read_pagelist(struct nfs_pgio_data *rdata)
+objlayout_read_pagelist(struct nfs_pgio_header *hdr)
 {
-	struct nfs_pgio_header *hdr = rdata->header;
 	struct inode *inode = hdr->inode;
-	loff_t offset = rdata->args.offset;
-	size_t count = rdata->args.count;
+	loff_t offset = hdr->args.offset;
+	size_t count = hdr->args.count;
 	int err;
 	loff_t eof;
 
@@ -279,23 +278,23 @@ objlayout_read_pagelist(struct nfs_pgio_data *rdata)
 	if (unlikely(offset + count > eof)) {
 		if (offset >= eof) {
 			err = 0;
-			rdata->res.count = 0;
-			rdata->res.eof = 1;
+			hdr->res.count = 0;
+			hdr->res.eof = 1;
 			/*FIXME: do we need to call pnfs_ld_read_done() */
 			goto out;
 		}
 		count = eof - offset;
 	}
 
-	rdata->res.eof = (offset + count) >= eof;
-	_fix_verify_io_params(hdr->lseg, &rdata->args.pages,
-			      &rdata->args.pgbase,
-			      rdata->args.offset, rdata->args.count);
+	hdr->res.eof = (offset + count) >= eof;
+	_fix_verify_io_params(hdr->lseg, &hdr->args.pages,
+			      &hdr->args.pgbase,
+			      hdr->args.offset, hdr->args.count);
 
 	dprintk("%s: inode(%lx) offset 0x%llx count 0x%Zx eof=%d\n",
-		__func__, inode->i_ino, offset, count, rdata->res.eof);
+		__func__, inode->i_ino, offset, count, hdr->res.eof);
 
-	err = objio_read_pagelist(rdata);
+	err = objio_read_pagelist(hdr);
  out:
 	if (unlikely(err)) {
 		hdr->pnfs_error = err;
@@ -312,38 +311,38 @@ objlayout_read_pagelist(struct nfs_pgio_data *rdata)
 static void _rpc_write_complete(struct work_struct *work)
 {
 	struct rpc_task *task;
-	struct nfs_pgio_data *wdata;
+	struct nfs_pgio_header *hdr;
 
 	dprintk("%s enter\n", __func__);
 	task = container_of(work, struct rpc_task, u.tk_work);
-	wdata = container_of(task, struct nfs_pgio_data, task);
+	hdr = container_of(task, struct nfs_pgio_header, task);
 
-	pnfs_ld_write_done(wdata);
+	pnfs_ld_write_done(hdr);
 }
 
 void
 objlayout_write_done(struct objlayout_io_res *oir, ssize_t status, bool sync)
 {
-	struct nfs_pgio_data *wdata = oir->rpcdata;
+	struct nfs_pgio_header *hdr = oir->rpcdata;
 
-	oir->status = wdata->task.tk_status = status;
+	oir->status = hdr->task.tk_status = status;
 	if (status >= 0) {
-		wdata->res.count = status;
-		wdata->verf.committed = oir->committed;
+		hdr->res.count = status;
+		hdr->verf.committed = oir->committed;
 	} else {
-		wdata->header->pnfs_error = status;
+		hdr->pnfs_error = status;
 	}
 	objlayout_iodone(oir);
 	/* must not use oir after this point */
 
 	dprintk("%s: Return status %zd committed %d sync=%d\n", __func__,
-		status, wdata->verf.committed, sync);
+		status, hdr->verf.committed, sync);
 
 	if (sync)
-		pnfs_ld_write_done(wdata);
+		pnfs_ld_write_done(hdr);
 	else {
-		INIT_WORK(&wdata->task.u.tk_work, _rpc_write_complete);
-		schedule_work(&wdata->task.u.tk_work);
+		INIT_WORK(&hdr->task.u.tk_work, _rpc_write_complete);
+		schedule_work(&hdr->task.u.tk_work);
 	}
 }
 
@@ -351,17 +350,15 @@ objlayout_write_done(struct objlayout_io_res *oir, ssize_t status, bool sync)
  * Perform sync or async writes.
  */
 enum pnfs_try_status
-objlayout_write_pagelist(struct nfs_pgio_data *wdata,
-			 int how)
+objlayout_write_pagelist(struct nfs_pgio_header *hdr, int how)
 {
-	struct nfs_pgio_header *hdr = wdata->header;
 	int err;
 
-	_fix_verify_io_params(hdr->lseg, &wdata->args.pages,
-			      &wdata->args.pgbase,
-			      wdata->args.offset, wdata->args.count);
+	_fix_verify_io_params(hdr->lseg, &hdr->args.pages,
+			      &hdr->args.pgbase,
+			      hdr->args.offset, hdr->args.count);
 
-	err = objio_write_pagelist(wdata, how);
+	err = objio_write_pagelist(hdr, how);
 	if (unlikely(err)) {
 		hdr->pnfs_error = err;
 		dprintk("%s: Returned Error %d\n", __func__, err);

fs/nfs/objlayout/objlayout.h

@@ -119,8 +119,8 @@ extern void objio_free_lseg(struct pnfs_layout_segment *lseg);
  */
 extern void objio_free_result(struct objlayout_io_res *oir);
 
-extern int objio_read_pagelist(struct nfs_pgio_data *rdata);
-extern int objio_write_pagelist(struct nfs_pgio_data *wdata, int how);
+extern int objio_read_pagelist(struct nfs_pgio_header *rdata);
+extern int objio_write_pagelist(struct nfs_pgio_header *wdata, int how);
 
 /*
  * callback API
@@ -168,10 +168,10 @@ extern struct pnfs_layout_segment *objlayout_alloc_lseg(
 extern void objlayout_free_lseg(struct pnfs_layout_segment *);
 
 extern enum pnfs_try_status objlayout_read_pagelist(
-	struct nfs_pgio_data *);
+	struct nfs_pgio_header *);
 
 extern enum pnfs_try_status objlayout_write_pagelist(
-	struct nfs_pgio_data *,
+	struct nfs_pgio_header *,
 	int how);
 
 extern void objlayout_encode_layoutcommit(

fs/nfs/pagelist.c

@@ -141,16 +141,24 @@ nfs_iocounter_wait(struct nfs_io_counter *c)
  * @req - request in group that is to be locked
  *
  * this lock must be held if modifying the page group list
+ *
+ * returns result from wait_on_bit_lock: 0 on success, < 0 on error
  */
-void
-nfs_page_group_lock(struct nfs_page *req)
+int
+nfs_page_group_lock(struct nfs_page *req, bool wait)
 {
 	struct nfs_page *head = req->wb_head;
+	int ret;
 
 	WARN_ON_ONCE(head != head->wb_head);
 
-	wait_on_bit_lock(&head->wb_flags, PG_HEADLOCK,
+	do {
+		ret = wait_on_bit_lock(&head->wb_flags, PG_HEADLOCK,
 			TASK_UNINTERRUPTIBLE);
+	} while (wait && ret != 0);
+
+	WARN_ON_ONCE(ret > 0);
+	return ret;
 }
 
 /*
@@ -211,7 +219,7 @@ bool nfs_page_group_sync_on_bit(struct nfs_page *req, unsigned int bit)
 {
 	bool ret;
 
-	nfs_page_group_lock(req);
+	nfs_page_group_lock(req, true);
 	ret = nfs_page_group_sync_on_bit_locked(req, bit);
 	nfs_page_group_unlock(req);
 
@@ -454,123 +462,72 @@ size_t nfs_generic_pg_test(struct nfs_pageio_descriptor *desc,
 }
 EXPORT_SYMBOL_GPL(nfs_generic_pg_test);
 
-static inline struct nfs_rw_header *NFS_RW_HEADER(struct nfs_pgio_header *hdr)
-{
-	return container_of(hdr, struct nfs_rw_header, header);
-}
-
-/**
- * nfs_rw_header_alloc - Allocate a header for a read or write
- * @ops: Read or write function vector
- */
-struct nfs_rw_header *nfs_rw_header_alloc(const struct nfs_rw_ops *ops)
+struct nfs_pgio_header *nfs_pgio_header_alloc(const struct nfs_rw_ops *ops)
 {
-	struct nfs_rw_header *header = ops->rw_alloc_header();
-
-	if (header) {
-		struct nfs_pgio_header *hdr = &header->header;
+	struct nfs_pgio_header *hdr = ops->rw_alloc_header();
 
+	if (hdr) {
 		INIT_LIST_HEAD(&hdr->pages);
 		spin_lock_init(&hdr->lock);
-		atomic_set(&hdr->refcnt, 0);
 		hdr->rw_ops = ops;
 	}
-	return header;
+	return hdr;
 }
-EXPORT_SYMBOL_GPL(nfs_rw_header_alloc);
+EXPORT_SYMBOL_GPL(nfs_pgio_header_alloc);
 
 /*
- * nfs_rw_header_free - Free a read or write header
+ * nfs_pgio_header_free - Free a read or write header
  * @hdr: The header to free
  */
-void nfs_rw_header_free(struct nfs_pgio_header *hdr)
+void nfs_pgio_header_free(struct nfs_pgio_header *hdr)
 {
-	hdr->rw_ops->rw_free_header(NFS_RW_HEADER(hdr));
+	hdr->rw_ops->rw_free_header(hdr);
 }
-EXPORT_SYMBOL_GPL(nfs_rw_header_free);
+EXPORT_SYMBOL_GPL(nfs_pgio_header_free);
 
 /**
- * nfs_pgio_data_alloc - Allocate pageio data
- * @hdr: The header making a request
- * @pagecount: Number of pages to create
- */
-static struct nfs_pgio_data *nfs_pgio_data_alloc(struct nfs_pgio_header *hdr,
-						 unsigned int pagecount)
-{
-	struct nfs_pgio_data *data, *prealloc;
-
-	prealloc = &NFS_RW_HEADER(hdr)->rpc_data;
-	if (prealloc->header == NULL)
-		data = prealloc;
-	else
-		data = kzalloc(sizeof(*data), GFP_KERNEL);
-	if (!data)
-		goto out;
-
-	if (nfs_pgarray_set(&data->pages, pagecount)) {
-		data->header = hdr;
-		atomic_inc(&hdr->refcnt);
-	} else {
-		if (data != prealloc)
-			kfree(data);
-		data = NULL;
-	}
-out:
-	return data;
-}
-
-/**
- * nfs_pgio_data_release - Properly free pageio data
- * @data: The data to release
+ * nfs_pgio_data_destroy - make @hdr suitable for reuse
+ *
+ * Frees memory and releases refs from nfs_generic_pgio, so that it may
+ * be called again.
+ *
+ * @hdr: A header that has had nfs_generic_pgio called
  */
-void nfs_pgio_data_release(struct nfs_pgio_data *data)
+void nfs_pgio_data_destroy(struct nfs_pgio_header *hdr)
 {
-	struct nfs_pgio_header *hdr = data->header;
-	struct nfs_rw_header *pageio_header = NFS_RW_HEADER(hdr);
-
-	put_nfs_open_context(data->args.context);
-	if (data->pages.pagevec != data->pages.page_array)
-		kfree(data->pages.pagevec);
-	if (data == &pageio_header->rpc_data) {
-		data->header = NULL;
-		data = NULL;
-	}
-	if (atomic_dec_and_test(&hdr->refcnt))
-		hdr->completion_ops->completion(hdr);
-	/* Note: we only free the rpc_task after callbacks are done.
-	 * See the comment in rpc_free_task() for why
-	 */
-	kfree(data);
+	put_nfs_open_context(hdr->args.context);
+	if (hdr->page_array.pagevec != hdr->page_array.page_array)
+		kfree(hdr->page_array.pagevec);
 }
-EXPORT_SYMBOL_GPL(nfs_pgio_data_release);
+EXPORT_SYMBOL_GPL(nfs_pgio_data_destroy);
 
 /**
  * nfs_pgio_rpcsetup - Set up arguments for a pageio call
- * @data: The pageio data
+ * @hdr: The pageio hdr
  * @count: Number of bytes to read
  * @offset: Initial offset
  * @how: How to commit data (writes only)
  * @cinfo: Commit information for the call (writes only)
  */
-static void nfs_pgio_rpcsetup(struct nfs_pgio_data *data,
+static void nfs_pgio_rpcsetup(struct nfs_pgio_header *hdr,
 			      unsigned int count, unsigned int offset,
 			      int how, struct nfs_commit_info *cinfo)
 {
-	struct nfs_page *req = data->header->req;
+	struct nfs_page *req = hdr->req;
 
 	/* Set up the RPC argument and reply structs
-	 * NB: take care not to mess about with data->commit et al. */
+	 * NB: take care not to mess about with hdr->commit et al. */
 
-	data->args.fh     = NFS_FH(data->header->inode);
-	data->args.offset = req_offset(req) + offset;
+	hdr->args.fh     = NFS_FH(hdr->inode);
+	hdr->args.offset = req_offset(req) + offset;
 	/* pnfs_set_layoutcommit needs this */
-	data->mds_offset = data->args.offset;
-	data->args.pgbase = req->wb_pgbase + offset;
-	data->args.pages  = data->pages.pagevec;
-	data->args.count  = count;
-	data->args.context = get_nfs_open_context(req->wb_context);
-	data->args.lock_context = req->wb_lock_context;
-	data->args.stable  = NFS_UNSTABLE;
+	hdr->mds_offset = hdr->args.offset;
+	hdr->args.pgbase = req->wb_pgbase + offset;
+	hdr->args.pages  = hdr->page_array.pagevec;
+	hdr->args.count  = count;
+	hdr->args.context = get_nfs_open_context(req->wb_context);
+	hdr->args.lock_context = req->wb_lock_context;
+	hdr->args.stable  = NFS_UNSTABLE;
 	switch (how & (FLUSH_STABLE | FLUSH_COND_STABLE)) {
 	case 0:
 		break;
@@ -578,59 +535,59 @@ static void nfs_pgio_rpcsetup(struct nfs_pgio_data *data,
 		if (nfs_reqs_to_commit(cinfo))
 			break;
 	default:
-		data->args.stable = NFS_FILE_SYNC;
+		hdr->args.stable = NFS_FILE_SYNC;
 	}
 
-	data->res.fattr   = &data->fattr;
-	data->res.count   = count;
-	data->res.eof     = 0;
-	data->res.verf    = &data->verf;
-	nfs_fattr_init(&data->fattr);
+	hdr->res.fattr   = &hdr->fattr;
+	hdr->res.count   = count;
+	hdr->res.eof     = 0;
+	hdr->res.verf    = &hdr->verf;
+	nfs_fattr_init(&hdr->fattr);
 }
 
 /**
- * nfs_pgio_prepare - Prepare pageio data to go over the wire
+ * nfs_pgio_prepare - Prepare pageio hdr to go over the wire
  * @task: The current task
- * @calldata: pageio data to prepare
+ * @calldata: pageio header to prepare
  */
 static void nfs_pgio_prepare(struct rpc_task *task, void *calldata)
 {
-	struct nfs_pgio_data *data = calldata;
+	struct nfs_pgio_header *hdr = calldata;
 	int err;
-	err = NFS_PROTO(data->header->inode)->pgio_rpc_prepare(task, data);
+	err = NFS_PROTO(hdr->inode)->pgio_rpc_prepare(task, hdr);
 	if (err)
 		rpc_exit(task, err);
 }
 
-int nfs_initiate_pgio(struct rpc_clnt *clnt, struct nfs_pgio_data *data,
+int nfs_initiate_pgio(struct rpc_clnt *clnt, struct nfs_pgio_header *hdr,
 		      const struct rpc_call_ops *call_ops, int how, int flags)
 {
 	struct rpc_task *task;
 	struct rpc_message msg = {
-		.rpc_argp = &data->args,
-		.rpc_resp = &data->res,
-		.rpc_cred = data->header->cred,
+		.rpc_argp = &hdr->args,
+		.rpc_resp = &hdr->res,
+		.rpc_cred = hdr->cred,
 	};
 	struct rpc_task_setup task_setup_data = {
 		.rpc_client = clnt,
-		.task = &data->task,
+		.task = &hdr->task,
 		.rpc_message = &msg,
 		.callback_ops = call_ops,
-		.callback_data = data,
+		.callback_data = hdr,
 		.workqueue = nfsiod_workqueue,
 		.flags = RPC_TASK_ASYNC | flags,
 	};
 	int ret = 0;
 
-	data->header->rw_ops->rw_initiate(data, &msg, &task_setup_data, how);
+	hdr->rw_ops->rw_initiate(hdr, &msg, &task_setup_data, how);
 
 	dprintk("NFS: %5u initiated pgio call "
 		"(req %s/%llu, %u bytes @ offset %llu)\n",
-		data->task.tk_pid,
-		data->header->inode->i_sb->s_id,
-		(unsigned long long)NFS_FILEID(data->header->inode),
-		data->args.count,
-		(unsigned long long)data->args.offset);
+		hdr->task.tk_pid,
+		hdr->inode->i_sb->s_id,
+		(unsigned long long)NFS_FILEID(hdr->inode),
+		hdr->args.count,
+		(unsigned long long)hdr->args.offset);
 
 	task = rpc_run_task(&task_setup_data);
 	if (IS_ERR(task)) {
@@ -657,22 +614,23 @@ static int nfs_pgio_error(struct nfs_pageio_descriptor *desc,
 			  struct nfs_pgio_header *hdr)
 {
 	set_bit(NFS_IOHDR_REDO, &hdr->flags);
-	nfs_pgio_data_release(hdr->data);
-	hdr->data = NULL;
+	nfs_pgio_data_destroy(hdr);
+	hdr->completion_ops->completion(hdr);
 	desc->pg_completion_ops->error_cleanup(&desc->pg_list);
 	return -ENOMEM;
 }
 
 /**
  * nfs_pgio_release - Release pageio data
- * @calldata: The pageio data to release
+ * @calldata: The pageio header to release
  */
 static void nfs_pgio_release(void *calldata)
 {
-	struct nfs_pgio_data *data = calldata;
-	if (data->header->rw_ops->rw_release)
-		data->header->rw_ops->rw_release(data);
-	nfs_pgio_data_release(data);
+	struct nfs_pgio_header *hdr = calldata;
+	if (hdr->rw_ops->rw_release)
+		hdr->rw_ops->rw_release(hdr);
+	nfs_pgio_data_destroy(hdr);
+	hdr->completion_ops->completion(hdr);
 }
 
 /**
@@ -713,22 +671,22 @@ EXPORT_SYMBOL_GPL(nfs_pageio_init);
 /**
  * nfs_pgio_result - Basic pageio error handling
  * @task: The task that ran
- * @calldata: Pageio data to check
+ * @calldata: Pageio header to check
  */
 static void nfs_pgio_result(struct rpc_task *task, void *calldata)
 {
-	struct nfs_pgio_data *data = calldata;
-	struct inode *inode = data->header->inode;
+	struct nfs_pgio_header *hdr = calldata;
+	struct inode *inode = hdr->inode;
 
 	dprintk("NFS: %s: %5u, (status %d)\n", __func__,
 		task->tk_pid, task->tk_status);
 
-	if (data->header->rw_ops->rw_done(task, data, inode) != 0)
+	if (hdr->rw_ops->rw_done(task, hdr, inode) != 0)
 		return;
 	if (task->tk_status < 0)
-		nfs_set_pgio_error(data->header, task->tk_status, data->args.offset);
+		nfs_set_pgio_error(hdr, task->tk_status, hdr->args.offset);
 	else
-		data->header->rw_ops->rw_result(task, data);
+		hdr->rw_ops->rw_result(task, hdr);
 }
 
 /*
@@ -744,17 +702,16 @@ int nfs_generic_pgio(struct nfs_pageio_descriptor *desc,
 {
 	struct nfs_page		*req;
 	struct page		**pages;
-	struct nfs_pgio_data	*data;
 	struct list_head *head = &desc->pg_list;
 	struct nfs_commit_info cinfo;
+	unsigned int pagecount;
 
-	data = nfs_pgio_data_alloc(hdr, nfs_page_array_len(desc->pg_base,
-							   desc->pg_count));
-	if (!data)
+	pagecount = nfs_page_array_len(desc->pg_base, desc->pg_count);
+	if (!nfs_pgarray_set(&hdr->page_array, pagecount))
 		return nfs_pgio_error(desc, hdr);
 
 	nfs_init_cinfo(&cinfo, desc->pg_inode, desc->pg_dreq);
-	pages = data->pages.pagevec;
+	pages = hdr->page_array.pagevec;
 	while (!list_empty(head)) {
 		req = nfs_list_entry(head->next);
 		nfs_list_remove_request(req);
@@ -767,8 +724,7 @@ int nfs_generic_pgio(struct nfs_pageio_descriptor *desc,
 		desc->pg_ioflags &= ~FLUSH_COND_STABLE;
 
 	/* Set up the argument struct */
-	nfs_pgio_rpcsetup(data, desc->pg_count, 0, desc->pg_ioflags, &cinfo);
-	hdr->data = data;
+	nfs_pgio_rpcsetup(hdr, desc->pg_count, 0, desc->pg_ioflags, &cinfo);
 	desc->pg_rpc_callops = &nfs_pgio_common_ops;
 	return 0;
 }
@@ -776,25 +732,20 @@ EXPORT_SYMBOL_GPL(nfs_generic_pgio);
 
 static int nfs_generic_pg_pgios(struct nfs_pageio_descriptor *desc)
 {
-	struct nfs_rw_header *rw_hdr;
 	struct nfs_pgio_header *hdr;
 	int ret;
 
-	rw_hdr = nfs_rw_header_alloc(desc->pg_rw_ops);
-	if (!rw_hdr) {
+	hdr = nfs_pgio_header_alloc(desc->pg_rw_ops);
+	if (!hdr) {
 		desc->pg_completion_ops->error_cleanup(&desc->pg_list);
 		return -ENOMEM;
 	}
-	hdr = &rw_hdr->header;
-	nfs_pgheader_init(desc, hdr, nfs_rw_header_free);
-	atomic_inc(&hdr->refcnt);
+	nfs_pgheader_init(desc, hdr, nfs_pgio_header_free);
 	ret = nfs_generic_pgio(desc, hdr);
 	if (ret == 0)
 		ret = nfs_initiate_pgio(NFS_CLIENT(hdr->inode),
-					hdr->data, desc->pg_rpc_callops,
+					hdr, desc->pg_rpc_callops,
 					desc->pg_ioflags, 0);
-	if (atomic_dec_and_test(&hdr->refcnt))
-		hdr->completion_ops->completion(hdr);
 	return ret;
 }
 
@@ -907,8 +858,13 @@ static int __nfs_pageio_add_request(struct nfs_pageio_descriptor *desc,
 	struct nfs_page *subreq;
 	unsigned int bytes_left = 0;
 	unsigned int offset, pgbase;
+	int ret;
 
-	nfs_page_group_lock(req);
+	ret = nfs_page_group_lock(req, false);
+	if (ret < 0) {
+		desc->pg_error = ret;
+		return 0;
+	}
 
 	subreq = req;
 	bytes_left = subreq->wb_bytes;
@@ -930,7 +886,11 @@ static int __nfs_pageio_add_request(struct nfs_pageio_descriptor *desc,
 			if (desc->pg_recoalesce)
 				return 0;
 			/* retry add_request for this subreq */
-			nfs_page_group_lock(req);
+			ret = nfs_page_group_lock(req, false);
+			if (ret < 0) {
+				desc->pg_error = ret;
+				return 0;
+			}
 			continue;
 		}
 
@@ -1005,7 +965,38 @@ int nfs_pageio_add_request(struct nfs_pageio_descriptor *desc,
 	} while (ret);
 	return ret;
 }
-EXPORT_SYMBOL_GPL(nfs_pageio_add_request);
+
+/*
+ * nfs_pageio_resend - Transfer requests to new descriptor and resend
+ * @hdr - the pgio header to move request from
+ * @desc - the pageio descriptor to add requests to
+ *
+ * Try to move each request (nfs_page) from @hdr to @desc then attempt
+ * to send them.
+ *
+ * Returns 0 on success and < 0 on error.
+ */
+int nfs_pageio_resend(struct nfs_pageio_descriptor *desc,
+		      struct nfs_pgio_header *hdr)
+{
+	LIST_HEAD(failed);
+
+	desc->pg_dreq = hdr->dreq;
+	while (!list_empty(&hdr->pages)) {
+		struct nfs_page *req = nfs_list_entry(hdr->pages.next);
+
+		nfs_list_remove_request(req);
+		if (!nfs_pageio_add_request(desc, req))
+			nfs_list_add_request(req, &failed);
+	}
+	nfs_pageio_complete(desc);
+	if (!list_empty(&failed)) {
+		list_move(&failed, &hdr->pages);
+		return -EIO;
+	}
+	return 0;
+}
+EXPORT_SYMBOL_GPL(nfs_pageio_resend);
 
 /**
  * nfs_pageio_complete - Complete I/O on an nfs_pageio_descriptor
@@ -1021,7 +1012,6 @@ void nfs_pageio_complete(struct nfs_pageio_descriptor *desc)
 			break;
 	}
 }
-EXPORT_SYMBOL_GPL(nfs_pageio_complete);
 
 /**
  * nfs_pageio_cond_complete - Conditional I/O completion

fs/nfs/pnfs.c

@@ -361,6 +361,23 @@ pnfs_put_lseg(struct pnfs_layout_segment *lseg)
 }
 EXPORT_SYMBOL_GPL(pnfs_put_lseg);
 
+static void pnfs_put_lseg_async_work(struct work_struct *work)
+{
+	struct pnfs_layout_segment *lseg;
+
+	lseg = container_of(work, struct pnfs_layout_segment, pls_work);
+
+	pnfs_put_lseg(lseg);
+}
+
+void
+pnfs_put_lseg_async(struct pnfs_layout_segment *lseg)
+{
+	INIT_WORK(&lseg->pls_work, pnfs_put_lseg_async_work);
+	schedule_work(&lseg->pls_work);
+}
+EXPORT_SYMBOL_GPL(pnfs_put_lseg_async);
+
 static u64
 end_offset(u64 start, u64 len)
 {
@@ -1470,41 +1487,19 @@ pnfs_generic_pg_test(struct nfs_pageio_descriptor *pgio, struct nfs_page *prev,
 }
 EXPORT_SYMBOL_GPL(pnfs_generic_pg_test);
 
-int pnfs_write_done_resend_to_mds(struct inode *inode,
-				struct list_head *head,
-				const struct nfs_pgio_completion_ops *compl_ops,
-				struct nfs_direct_req *dreq)
+int pnfs_write_done_resend_to_mds(struct nfs_pgio_header *hdr)
 {
 	struct nfs_pageio_descriptor pgio;
-	LIST_HEAD(failed);
 
 	/* Resend all requests through the MDS */
-	nfs_pageio_init_write(&pgio, inode, FLUSH_STABLE, true, compl_ops);
-	pgio.pg_dreq = dreq;
-	while (!list_empty(head)) {
-		struct nfs_page *req = nfs_list_entry(head->next);
-
-		nfs_list_remove_request(req);
-		if (!nfs_pageio_add_request(&pgio, req))
-			nfs_list_add_request(req, &failed);
-	}
-	nfs_pageio_complete(&pgio);
-
-	if (!list_empty(&failed)) {
-		/* For some reason our attempt to resend pages. Mark the
-		 * overall send request as having failed, and let
-		 * nfs_writeback_release_full deal with the error.
-		 */
-		list_move(&failed, head);
-		return -EIO;
-	}
-	return 0;
+	nfs_pageio_init_write(&pgio, hdr->inode, FLUSH_STABLE, true,
+			      hdr->completion_ops);
+	return nfs_pageio_resend(&pgio, hdr);
 }
 EXPORT_SYMBOL_GPL(pnfs_write_done_resend_to_mds);
 
-static void pnfs_ld_handle_write_error(struct nfs_pgio_data *data)
+static void pnfs_ld_handle_write_error(struct nfs_pgio_header *hdr)
 {
-	struct nfs_pgio_header *hdr = data->header;
 
 	dprintk("pnfs write error = %d\n", hdr->pnfs_error);
 	if (NFS_SERVER(hdr->inode)->pnfs_curr_ld->flags &
@@ -1512,50 +1507,42 @@ static void pnfs_ld_handle_write_error(struct nfs_pgio_data *data)
 		pnfs_return_layout(hdr->inode);
 	}
 	if (!test_and_set_bit(NFS_IOHDR_REDO, &hdr->flags))
-		data->task.tk_status = pnfs_write_done_resend_to_mds(hdr->inode,
-							&hdr->pages,
-							hdr->completion_ops,
-							hdr->dreq);
+		hdr->task.tk_status = pnfs_write_done_resend_to_mds(hdr);
 }
 
 /*
  * Called by non rpc-based layout drivers
  */
-void pnfs_ld_write_done(struct nfs_pgio_data *data)
+void pnfs_ld_write_done(struct nfs_pgio_header *hdr)
 {
-	struct nfs_pgio_header *hdr = data->header;
-
-	trace_nfs4_pnfs_write(data, hdr->pnfs_error);
+	trace_nfs4_pnfs_write(hdr, hdr->pnfs_error);
 	if (!hdr->pnfs_error) {
-		pnfs_set_layoutcommit(data);
-		hdr->mds_ops->rpc_call_done(&data->task, data);
+		pnfs_set_layoutcommit(hdr);
+		hdr->mds_ops->rpc_call_done(&hdr->task, hdr);
 	} else
-		pnfs_ld_handle_write_error(data);
-	hdr->mds_ops->rpc_release(data);
+		pnfs_ld_handle_write_error(hdr);
+	hdr->mds_ops->rpc_release(hdr);
 }
 EXPORT_SYMBOL_GPL(pnfs_ld_write_done);
 
 static void
 pnfs_write_through_mds(struct nfs_pageio_descriptor *desc,
-		struct nfs_pgio_data *data)
+		struct nfs_pgio_header *hdr)
 {
-	struct nfs_pgio_header *hdr = data->header;
-
 	if (!test_and_set_bit(NFS_IOHDR_REDO, &hdr->flags)) {
 		list_splice_tail_init(&hdr->pages, &desc->pg_list);
 		nfs_pageio_reset_write_mds(desc);
 		desc->pg_recoalesce = 1;
 	}
-	nfs_pgio_data_release(data);
+	nfs_pgio_data_destroy(hdr);
 }
 
 static enum pnfs_try_status
-pnfs_try_to_write_data(struct nfs_pgio_data *wdata,
+pnfs_try_to_write_data(struct nfs_pgio_header *hdr,
 			const struct rpc_call_ops *call_ops,
 			struct pnfs_layout_segment *lseg,
 			int how)
 {
-	struct nfs_pgio_header *hdr = wdata->header;
 	struct inode *inode = hdr->inode;
 	enum pnfs_try_status trypnfs;
 	struct nfs_server *nfss = NFS_SERVER(inode);
@@ -1563,8 +1550,8 @@ pnfs_try_to_write_data(struct nfs_pgio_data *wdata,
 	hdr->mds_ops = call_ops;
 
 	dprintk("%s: Writing ino:%lu %u@%llu (how %d)\n", __func__,
-		inode->i_ino, wdata->args.count, wdata->args.offset, how);
-	trypnfs = nfss->pnfs_curr_ld->write_pagelist(wdata, how);
+		inode->i_ino, hdr->args.count, hdr->args.offset, how);
+	trypnfs = nfss->pnfs_curr_ld->write_pagelist(hdr, how);
 	if (trypnfs != PNFS_NOT_ATTEMPTED)
 		nfs_inc_stats(inode, NFSIOS_PNFS_WRITE);
 	dprintk("%s End (trypnfs:%d)\n", __func__, trypnfs);
@@ -1575,139 +1562,105 @@ static void
 pnfs_do_write(struct nfs_pageio_descriptor *desc,
 	      struct nfs_pgio_header *hdr, int how)
 {
-	struct nfs_pgio_data *data = hdr->data;
 	const struct rpc_call_ops *call_ops = desc->pg_rpc_callops;
 	struct pnfs_layout_segment *lseg = desc->pg_lseg;
 	enum pnfs_try_status trypnfs;
 
 	desc->pg_lseg = NULL;
-	trypnfs = pnfs_try_to_write_data(data, call_ops, lseg, how);
+	trypnfs = pnfs_try_to_write_data(hdr, call_ops, lseg, how);
 	if (trypnfs == PNFS_NOT_ATTEMPTED)
-		pnfs_write_through_mds(desc, data);
+		pnfs_write_through_mds(desc, hdr);
 	pnfs_put_lseg(lseg);
 }
 
 static void pnfs_writehdr_free(struct nfs_pgio_header *hdr)
 {
 	pnfs_put_lseg(hdr->lseg);
-	nfs_rw_header_free(hdr);
+	nfs_pgio_header_free(hdr);
 }
 EXPORT_SYMBOL_GPL(pnfs_writehdr_free);
 
 int
 pnfs_generic_pg_writepages(struct nfs_pageio_descriptor *desc)
 {
-	struct nfs_rw_header *whdr;
 	struct nfs_pgio_header *hdr;
 	int ret;
 
-	whdr = nfs_rw_header_alloc(desc->pg_rw_ops);
-	if (!whdr) {
+	hdr = nfs_pgio_header_alloc(desc->pg_rw_ops);
+	if (!hdr) {
 		desc->pg_completion_ops->error_cleanup(&desc->pg_list);
 		pnfs_put_lseg(desc->pg_lseg);
 		desc->pg_lseg = NULL;
 		return -ENOMEM;
 	}
-	hdr = &whdr->header;
 	nfs_pgheader_init(desc, hdr, pnfs_writehdr_free);
 	hdr->lseg = pnfs_get_lseg(desc->pg_lseg);
-	atomic_inc(&hdr->refcnt);
 	ret = nfs_generic_pgio(desc, hdr);
 	if (ret != 0) {
 		pnfs_put_lseg(desc->pg_lseg);
 		desc->pg_lseg = NULL;
 	} else
 		pnfs_do_write(desc, hdr, desc->pg_ioflags);
-	if (atomic_dec_and_test(&hdr->refcnt))
-		hdr->completion_ops->completion(hdr);
 	return ret;
 }
 EXPORT_SYMBOL_GPL(pnfs_generic_pg_writepages);
 
-int pnfs_read_done_resend_to_mds(struct inode *inode,
-				struct list_head *head,
-				const struct nfs_pgio_completion_ops *compl_ops,
-				struct nfs_direct_req *dreq)
+int pnfs_read_done_resend_to_mds(struct nfs_pgio_header *hdr)
 {
 	struct nfs_pageio_descriptor pgio;
-	LIST_HEAD(failed);
 
 	/* Resend all requests through the MDS */
-	nfs_pageio_init_read(&pgio, inode, true, compl_ops);
-	pgio.pg_dreq = dreq;
-	while (!list_empty(head)) {
-		struct nfs_page *req = nfs_list_entry(head->next);
-
-		nfs_list_remove_request(req);
-		if (!nfs_pageio_add_request(&pgio, req))
-			nfs_list_add_request(req, &failed);
-	}
-	nfs_pageio_complete(&pgio);
-
-	if (!list_empty(&failed)) {
-		list_move(&failed, head);
-		return -EIO;
-	}
-	return 0;
+	nfs_pageio_init_read(&pgio, hdr->inode, true, hdr->completion_ops);
+	return nfs_pageio_resend(&pgio, hdr);
 }
 EXPORT_SYMBOL_GPL(pnfs_read_done_resend_to_mds);
 
-static void pnfs_ld_handle_read_error(struct nfs_pgio_data *data)
+static void pnfs_ld_handle_read_error(struct nfs_pgio_header *hdr)
 {
-	struct nfs_pgio_header *hdr = data->header;
-
 	dprintk("pnfs read error = %d\n", hdr->pnfs_error);
 	if (NFS_SERVER(hdr->inode)->pnfs_curr_ld->flags &
 	    PNFS_LAYOUTRET_ON_ERROR) {
 		pnfs_return_layout(hdr->inode);
 	}
 	if (!test_and_set_bit(NFS_IOHDR_REDO, &hdr->flags))
-		data->task.tk_status = pnfs_read_done_resend_to_mds(hdr->inode,
-							&hdr->pages,
-							hdr->completion_ops,
-							hdr->dreq);
+		hdr->task.tk_status = pnfs_read_done_resend_to_mds(hdr);
 }
 
 /*
  * Called by non rpc-based layout drivers
  */
-void pnfs_ld_read_done(struct nfs_pgio_data *data)
+void pnfs_ld_read_done(struct nfs_pgio_header *hdr)
 {
-	struct nfs_pgio_header *hdr = data->header;
-
-	trace_nfs4_pnfs_read(data, hdr->pnfs_error);
+	trace_nfs4_pnfs_read(hdr, hdr->pnfs_error);
 	if (likely(!hdr->pnfs_error)) {
-		__nfs4_read_done_cb(data);
-		hdr->mds_ops->rpc_call_done(&data->task, data);
+		__nfs4_read_done_cb(hdr);
+		hdr->mds_ops->rpc_call_done(&hdr->task, hdr);
 	} else
-		pnfs_ld_handle_read_error(data);
-	hdr->mds_ops->rpc_release(data);
+		pnfs_ld_handle_read_error(hdr);
+	hdr->mds_ops->rpc_release(hdr);
 }
 EXPORT_SYMBOL_GPL(pnfs_ld_read_done);
 
 static void
 pnfs_read_through_mds(struct nfs_pageio_descriptor *desc,
-		struct nfs_pgio_data *data)
+		struct nfs_pgio_header *hdr)
 {
-	struct nfs_pgio_header *hdr = data->header;
-
 	if (!test_and_set_bit(NFS_IOHDR_REDO, &hdr->flags)) {
 		list_splice_tail_init(&hdr->pages, &desc->pg_list);
 		nfs_pageio_reset_read_mds(desc);
 		desc->pg_recoalesce = 1;
 	}
-	nfs_pgio_data_release(data);
+	nfs_pgio_data_destroy(hdr);
 }
 
 /*
  * Call the appropriate parallel I/O subsystem read function.
  */
 static enum pnfs_try_status
-pnfs_try_to_read_data(struct nfs_pgio_data *rdata,
+pnfs_try_to_read_data(struct nfs_pgio_header *hdr,
 		       const struct rpc_call_ops *call_ops,
 		       struct pnfs_layout_segment *lseg)
 {
-	struct nfs_pgio_header *hdr = rdata->header;
 	struct inode *inode = hdr->inode;
 	struct nfs_server *nfss = NFS_SERVER(inode);
 	enum pnfs_try_status trypnfs;
@@ -1715,9 +1668,9 @@ pnfs_try_to_read_data(struct nfs_pgio_data *rdata,
 	hdr->mds_ops = call_ops;
 
 	dprintk("%s: Reading ino:%lu %u@%llu\n",
-		__func__, inode->i_ino, rdata->args.count, rdata->args.offset);
+		__func__, inode->i_ino, hdr->args.count, hdr->args.offset);
 
-	trypnfs = nfss->pnfs_curr_ld->read_pagelist(rdata);
+	trypnfs = nfss->pnfs_curr_ld->read_pagelist(hdr);
 	if (trypnfs != PNFS_NOT_ATTEMPTED)
 		nfs_inc_stats(inode, NFSIOS_PNFS_READ);
 	dprintk("%s End (trypnfs:%d)\n", __func__, trypnfs);
@@ -1727,52 +1680,46 @@ pnfs_try_to_read_data(struct nfs_pgio_data *rdata,
 static void
 pnfs_do_read(struct nfs_pageio_descriptor *desc, struct nfs_pgio_header *hdr)
 {
-	struct nfs_pgio_data *data = hdr->data;
 	const struct rpc_call_ops *call_ops = desc->pg_rpc_callops;
 	struct pnfs_layout_segment *lseg = desc->pg_lseg;
 	enum pnfs_try_status trypnfs;
 
 	desc->pg_lseg = NULL;
-	trypnfs = pnfs_try_to_read_data(data, call_ops, lseg);
+	trypnfs = pnfs_try_to_read_data(hdr, call_ops, lseg);
 	if (trypnfs == PNFS_NOT_ATTEMPTED)
-		pnfs_read_through_mds(desc, data);
+		pnfs_read_through_mds(desc, hdr);
 	pnfs_put_lseg(lseg);
 }
 
 static void pnfs_readhdr_free(struct nfs_pgio_header *hdr)
 {
 	pnfs_put_lseg(hdr->lseg);
-	nfs_rw_header_free(hdr);
+	nfs_pgio_header_free(hdr);
 }
 EXPORT_SYMBOL_GPL(pnfs_readhdr_free);
 
 int
 pnfs_generic_pg_readpages(struct nfs_pageio_descriptor *desc)
 {
-	struct nfs_rw_header *rhdr;
 	struct nfs_pgio_header *hdr;
 	int ret;
 
-	rhdr = nfs_rw_header_alloc(desc->pg_rw_ops);
-	if (!rhdr) {
+	hdr = nfs_pgio_header_alloc(desc->pg_rw_ops);
+	if (!hdr) {
 		desc->pg_completion_ops->error_cleanup(&desc->pg_list);
 		ret = -ENOMEM;
 		pnfs_put_lseg(desc->pg_lseg);
 		desc->pg_lseg = NULL;
 		return ret;
 	}
-	hdr = &rhdr->header;
 	nfs_pgheader_init(desc, hdr, pnfs_readhdr_free);
 	hdr->lseg = pnfs_get_lseg(desc->pg_lseg);
-	atomic_inc(&hdr->refcnt);
 	ret = nfs_generic_pgio(desc, hdr);
 	if (ret != 0) {
 		pnfs_put_lseg(desc->pg_lseg);
 		desc->pg_lseg = NULL;
 	} else
 		pnfs_do_read(desc, hdr);
-	if (atomic_dec_and_test(&hdr->refcnt))
-		hdr->completion_ops->completion(hdr);
 	return ret;
 }
 EXPORT_SYMBOL_GPL(pnfs_generic_pg_readpages);
@@ -1820,12 +1767,11 @@ void pnfs_set_lo_fail(struct pnfs_layout_segment *lseg)
 EXPORT_SYMBOL_GPL(pnfs_set_lo_fail);
 
 void
-pnfs_set_layoutcommit(struct nfs_pgio_data *wdata)
+pnfs_set_layoutcommit(struct nfs_pgio_header *hdr)
 {
-	struct nfs_pgio_header *hdr = wdata->header;
 	struct inode *inode = hdr->inode;
 	struct nfs_inode *nfsi = NFS_I(inode);
-	loff_t end_pos = wdata->mds_offset + wdata->res.count;
+	loff_t end_pos = hdr->mds_offset + hdr->res.count;
 	bool mark_as_dirty = false;
 
 	spin_lock(&inode->i_lock);

fs/nfs/pnfs.h

@@ -32,6 +32,7 @@
 
 #include <linux/nfs_fs.h>
 #include <linux/nfs_page.h>
+#include <linux/workqueue.h>
 
 enum {
 	NFS_LSEG_VALID = 0,	/* cleared when lseg is recalled/returned */
@@ -46,6 +47,7 @@ struct pnfs_layout_segment {
 	atomic_t pls_refcount;
 	unsigned long pls_flags;
 	struct pnfs_layout_hdr *pls_layout;
+	struct work_struct pls_work;
 };
 
 enum pnfs_try_status {
@@ -104,6 +106,8 @@ struct pnfs_layoutdriver_type {
 				  int max);
 	void (*recover_commit_reqs) (struct list_head *list,
 				     struct nfs_commit_info *cinfo);
+	struct nfs_page * (*search_commit_reqs)(struct nfs_commit_info *cinfo,
+						struct page *page);
 	int (*commit_pagelist)(struct inode *inode,
 			       struct list_head *mds_pages,
 			       int how,
@@ -113,8 +117,8 @@ struct pnfs_layoutdriver_type {
 	 * Return PNFS_ATTEMPTED to indicate the layout code has attempted
 	 * I/O, else return PNFS_NOT_ATTEMPTED to fall back to normal NFS
 	 */
-	enum pnfs_try_status (*read_pagelist) (struct nfs_pgio_data *nfs_data);
-	enum pnfs_try_status (*write_pagelist) (struct nfs_pgio_data *nfs_data, int how);
+	enum pnfs_try_status (*read_pagelist)(struct nfs_pgio_header *);
+	enum pnfs_try_status (*write_pagelist)(struct nfs_pgio_header *, int);
 
 	void (*free_deviceid_node) (struct nfs4_deviceid_node *);
 
@@ -179,6 +183,7 @@ extern int nfs4_proc_layoutreturn(struct nfs4_layoutreturn *lrp);
 /* pnfs.c */
 void pnfs_get_layout_hdr(struct pnfs_layout_hdr *lo);
 void pnfs_put_lseg(struct pnfs_layout_segment *lseg);
+void pnfs_put_lseg_async(struct pnfs_layout_segment *lseg);
 
 void set_pnfs_layoutdriver(struct nfs_server *, const struct nfs_fh *, u32);
 void unset_pnfs_layoutdriver(struct nfs_server *);
@@ -213,13 +218,13 @@ bool pnfs_roc(struct inode *ino);
 void pnfs_roc_release(struct inode *ino);
 void pnfs_roc_set_barrier(struct inode *ino, u32 barrier);
 bool pnfs_roc_drain(struct inode *ino, u32 *barrier, struct rpc_task *task);
-void pnfs_set_layoutcommit(struct nfs_pgio_data *wdata);
+void pnfs_set_layoutcommit(struct nfs_pgio_header *);
 void pnfs_cleanup_layoutcommit(struct nfs4_layoutcommit_data *data);
 int pnfs_layoutcommit_inode(struct inode *inode, bool sync);
 int _pnfs_return_layout(struct inode *);
 int pnfs_commit_and_return_layout(struct inode *);
-void pnfs_ld_write_done(struct nfs_pgio_data *);
-void pnfs_ld_read_done(struct nfs_pgio_data *);
+void pnfs_ld_write_done(struct nfs_pgio_header *);
+void pnfs_ld_read_done(struct nfs_pgio_header *);
 struct pnfs_layout_segment *pnfs_update_layout(struct inode *ino,
 					       struct nfs_open_context *ctx,
 					       loff_t pos,
@@ -228,12 +233,8 @@ struct pnfs_layout_segment *pnfs_update_layout(struct inode *ino,
 					       gfp_t gfp_flags);
 
 void nfs4_deviceid_mark_client_invalid(struct nfs_client *clp);
-int pnfs_read_done_resend_to_mds(struct inode *inode, struct list_head *head,
-			const struct nfs_pgio_completion_ops *compl_ops,
-			struct nfs_direct_req *dreq);
-int pnfs_write_done_resend_to_mds(struct inode *inode, struct list_head *head,
-			const struct nfs_pgio_completion_ops *compl_ops,
-			struct nfs_direct_req *dreq);
+int pnfs_read_done_resend_to_mds(struct nfs_pgio_header *);
+int pnfs_write_done_resend_to_mds(struct nfs_pgio_header *);
 struct nfs4_threshold *pnfs_mdsthreshold_alloc(void);
 
 /* nfs4_deviceid_flags */
@@ -345,6 +346,17 @@ pnfs_recover_commit_reqs(struct inode *inode, struct list_head *list,
 	NFS_SERVER(inode)->pnfs_curr_ld->recover_commit_reqs(list, cinfo);
 }
 
+static inline struct nfs_page *
+pnfs_search_commit_reqs(struct inode *inode, struct nfs_commit_info *cinfo,
+			struct page *page)
+{
+	struct pnfs_layoutdriver_type *ld = NFS_SERVER(inode)->pnfs_curr_ld;
+
+	if (ld == NULL || ld->search_commit_reqs == NULL)
+		return NULL;
+	return ld->search_commit_reqs(cinfo, page);
+}
+
 /* Should the pNFS client commit and return the layout upon a setattr */
 static inline bool
 pnfs_ld_layoutret_on_setattr(struct inode *inode)
@@ -410,6 +422,10 @@ static inline void pnfs_put_lseg(struct pnfs_layout_segment *lseg)
 {
 }
 
+static inline void pnfs_put_lseg_async(struct pnfs_layout_segment *lseg)
+{
+}
+
 static inline int pnfs_return_layout(struct inode *ino)
 {
 	return 0;
@@ -496,6 +512,13 @@ pnfs_recover_commit_reqs(struct inode *inode, struct list_head *list,
 {
 }
 
+static inline struct nfs_page *
+pnfs_search_commit_reqs(struct inode *inode, struct nfs_commit_info *cinfo,
+			struct page *page)
+{
+	return NULL;
+}
+
 static inline int pnfs_layoutcommit_inode(struct inode *inode, bool sync)
 {
 	return 0;

fs/nfs/proc.c

@@ -578,46 +578,49 @@ nfs_proc_pathconf(struct nfs_server *server, struct nfs_fh *fhandle,
 	return 0;
 }
 
-static int nfs_read_done(struct rpc_task *task, struct nfs_pgio_data *data)
+static int nfs_read_done(struct rpc_task *task, struct nfs_pgio_header *hdr)
 {
-	struct inode *inode = data->header->inode;
+	struct inode *inode = hdr->inode;
 
 	nfs_invalidate_atime(inode);
 	if (task->tk_status >= 0) {
-		nfs_refresh_inode(inode, data->res.fattr);
+		nfs_refresh_inode(inode, hdr->res.fattr);
 		/* Emulate the eof flag, which isn't normally needed in NFSv2
 		 * as it is guaranteed to always return the file attributes
 		 */
-		if (data->args.offset + data->res.count >= data->res.fattr->size)
-			data->res.eof = 1;
+		if (hdr->args.offset + hdr->res.count >= hdr->res.fattr->size)
+			hdr->res.eof = 1;
 	}
 	return 0;
 }
 
-static void nfs_proc_read_setup(struct nfs_pgio_data *data, struct rpc_message *msg)
+static void nfs_proc_read_setup(struct nfs_pgio_header *hdr,
+				struct rpc_message *msg)
 {
 	msg->rpc_proc = &nfs_procedures[NFSPROC_READ];
 }
 
-static int nfs_proc_pgio_rpc_prepare(struct rpc_task *task, struct nfs_pgio_data *data)
+static int nfs_proc_pgio_rpc_prepare(struct rpc_task *task,
+				     struct nfs_pgio_header *hdr)
 {
 	rpc_call_start(task);
 	return 0;
 }
 
-static int nfs_write_done(struct rpc_task *task, struct nfs_pgio_data *data)
+static int nfs_write_done(struct rpc_task *task, struct nfs_pgio_header *hdr)
 {
-	struct inode *inode = data->header->inode;
+	struct inode *inode = hdr->inode;
 
 	if (task->tk_status >= 0)
-		nfs_post_op_update_inode_force_wcc(inode, data->res.fattr);
+		nfs_post_op_update_inode_force_wcc(inode, hdr->res.fattr);
 	return 0;
 }
 
-static void nfs_proc_write_setup(struct nfs_pgio_data *data, struct rpc_message *msg)
+static void nfs_proc_write_setup(struct nfs_pgio_header *hdr,
+				 struct rpc_message *msg)
 {
 	/* Note: NFSv2 ignores @stable and always uses NFS_FILE_SYNC */
-	data->args.stable = NFS_FILE_SYNC;
+	hdr->args.stable = NFS_FILE_SYNC;
 	msg->rpc_proc = &nfs_procedures[NFSPROC_WRITE];
 }
 

fs/nfs/read.c

@@ -33,12 +33,12 @@ static const struct nfs_rw_ops nfs_rw_read_ops;
 
 static struct kmem_cache *nfs_rdata_cachep;
 
-static struct nfs_rw_header *nfs_readhdr_alloc(void)
+static struct nfs_pgio_header *nfs_readhdr_alloc(void)
 {
 	return kmem_cache_zalloc(nfs_rdata_cachep, GFP_KERNEL);
 }
 
-static void nfs_readhdr_free(struct nfs_rw_header *rhdr)
+static void nfs_readhdr_free(struct nfs_pgio_header *rhdr)
 {
 	kmem_cache_free(nfs_rdata_cachep, rhdr);
 }
@@ -115,12 +115,6 @@ static void nfs_readpage_release(struct nfs_page *req)
 
 		unlock_page(req->wb_page);
 	}
-
-	dprintk("NFS: read done (%s/%Lu %d@%Ld)\n",
-			req->wb_context->dentry->d_inode->i_sb->s_id,
-			(unsigned long long)NFS_FILEID(req->wb_context->dentry->d_inode),
-			req->wb_bytes,
-			(long long)req_offset(req));
 	nfs_release_request(req);
 }
 
@@ -172,14 +166,15 @@ out:
 	hdr->release(hdr);
 }
 
-static void nfs_initiate_read(struct nfs_pgio_data *data, struct rpc_message *msg,
+static void nfs_initiate_read(struct nfs_pgio_header *hdr,
+			      struct rpc_message *msg,
 			      struct rpc_task_setup *task_setup_data, int how)
 {
-	struct inode *inode = data->header->inode;
+	struct inode *inode = hdr->inode;
 	int swap_flags = IS_SWAPFILE(inode) ? NFS_RPC_SWAPFLAGS : 0;
 
 	task_setup_data->flags |= swap_flags;
-	NFS_PROTO(inode)->read_setup(data, msg);
+	NFS_PROTO(inode)->read_setup(hdr, msg);
 }
 
 static void
@@ -203,14 +198,15 @@ static const struct nfs_pgio_completion_ops nfs_async_read_completion_ops = {
  * This is the callback from RPC telling us whether a reply was
  * received or some error occurred (timeout or socket shutdown).
  */
-static int nfs_readpage_done(struct rpc_task *task, struct nfs_pgio_data *data,
+static int nfs_readpage_done(struct rpc_task *task,
+			     struct nfs_pgio_header *hdr,
 			     struct inode *inode)
 {
-	int status = NFS_PROTO(inode)->read_done(task, data);
+	int status = NFS_PROTO(inode)->read_done(task, hdr);
 	if (status != 0)
 		return status;
 
-	nfs_add_stats(inode, NFSIOS_SERVERREADBYTES, data->res.count);
+	nfs_add_stats(inode, NFSIOS_SERVERREADBYTES, hdr->res.count);
 
 	if (task->tk_status == -ESTALE) {
 		set_bit(NFS_INO_STALE, &NFS_I(inode)->flags);
@@ -219,34 +215,34 @@ static int nfs_readpage_done(struct rpc_task *task, struct nfs_pgio_data *data,
 	return 0;
 }
 
-static void nfs_readpage_retry(struct rpc_task *task, struct nfs_pgio_data *data)
+static void nfs_readpage_retry(struct rpc_task *task,
+			       struct nfs_pgio_header *hdr)
 {
-	struct nfs_pgio_args *argp = &data->args;
-	struct nfs_pgio_res  *resp = &data->res;
+	struct nfs_pgio_args *argp = &hdr->args;
+	struct nfs_pgio_res  *resp = &hdr->res;
 
 	/* This is a short read! */
-	nfs_inc_stats(data->header->inode, NFSIOS_SHORTREAD);
+	nfs_inc_stats(hdr->inode, NFSIOS_SHORTREAD);
 	/* Has the server at least made some progress? */
 	if (resp->count == 0) {
-		nfs_set_pgio_error(data->header, -EIO, argp->offset);
+		nfs_set_pgio_error(hdr, -EIO, argp->offset);
 		return;
 	}
-	/* Yes, so retry the read at the end of the data */
-	data->mds_offset += resp->count;
+	/* Yes, so retry the read at the end of the hdr */
+	hdr->mds_offset += resp->count;
 	argp->offset += resp->count;
 	argp->pgbase += resp->count;
 	argp->count -= resp->count;
 	rpc_restart_call_prepare(task);
 }
 
-static void nfs_readpage_result(struct rpc_task *task, struct nfs_pgio_data *data)
+static void nfs_readpage_result(struct rpc_task *task,
+				struct nfs_pgio_header *hdr)
 {
-	struct nfs_pgio_header *hdr = data->header;
-
-	if (data->res.eof) {
+	if (hdr->res.eof) {
 		loff_t bound;
 
-		bound = data->args.offset + data->res.count;
+		bound = hdr->args.offset + hdr->res.count;
 		spin_lock(&hdr->lock);
 		if (bound < hdr->io_start + hdr->good_bytes) {
 			set_bit(NFS_IOHDR_EOF, &hdr->flags);
@@ -254,8 +250,8 @@ static void nfs_readpage_result(struct rpc_task *task, struct nfs_pgio_data *dat
 			hdr->good_bytes = bound - hdr->io_start;
 		}
 		spin_unlock(&hdr->lock);
-	} else if (data->res.count != data->args.count)
-		nfs_readpage_retry(task, data);
+	} else if (hdr->res.count != hdr->args.count)
+		nfs_readpage_retry(task, hdr);
 }
 
 /*
@@ -404,7 +400,7 @@ out:
 int __init nfs_init_readpagecache(void)
 {
 	nfs_rdata_cachep = kmem_cache_create("nfs_read_data",
-					     sizeof(struct nfs_rw_header),
+					     sizeof(struct nfs_pgio_header),
 					     0, SLAB_HWCACHE_ALIGN,
 					     NULL);
 	if (nfs_rdata_cachep == NULL)

fs/nfs/super.c

@@ -1027,8 +1027,7 @@ static bool nfs_auth_info_add(struct nfs_auth_info *auth_info,
 			      rpc_authflavor_t flavor)
 {
 	unsigned int i;
-	unsigned int max_flavor_len = (sizeof(auth_info->flavors) /
-				       sizeof(auth_info->flavors[0]));
+	unsigned int max_flavor_len = ARRAY_SIZE(auth_info->flavors);
 
 	/* make sure this flavor isn't already in the list */
 	for (i = 0; i < auth_info->flavor_len; i++) {
@@ -2180,7 +2179,7 @@ out_no_address:
 	return -EINVAL;
 }
 
-#define NFS_MOUNT_CMP_FLAGMASK ~(NFS_MOUNT_INTR \
+#define NFS_REMOUNT_CMP_FLAGMASK ~(NFS_MOUNT_INTR \
 		| NFS_MOUNT_SECURE \
 		| NFS_MOUNT_TCP \
 		| NFS_MOUNT_VER3 \
@@ -2188,15 +2187,16 @@ out_no_address:
 		| NFS_MOUNT_NONLM \
 		| NFS_MOUNT_BROKEN_SUID \
 		| NFS_MOUNT_STRICTLOCK \
-		| NFS_MOUNT_UNSHARED \
-		| NFS_MOUNT_NORESVPORT \
 		| NFS_MOUNT_LEGACY_INTERFACE)
 
+#define NFS_MOUNT_CMP_FLAGMASK (NFS_REMOUNT_CMP_FLAGMASK & \
+		~(NFS_MOUNT_UNSHARED | NFS_MOUNT_NORESVPORT))
+
 static int
 nfs_compare_remount_data(struct nfs_server *nfss,
 			 struct nfs_parsed_mount_data *data)
 {
-	if ((data->flags ^ nfss->flags) & NFS_MOUNT_CMP_FLAGMASK ||
+	if ((data->flags ^ nfss->flags) & NFS_REMOUNT_CMP_FLAGMASK ||
 	    data->rsize != nfss->rsize ||
 	    data->wsize != nfss->wsize ||
 	    data->version != nfss->nfs_client->rpc_ops->version ||

fs/nfs/write.c

@@ -47,6 +47,8 @@ static const struct nfs_pgio_completion_ops nfs_async_write_completion_ops;
 static const struct nfs_commit_completion_ops nfs_commit_completion_ops;
 static const struct nfs_rw_ops nfs_rw_write_ops;
 static void nfs_clear_request_commit(struct nfs_page *req);
+static void nfs_init_cinfo_from_inode(struct nfs_commit_info *cinfo,
+				      struct inode *inode);
 
 static struct kmem_cache *nfs_wdata_cachep;
 static mempool_t *nfs_wdata_mempool;
@@ -71,18 +73,18 @@ void nfs_commit_free(struct nfs_commit_data *p)
 }
 EXPORT_SYMBOL_GPL(nfs_commit_free);
 
-static struct nfs_rw_header *nfs_writehdr_alloc(void)
+static struct nfs_pgio_header *nfs_writehdr_alloc(void)
 {
-	struct nfs_rw_header *p = mempool_alloc(nfs_wdata_mempool, GFP_NOIO);
+	struct nfs_pgio_header *p = mempool_alloc(nfs_wdata_mempool, GFP_NOIO);
 
 	if (p)
 		memset(p, 0, sizeof(*p));
 	return p;
 }
 
-static void nfs_writehdr_free(struct nfs_rw_header *whdr)
+static void nfs_writehdr_free(struct nfs_pgio_header *hdr)
 {
-	mempool_free(whdr, nfs_wdata_mempool);
+	mempool_free(hdr, nfs_wdata_mempool);
 }
 
 static void nfs_context_set_write_error(struct nfs_open_context *ctx, int error)
@@ -92,6 +94,38 @@ static void nfs_context_set_write_error(struct nfs_open_context *ctx, int error)
 	set_bit(NFS_CONTEXT_ERROR_WRITE, &ctx->flags);
 }
 
+/*
+ * nfs_page_search_commits_for_head_request_locked
+ *
+ * Search through commit lists on @inode for the head request for @page.
+ * Must be called while holding the inode (which is cinfo) lock.
+ *
+ * Returns the head request if found, or NULL if not found.
+ */
+static struct nfs_page *
+nfs_page_search_commits_for_head_request_locked(struct nfs_inode *nfsi,
+						struct page *page)
+{
+	struct nfs_page *freq, *t;
+	struct nfs_commit_info cinfo;
+	struct inode *inode = &nfsi->vfs_inode;
+
+	nfs_init_cinfo_from_inode(&cinfo, inode);
+
+	/* search through pnfs commit lists */
+	freq = pnfs_search_commit_reqs(inode, &cinfo, page);
+	if (freq)
+		return freq->wb_head;
+
+	/* Linearly search the commit list for the correct request */
+	list_for_each_entry_safe(freq, t, &cinfo.mds->list, wb_list) {
+		if (freq->wb_page == page)
+			return freq->wb_head;
+	}
+
+	return NULL;
+}
+
 /*
  * nfs_page_find_head_request_locked - find head request associated with @page
  *
@@ -106,21 +140,12 @@ nfs_page_find_head_request_locked(struct nfs_inode *nfsi, struct page *page)
 
 	if (PagePrivate(page))
 		req = (struct nfs_page *)page_private(page);
-	else if (unlikely(PageSwapCache(page))) {
-		struct nfs_page *freq, *t;
-
-		/* Linearly search the commit list for the correct req */
-		list_for_each_entry_safe(freq, t, &nfsi->commit_info.list, wb_list) {
-			if (freq->wb_page == page) {
-				req = freq->wb_head;
-				break;
-			}
-		}
-	}
+	else if (unlikely(PageSwapCache(page)))
+		req = nfs_page_search_commits_for_head_request_locked(nfsi,
+			page);
 
 	if (req) {
 		WARN_ON_ONCE(req->wb_head != req);
-
 		kref_get(&req->wb_kref);
 	}
 
@@ -216,7 +241,7 @@ static bool nfs_page_group_covers_page(struct nfs_page *req)
 	unsigned int pos = 0;
 	unsigned int len = nfs_page_length(req->wb_page);
 
-	nfs_page_group_lock(req);
+	nfs_page_group_lock(req, true);
 
 	do {
 		tmp = nfs_page_group_search_locked(req->wb_head, pos);
@@ -379,8 +404,6 @@ nfs_destroy_unlinked_subrequests(struct nfs_page *destroy_list,
 		subreq->wb_head = subreq;
 		subreq->wb_this_page = subreq;
 
-		nfs_clear_request_commit(subreq);
-
 		/* subreq is now totally disconnected from page group or any
 		 * write / commit lists. last chance to wake any waiters */
 		nfs_unlock_request(subreq);
@@ -456,7 +479,9 @@ try_again:
 	}
 
 	/* lock each request in the page group */
-	nfs_page_group_lock(head);
+	ret = nfs_page_group_lock(head, false);
+	if (ret < 0)
+		return ERR_PTR(ret);
 	subreq = head;
 	do {
 		/*
@@ -488,7 +513,7 @@ try_again:
 	 * Commit list removal accounting is done after locks are dropped */
 	subreq = head;
 	do {
-		nfs_list_remove_request(subreq);
+		nfs_clear_request_commit(subreq);
 		subreq = subreq->wb_this_page;
 	} while (subreq != head);
 
@@ -518,15 +543,11 @@ try_again:
 
 	nfs_page_group_unlock(head);
 
-	/* drop lock to clear_request_commit the head req and clean up
-	 * requests on destroy list */
+	/* drop lock to clean uprequests on destroy list */
 	spin_unlock(&inode->i_lock);
 
 	nfs_destroy_unlinked_subrequests(destroy_list, head);
 
-	/* clean up commit list state */
-	nfs_clear_request_commit(head);
-
 	/* still holds ref on head from nfs_page_find_head_request_locked
 	 * and still has lock on head from lock loop */
 	return head;
@@ -705,6 +726,8 @@ static void nfs_inode_remove_request(struct nfs_page *req)
 
 	if (test_and_clear_bit(PG_INODE_REF, &req->wb_flags))
 		nfs_release_request(req);
+	else
+		WARN_ON_ONCE(1);
 }
 
 static void
@@ -808,6 +831,7 @@ nfs_clear_page_commit(struct page *page)
 	dec_bdi_stat(page_file_mapping(page)->backing_dev_info, BDI_RECLAIMABLE);
 }
 
+/* Called holding inode (/cinfo) lock */
 static void
 nfs_clear_request_commit(struct nfs_page *req)
 {
@@ -817,20 +841,17 @@ nfs_clear_request_commit(struct nfs_page *req)
 
 		nfs_init_cinfo_from_inode(&cinfo, inode);
 		if (!pnfs_clear_request_commit(req, &cinfo)) {
-			spin_lock(cinfo.lock);
 			nfs_request_remove_commit_list(req, &cinfo);
-			spin_unlock(cinfo.lock);
 		}
 		nfs_clear_page_commit(req->wb_page);
 	}
 }
 
-static inline
-int nfs_write_need_commit(struct nfs_pgio_data *data)
+int nfs_write_need_commit(struct nfs_pgio_header *hdr)
 {
-	if (data->verf.committed == NFS_DATA_SYNC)
-		return data->header->lseg == NULL;
-	return data->verf.committed != NFS_FILE_SYNC;
+	if (hdr->verf.committed == NFS_DATA_SYNC)
+		return hdr->lseg == NULL;
+	return hdr->verf.committed != NFS_FILE_SYNC;
 }
 
 #else
@@ -856,8 +877,7 @@ nfs_clear_request_commit(struct nfs_page *req)
 {
 }
 
-static inline
-int nfs_write_need_commit(struct nfs_pgio_data *data)
+int nfs_write_need_commit(struct nfs_pgio_header *hdr)
 {
 	return 0;
 }
@@ -883,11 +903,7 @@ static void nfs_write_completion(struct nfs_pgio_header *hdr)
 			nfs_context_set_write_error(req->wb_context, hdr->error);
 			goto remove_req;
 		}
-		if (test_bit(NFS_IOHDR_NEED_RESCHED, &hdr->flags)) {
-			nfs_mark_request_dirty(req);
-			goto next;
-		}
-		if (test_bit(NFS_IOHDR_NEED_COMMIT, &hdr->flags)) {
+		if (nfs_write_need_commit(hdr)) {
 			memcpy(&req->wb_verf, &hdr->verf.verifier, sizeof(req->wb_verf));
 			nfs_mark_request_commit(req, hdr->lseg, &cinfo);
 			goto next;
@@ -1038,9 +1054,9 @@ static struct nfs_page *nfs_try_to_update_request(struct inode *inode,
 	else
 		req->wb_bytes = rqend - req->wb_offset;
 out_unlock:
-	spin_unlock(&inode->i_lock);
 	if (req)
 		nfs_clear_request_commit(req);
+	spin_unlock(&inode->i_lock);
 	return req;
 out_flushme:
 	spin_unlock(&inode->i_lock);
@@ -1241,17 +1257,18 @@ static int flush_task_priority(int how)
 	return RPC_PRIORITY_NORMAL;
 }
 
-static void nfs_initiate_write(struct nfs_pgio_data *data, struct rpc_message *msg,
+static void nfs_initiate_write(struct nfs_pgio_header *hdr,
+			       struct rpc_message *msg,
 			       struct rpc_task_setup *task_setup_data, int how)
 {
-	struct inode *inode = data->header->inode;
+	struct inode *inode = hdr->inode;
 	int priority = flush_task_priority(how);
 
 	task_setup_data->priority = priority;
-	NFS_PROTO(inode)->write_setup(data, msg);
+	NFS_PROTO(inode)->write_setup(hdr, msg);
 
 	nfs4_state_protect_write(NFS_SERVER(inode)->nfs_client,
-				 &task_setup_data->rpc_client, msg, data);
+				 &task_setup_data->rpc_client, msg, hdr);
 }
 
 /* If a nfs_flush_* function fails, it should remove reqs from @head and
@@ -1313,21 +1330,9 @@ void nfs_commit_prepare(struct rpc_task *task, void *calldata)
 	NFS_PROTO(data->inode)->commit_rpc_prepare(task, data);
 }
 
-static void nfs_writeback_release_common(struct nfs_pgio_data *data)
+static void nfs_writeback_release_common(struct nfs_pgio_header *hdr)
 {
-	struct nfs_pgio_header *hdr = data->header;
-	int status = data->task.tk_status;
-
-	if ((status >= 0) && nfs_write_need_commit(data)) {
-		spin_lock(&hdr->lock);
-		if (test_bit(NFS_IOHDR_NEED_RESCHED, &hdr->flags))
-			; /* Do nothing */
-		else if (!test_and_set_bit(NFS_IOHDR_NEED_COMMIT, &hdr->flags))
-			memcpy(&hdr->verf, &data->verf, sizeof(hdr->verf));
-		else if (memcmp(&hdr->verf, &data->verf, sizeof(hdr->verf)))
-			set_bit(NFS_IOHDR_NEED_RESCHED, &hdr->flags);
-		spin_unlock(&hdr->lock);
-	}
+	/* do nothing! */
 }
 
 /*
@@ -1358,7 +1363,8 @@ static int nfs_should_remove_suid(const struct inode *inode)
 /*
  * This function is called when the WRITE call is complete.
  */
-static int nfs_writeback_done(struct rpc_task *task, struct nfs_pgio_data *data,
+static int nfs_writeback_done(struct rpc_task *task,
+			      struct nfs_pgio_header *hdr,
 			      struct inode *inode)
 {
 	int status;
@@ -1370,13 +1376,14 @@ static int nfs_writeback_done(struct rpc_task *task, struct nfs_pgio_data *data,
 	 * another writer had changed the file, but some applications
 	 * depend on tighter cache coherency when writing.
 	 */
-	status = NFS_PROTO(inode)->write_done(task, data);
+	status = NFS_PROTO(inode)->write_done(task, hdr);
 	if (status != 0)
 		return status;
-	nfs_add_stats(inode, NFSIOS_SERVERWRITTENBYTES, data->res.count);
+	nfs_add_stats(inode, NFSIOS_SERVERWRITTENBYTES, hdr->res.count);
 
 #if IS_ENABLED(CONFIG_NFS_V3) || IS_ENABLED(CONFIG_NFS_V4)
-	if (data->res.verf->committed < data->args.stable && task->tk_status >= 0) {
+	if (hdr->res.verf->committed < hdr->args.stable &&
+	    task->tk_status >= 0) {
 		/* We tried a write call, but the server did not
 		 * commit data to stable storage even though we
 		 * requested it.
@@ -1392,7 +1399,7 @@ static int nfs_writeback_done(struct rpc_task *task, struct nfs_pgio_data *data,
 			dprintk("NFS:       faulty NFS server %s:"
 				" (committed = %d) != (stable = %d)\n",
 				NFS_SERVER(inode)->nfs_client->cl_hostname,
-				data->res.verf->committed, data->args.stable);
+				hdr->res.verf->committed, hdr->args.stable);
 			complain = jiffies + 300 * HZ;
 		}
 	}
@@ -1407,16 +1414,17 @@ static int nfs_writeback_done(struct rpc_task *task, struct nfs_pgio_data *data,
 /*
  * This function is called when the WRITE call is complete.
  */
-static void nfs_writeback_result(struct rpc_task *task, struct nfs_pgio_data *data)
+static void nfs_writeback_result(struct rpc_task *task,
+				 struct nfs_pgio_header *hdr)
 {
-	struct nfs_pgio_args	*argp = &data->args;
-	struct nfs_pgio_res	*resp = &data->res;
+	struct nfs_pgio_args	*argp = &hdr->args;
+	struct nfs_pgio_res	*resp = &hdr->res;
 
 	if (resp->count < argp->count) {
 		static unsigned long    complain;
 
 		/* This a short write! */
-		nfs_inc_stats(data->header->inode, NFSIOS_SHORTWRITE);
+		nfs_inc_stats(hdr->inode, NFSIOS_SHORTWRITE);
 
 		/* Has the server at least made some progress? */
 		if (resp->count == 0) {
@@ -1426,14 +1434,14 @@ static void nfs_writeback_result(struct rpc_task *task, struct nfs_pgio_data *da
 				       argp->count);
 				complain = jiffies + 300 * HZ;
 			}
-			nfs_set_pgio_error(data->header, -EIO, argp->offset);
+			nfs_set_pgio_error(hdr, -EIO, argp->offset);
 			task->tk_status = -EIO;
 			return;
 		}
 		/* Was this an NFSv2 write or an NFSv3 stable write? */
 		if (resp->verf->committed != NFS_UNSTABLE) {
 			/* Resend from where the server left off */
-			data->mds_offset += resp->count;
+			hdr->mds_offset += resp->count;
 			argp->offset += resp->count;
 			argp->pgbase += resp->count;
 			argp->count -= resp->count;
@@ -1884,7 +1892,7 @@ int nfs_migrate_page(struct address_space *mapping, struct page *newpage,
 int __init nfs_init_writepagecache(void)
 {
 	nfs_wdata_cachep = kmem_cache_create("nfs_write_data",
-					     sizeof(struct nfs_rw_header),
+					     sizeof(struct nfs_pgio_header),
 					     0, SLAB_HWCACHE_ALIGN,
 					     NULL);
 	if (nfs_wdata_cachep == NULL)

fs/nfs_common/nfsacl.c

@@ -30,9 +30,6 @@
 
 MODULE_LICENSE("GPL");
 
-EXPORT_SYMBOL_GPL(nfsacl_encode);
-EXPORT_SYMBOL_GPL(nfsacl_decode);
-
 struct nfsacl_encode_desc {
 	struct xdr_array2_desc desc;
 	unsigned int count;
@@ -136,6 +133,7 @@ int nfsacl_encode(struct xdr_buf *buf, unsigned int base, struct inode *inode,
 			  nfsacl_desc.desc.array_len;
 	return err;
 }
+EXPORT_SYMBOL_GPL(nfsacl_encode);
 
 struct nfsacl_decode_desc {
 	struct xdr_array2_desc desc;
@@ -295,3 +293,4 @@ int nfsacl_decode(struct xdr_buf *buf, unsigned int base, unsigned int *aclcnt,
 	return 8 + nfsacl_desc.desc.elem_size *
 		   nfsacl_desc.desc.array_len;
 }
+EXPORT_SYMBOL_GPL(nfsacl_decode);

include/linux/nfs_fs.h

@@ -52,6 +52,7 @@ struct nfs_access_entry {
 	unsigned long		jiffies;
 	struct rpc_cred *	cred;
 	int			mask;
+	struct rcu_head		rcu_head;
 };
 
 struct nfs_lockowner {
@@ -352,6 +353,7 @@ extern int nfs_release(struct inode *, struct file *);
 extern int nfs_attribute_timeout(struct inode *inode);
 extern int nfs_attribute_cache_expired(struct inode *inode);
 extern int nfs_revalidate_inode(struct nfs_server *server, struct inode *inode);
+extern int nfs_revalidate_inode_rcu(struct nfs_server *server, struct inode *inode);
 extern int __nfs_revalidate_inode(struct nfs_server *, struct inode *);
 extern int nfs_revalidate_mapping(struct inode *inode, struct address_space *mapping);
 extern int nfs_setattr(struct dentry *, struct iattr *);

include/linux/nfs_fs_sb.h

@@ -45,6 +45,7 @@ struct nfs_client {
 	struct sockaddr_storage	cl_addr;	/* server identifier */
 	size_t			cl_addrlen;
 	char *			cl_hostname;	/* hostname of server */
+	char *			cl_acceptor;	/* GSSAPI acceptor name */
 	struct list_head	cl_share_link;	/* link in global client list */
 	struct list_head	cl_superblocks;	/* List of nfs_server structs */
 

include/linux/nfs_page.h

@@ -26,7 +26,7 @@ enum {
 	PG_MAPPED,		/* page private set for buffered io */
 	PG_CLEAN,		/* write succeeded */
 	PG_COMMIT_TO_DS,	/* used by pnfs layouts */
-	PG_INODE_REF,		/* extra ref held by inode (head req only) */
+	PG_INODE_REF,		/* extra ref held by inode when in writeback */
 	PG_HEADLOCK,		/* page group lock of wb_head */
 	PG_TEARDOWN,		/* page group sync for destroy */
 	PG_UNLOCKPAGE,		/* page group sync bit in read path */
@@ -62,12 +62,13 @@ struct nfs_pageio_ops {
 
 struct nfs_rw_ops {
 	const fmode_t rw_mode;
-	struct nfs_rw_header *(*rw_alloc_header)(void);
-	void (*rw_free_header)(struct nfs_rw_header *);
-	void (*rw_release)(struct nfs_pgio_data *);
-	int  (*rw_done)(struct rpc_task *, struct nfs_pgio_data *, struct inode *);
-	void (*rw_result)(struct rpc_task *, struct nfs_pgio_data *);
-	void (*rw_initiate)(struct nfs_pgio_data *, struct rpc_message *,
+	struct nfs_pgio_header *(*rw_alloc_header)(void);
+	void (*rw_free_header)(struct nfs_pgio_header *);
+	void (*rw_release)(struct nfs_pgio_header *);
+	int  (*rw_done)(struct rpc_task *, struct nfs_pgio_header *,
+			struct inode *);
+	void (*rw_result)(struct rpc_task *, struct nfs_pgio_header *);
+	void (*rw_initiate)(struct nfs_pgio_header *, struct rpc_message *,
 			    struct rpc_task_setup *, int);
 };
 
@@ -111,6 +112,8 @@ extern	void nfs_pageio_init(struct nfs_pageio_descriptor *desc,
 			     int how);
 extern	int nfs_pageio_add_request(struct nfs_pageio_descriptor *,
 				   struct nfs_page *);
+extern  int nfs_pageio_resend(struct nfs_pageio_descriptor *,
+			      struct nfs_pgio_header *);
 extern	void nfs_pageio_complete(struct nfs_pageio_descriptor *desc);
 extern	void nfs_pageio_cond_complete(struct nfs_pageio_descriptor *, pgoff_t);
 extern size_t nfs_generic_pg_test(struct nfs_pageio_descriptor *desc,
@@ -119,7 +122,7 @@ extern size_t nfs_generic_pg_test(struct nfs_pageio_descriptor *desc,
 extern  int nfs_wait_on_request(struct nfs_page *);
 extern	void nfs_unlock_request(struct nfs_page *req);
 extern	void nfs_unlock_and_release_request(struct nfs_page *);
-extern void nfs_page_group_lock(struct nfs_page *);
+extern int nfs_page_group_lock(struct nfs_page *, bool);
 extern void nfs_page_group_unlock(struct nfs_page *);
 extern bool nfs_page_group_sync_on_bit(struct nfs_page *, unsigned int);
 

include/linux/nfs_xdr.h

@@ -993,6 +993,7 @@ struct nfs4_setclientid {
 	unsigned int			sc_uaddr_len;
 	char				sc_uaddr[RPCBIND_MAXUADDRLEN + 1];
 	u32				sc_cb_ident;
+	struct rpc_cred			*sc_cred;
 };
 
 struct nfs4_setclientid_res {
@@ -1253,18 +1254,12 @@ enum {
 	NFS_IOHDR_ERROR = 0,
 	NFS_IOHDR_EOF,
 	NFS_IOHDR_REDO,
-	NFS_IOHDR_NEED_COMMIT,
-	NFS_IOHDR_NEED_RESCHED,
 };
 
-struct nfs_pgio_data;
-
 struct nfs_pgio_header {
 	struct inode		*inode;
 	struct rpc_cred		*cred;
 	struct list_head	pages;
-	struct nfs_pgio_data	*data;
-	atomic_t		refcnt;
 	struct nfs_page		*req;
 	struct nfs_writeverf	verf;		/* Used for writes */
 	struct pnfs_layout_segment *lseg;
@@ -1281,28 +1276,22 @@ struct nfs_pgio_header {
 	int			error;		/* merge with pnfs_error */
 	unsigned long		good_bytes;	/* boundary of good data */
 	unsigned long		flags;
-};
 
-struct nfs_pgio_data {
-	struct nfs_pgio_header	*header;
+	/*
+	 * rpc data
+	 */
 	struct rpc_task		task;
 	struct nfs_fattr	fattr;
-	struct nfs_writeverf	verf;		/* Used for writes */
 	struct nfs_pgio_args	args;		/* argument struct */
 	struct nfs_pgio_res	res;		/* result struct */
 	unsigned long		timestamp;	/* For lease renewal */
-	int (*pgio_done_cb) (struct rpc_task *task, struct nfs_pgio_data *data);
+	int (*pgio_done_cb)(struct rpc_task *, struct nfs_pgio_header *);
 	__u64			mds_offset;	/* Filelayout dense stripe */
-	struct nfs_page_array	pages;
+	struct nfs_page_array	page_array;
 	struct nfs_client	*ds_clp;	/* pNFS data server */
 	int			ds_idx;		/* ds index if ds_clp is set */
 };
 
-struct nfs_rw_header {
-	struct nfs_pgio_header	header;
-	struct nfs_pgio_data	rpc_data;
-};
-
 struct nfs_mds_commit_info {
 	atomic_t rpcs_out;
 	unsigned long		ncommit;
@@ -1432,11 +1421,12 @@ struct nfs_rpc_ops {
 			     struct nfs_pathconf *);
 	int	(*set_capabilities)(struct nfs_server *, struct nfs_fh *);
 	int	(*decode_dirent)(struct xdr_stream *, struct nfs_entry *, int);
-	int	(*pgio_rpc_prepare)(struct rpc_task *, struct nfs_pgio_data *);
-	void	(*read_setup)   (struct nfs_pgio_data *, struct rpc_message *);
-	int	(*read_done)  (struct rpc_task *, struct nfs_pgio_data *);
-	void	(*write_setup)  (struct nfs_pgio_data *, struct rpc_message *);
-	int	(*write_done)  (struct rpc_task *, struct nfs_pgio_data *);
+	int	(*pgio_rpc_prepare)(struct rpc_task *,
+				    struct nfs_pgio_header *);
+	void	(*read_setup)(struct nfs_pgio_header *, struct rpc_message *);
+	int	(*read_done)(struct rpc_task *, struct nfs_pgio_header *);
+	void	(*write_setup)(struct nfs_pgio_header *, struct rpc_message *);
+	int	(*write_done)(struct rpc_task *, struct nfs_pgio_header *);
 	void	(*commit_setup) (struct nfs_commit_data *, struct rpc_message *);
 	void	(*commit_rpc_prepare)(struct rpc_task *, struct nfs_commit_data *);
 	int	(*commit_done) (struct rpc_task *, struct nfs_commit_data *);

include/linux/sunrpc/auth.h

@@ -103,6 +103,7 @@ struct rpc_auth_create_args {
 
 /* Flags for rpcauth_lookupcred() */
 #define RPCAUTH_LOOKUP_NEW		0x01	/* Accept an uninitialised cred */
+#define RPCAUTH_LOOKUP_RCU		0x02	/* lock-less lookup */
 
 /*
  * Client authentication ops
@@ -140,6 +141,7 @@ struct rpc_credops {
 						void *, __be32 *, void *);
 	int			(*crkey_timeout)(struct rpc_cred *);
 	bool			(*crkey_to_expire)(struct rpc_cred *);
+	char *			(*crstringify_acceptor)(struct rpc_cred *);
 };
 
 extern const struct rpc_authops	authunix_ops;
@@ -153,6 +155,7 @@ void			rpc_destroy_generic_auth(void);
 void 			rpc_destroy_authunix(void);
 
 struct rpc_cred *	rpc_lookup_cred(void);
+struct rpc_cred *	rpc_lookup_cred_nonblock(void);
 struct rpc_cred *	rpc_lookup_machine_cred(const char *service_name);
 int			rpcauth_register(const struct rpc_authops *);
 int			rpcauth_unregister(const struct rpc_authops *);
@@ -182,6 +185,7 @@ void			rpcauth_clear_credcache(struct rpc_cred_cache *);
 int			rpcauth_key_timeout_notify(struct rpc_auth *,
 						struct rpc_cred *);
 bool			rpcauth_cred_key_to_expire(struct rpc_cred *);
+char *			rpcauth_stringify_acceptor(struct rpc_cred *);
 
 static inline
 struct rpc_cred *	get_rpccred(struct rpc_cred *cred)

include/linux/sunrpc/auth_gss.h

@@ -69,8 +69,9 @@ struct gss_cl_ctx {
 	enum rpc_gss_proc	gc_proc;
 	u32			gc_seq;
 	spinlock_t		gc_seq_lock;
-	struct gss_ctx __rcu	*gc_gss_ctx;
+	struct gss_ctx		*gc_gss_ctx;
 	struct xdr_netobj	gc_wire_ctx;
+	struct xdr_netobj	gc_acceptor;
 	u32			gc_win;
 	unsigned long		gc_expiry;
 	struct rcu_head		gc_rcu;

include/linux/sunrpc/gss_krb5.h

@@ -81,7 +81,7 @@ struct gss_krb5_enctype {
 		       struct xdr_netobj *in,
 		       struct xdr_netobj *out);	/* complete key generation */
 	u32 (*encrypt_v2) (struct krb5_ctx *kctx, u32 offset,
-			   struct xdr_buf *buf, int ec,
+			   struct xdr_buf *buf,
 			   struct page **pages); /* v2 encryption function */
 	u32 (*decrypt_v2) (struct krb5_ctx *kctx, u32 offset,
 			   struct xdr_buf *buf, u32 *headskip,
@@ -310,7 +310,7 @@ gss_krb5_aes_make_key(const struct gss_krb5_enctype *gk5e,
 
 u32
 gss_krb5_aes_encrypt(struct krb5_ctx *kctx, u32 offset,
-		     struct xdr_buf *buf, int ec,
+		     struct xdr_buf *buf,
 		     struct page **pages);
 
 u32

include/linux/sunrpc/xprtrdma.h

@@ -62,8 +62,6 @@
 #define RPCRDMA_INLINE_PAD_THRESH  (512)/* payload threshold to pad (bytes) */
 
 /* memory registration strategies */
-#define RPCRDMA_PERSISTENT_REGISTRATION (1)
-
 enum rpcrdma_memreg {
 	RPCRDMA_BOUNCEBUFFERS = 0,
 	RPCRDMA_REGISTER,

net/sunrpc/addr.c

@@ -176,7 +176,7 @@ static int rpc_parse_scope_id(struct net *net, const char *buf,
 	len = (buf + buflen) - delim - 1;
 	p = kstrndup(delim + 1, len, GFP_KERNEL);
 	if (p) {
-		unsigned long scope_id = 0;
+		u32 scope_id = 0;
 		struct net_device *dev;
 
 		dev = dev_get_by_name(net, p);
@@ -184,7 +184,7 @@ static int rpc_parse_scope_id(struct net *net, const char *buf,
 			scope_id = dev->ifindex;
 			dev_put(dev);
 		} else {
-			if (strict_strtoul(p, 10, &scope_id) == 0) {
+			if (kstrtou32(p, 10, &scope_id) == 0) {
 				kfree(p);
 				return 0;
 			}
@@ -304,7 +304,7 @@ char *rpc_sockaddr2uaddr(const struct sockaddr *sap, gfp_t gfp_flags)
  * @sap: buffer into which to plant socket address
  * @salen: size of buffer
  *
- * @uaddr does not have to be '\0'-terminated, but strict_strtoul() and
+ * @uaddr does not have to be '\0'-terminated, but kstrtou8() and
  * rpc_pton() require proper string termination to be successful.
  *
  * Returns the size of the socket address if successful; otherwise
@@ -315,7 +315,7 @@ size_t rpc_uaddr2sockaddr(struct net *net, const char *uaddr,
 			  const size_t salen)
 {
 	char *c, buf[RPCBIND_MAXUADDRLEN + sizeof('\0')];
-	unsigned long portlo, porthi;
+	u8 portlo, porthi;
 	unsigned short port;
 
 	if (uaddr_len > RPCBIND_MAXUADDRLEN)
@@ -327,18 +327,14 @@ size_t rpc_uaddr2sockaddr(struct net *net, const char *uaddr,
 	c = strrchr(buf, '.');
 	if (unlikely(c == NULL))
 		return 0;
-	if (unlikely(strict_strtoul(c + 1, 10, &portlo) != 0))
-		return 0;
-	if (unlikely(portlo > 255))
+	if (unlikely(kstrtou8(c + 1, 10, &portlo) != 0))
 		return 0;
 
 	*c = '\0';
 	c = strrchr(buf, '.');
 	if (unlikely(c == NULL))
 		return 0;
-	if (unlikely(strict_strtoul(c + 1, 10, &porthi) != 0))
-		return 0;
-	if (unlikely(porthi > 255))
+	if (unlikely(kstrtou8(c + 1, 10, &porthi) != 0))
 		return 0;
 
 	port = (unsigned short)((porthi << 8) | portlo);

net/sunrpc/auth.c

@@ -48,7 +48,7 @@ static int param_set_hashtbl_sz(const char *val, const struct kernel_param *kp)
 
 	if (!val)
 		goto out_inval;
-	ret = strict_strtoul(val, 0, &num);
+	ret = kstrtoul(val, 0, &num);
 	if (ret == -EINVAL)
 		goto out_inval;
 	nbits = fls(num);
@@ -80,6 +80,10 @@ static struct kernel_param_ops param_ops_hashtbl_sz = {
 module_param_named(auth_hashtable_size, auth_hashbits, hashtbl_sz, 0644);
 MODULE_PARM_DESC(auth_hashtable_size, "RPC credential cache hashtable size");
 
+static unsigned long auth_max_cred_cachesize = ULONG_MAX;
+module_param(auth_max_cred_cachesize, ulong, 0644);
+MODULE_PARM_DESC(auth_max_cred_cachesize, "RPC credential maximum total cache size");
+
 static u32
 pseudoflavor_to_flavor(u32 flavor) {
 	if (flavor > RPC_AUTH_MAXFLAVOR)
@@ -363,6 +367,15 @@ rpcauth_cred_key_to_expire(struct rpc_cred *cred)
 }
 EXPORT_SYMBOL_GPL(rpcauth_cred_key_to_expire);
 
+char *
+rpcauth_stringify_acceptor(struct rpc_cred *cred)
+{
+	if (!cred->cr_ops->crstringify_acceptor)
+		return NULL;
+	return cred->cr_ops->crstringify_acceptor(cred);
+}
+EXPORT_SYMBOL_GPL(rpcauth_stringify_acceptor);
+
 /*
  * Destroy a list of credentials
  */
@@ -472,6 +485,20 @@ rpcauth_prune_expired(struct list_head *free, int nr_to_scan)
 	return freed;
 }
 
+static unsigned long
+rpcauth_cache_do_shrink(int nr_to_scan)
+{
+	LIST_HEAD(free);
+	unsigned long freed;
+
+	spin_lock(&rpc_credcache_lock);
+	freed = rpcauth_prune_expired(&free, nr_to_scan);
+	spin_unlock(&rpc_credcache_lock);
+	rpcauth_destroy_credlist(&free);
+
+	return freed;
+}
+
 /*
  * Run memory cache shrinker.
  */
@@ -479,9 +506,6 @@ static unsigned long
 rpcauth_cache_shrink_scan(struct shrinker *shrink, struct shrink_control *sc)
 
 {
-	LIST_HEAD(free);
-	unsigned long freed;
-
 	if ((sc->gfp_mask & GFP_KERNEL) != GFP_KERNEL)
 		return SHRINK_STOP;
 
@@ -489,12 +513,7 @@ rpcauth_cache_shrink_scan(struct shrinker *shrink, struct shrink_control *sc)
 	if (list_empty(&cred_unused))
 		return SHRINK_STOP;
 
-	spin_lock(&rpc_credcache_lock);
-	freed = rpcauth_prune_expired(&free, sc->nr_to_scan);
-	spin_unlock(&rpc_credcache_lock);
-	rpcauth_destroy_credlist(&free);
-
-	return freed;
+	return rpcauth_cache_do_shrink(sc->nr_to_scan);
 }
 
 static unsigned long
@@ -504,6 +523,21 @@ rpcauth_cache_shrink_count(struct shrinker *shrink, struct shrink_control *sc)
 	return (number_cred_unused / 100) * sysctl_vfs_cache_pressure;
 }
 
+static void
+rpcauth_cache_enforce_limit(void)
+{
+	unsigned long diff;
+	unsigned int nr_to_scan;
+
+	if (number_cred_unused <= auth_max_cred_cachesize)
+		return;
+	diff = number_cred_unused - auth_max_cred_cachesize;
+	nr_to_scan = 100;
+	if (diff < nr_to_scan)
+		nr_to_scan = diff;
+	rpcauth_cache_do_shrink(nr_to_scan);
+}
+
 /*
  * Look up a process' credentials in the authentication cache
  */
@@ -523,6 +557,12 @@ rpcauth_lookup_credcache(struct rpc_auth *auth, struct auth_cred * acred,
 	hlist_for_each_entry_rcu(entry, &cache->hashtable[nr], cr_hash) {
 		if (!entry->cr_ops->crmatch(acred, entry, flags))
 			continue;
+		if (flags & RPCAUTH_LOOKUP_RCU) {
+			if (test_bit(RPCAUTH_CRED_HASHED, &entry->cr_flags) &&
+			    !test_bit(RPCAUTH_CRED_NEW, &entry->cr_flags))
+				cred = entry;
+			break;
+		}
 		spin_lock(&cache->lock);
 		if (test_bit(RPCAUTH_CRED_HASHED, &entry->cr_flags) == 0) {
 			spin_unlock(&cache->lock);
@@ -537,6 +577,9 @@ rpcauth_lookup_credcache(struct rpc_auth *auth, struct auth_cred * acred,
 	if (cred != NULL)
 		goto found;
 
+	if (flags & RPCAUTH_LOOKUP_RCU)
+		return ERR_PTR(-ECHILD);
+
 	new = auth->au_ops->crcreate(auth, acred, flags);
 	if (IS_ERR(new)) {
 		cred = new;
@@ -557,6 +600,7 @@ rpcauth_lookup_credcache(struct rpc_auth *auth, struct auth_cred * acred,
 	} else
 		list_add_tail(&new->cr_lru, &free);
 	spin_unlock(&cache->lock);
+	rpcauth_cache_enforce_limit();
 found:
 	if (test_bit(RPCAUTH_CRED_NEW, &cred->cr_flags) &&
 	    cred->cr_ops->cr_init != NULL &&
@@ -586,10 +630,8 @@ rpcauth_lookupcred(struct rpc_auth *auth, int flags)
 	memset(&acred, 0, sizeof(acred));
 	acred.uid = cred->fsuid;
 	acred.gid = cred->fsgid;
-	acred.group_info = get_group_info(((struct cred *)cred)->group_info);
-
+	acred.group_info = cred->group_info;
 	ret = auth->au_ops->lookup_cred(auth, &acred, flags);
-	put_group_info(acred.group_info);
 	return ret;
 }
 EXPORT_SYMBOL_GPL(rpcauth_lookupcred);

net/sunrpc/auth_generic.c

@@ -38,6 +38,12 @@ struct rpc_cred *rpc_lookup_cred(void)
 }
 EXPORT_SYMBOL_GPL(rpc_lookup_cred);
 
+struct rpc_cred *rpc_lookup_cred_nonblock(void)
+{
+	return rpcauth_lookupcred(&generic_auth, RPCAUTH_LOOKUP_RCU);
+}
+EXPORT_SYMBOL_GPL(rpc_lookup_cred_nonblock);
+
 /*
  * Public call interface for looking up machine creds.
  */

net/sunrpc/auth_gss/auth_gss.c

@@ -183,8 +183,9 @@ gss_cred_get_ctx(struct rpc_cred *cred)
 	struct gss_cl_ctx *ctx = NULL;
 
 	rcu_read_lock();
-	if (gss_cred->gc_ctx)
-		ctx = gss_get_ctx(gss_cred->gc_ctx);
+	ctx = rcu_dereference(gss_cred->gc_ctx);
+	if (ctx)
+		gss_get_ctx(ctx);
 	rcu_read_unlock();
 	return ctx;
 }
@@ -262,9 +263,22 @@ gss_fill_context(const void *p, const void *end, struct gss_cl_ctx *ctx, struct
 		p = ERR_PTR(ret);
 		goto err;
 	}
-	dprintk("RPC:       %s Success. gc_expiry %lu now %lu timeout %u\n",
-		__func__, ctx->gc_expiry, now, timeout);
-	return q;
+
+	/* is there any trailing data? */
+	if (q == end) {
+		p = q;
+		goto done;
+	}
+
+	/* pull in acceptor name (if there is one) */
+	p = simple_get_netobj(q, end, &ctx->gc_acceptor);
+	if (IS_ERR(p))
+		goto err;
+done:
+	dprintk("RPC:       %s Success. gc_expiry %lu now %lu timeout %u acceptor %.*s\n",
+		__func__, ctx->gc_expiry, now, timeout, ctx->gc_acceptor.len,
+		ctx->gc_acceptor.data);
+	return p;
 err:
 	dprintk("RPC:       %s returns error %ld\n", __func__, -PTR_ERR(p));
 	return p;
@@ -1194,13 +1208,13 @@ gss_destroying_context(struct rpc_cred *cred)
 {
 	struct gss_cred *gss_cred = container_of(cred, struct gss_cred, gc_base);
 	struct gss_auth *gss_auth = container_of(cred->cr_auth, struct gss_auth, rpc_auth);
+	struct gss_cl_ctx *ctx = rcu_dereference_protected(gss_cred->gc_ctx, 1);
 	struct rpc_task *task;
 
-	if (gss_cred->gc_ctx == NULL ||
-	    test_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags) == 0)
+	if (test_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags) == 0)
 		return 0;
 
-	gss_cred->gc_ctx->gc_proc = RPC_GSS_PROC_DESTROY;
+	ctx->gc_proc = RPC_GSS_PROC_DESTROY;
 	cred->cr_ops = &gss_nullops;
 
 	/* Take a reference to ensure the cred will be destroyed either
@@ -1225,6 +1239,7 @@ gss_do_free_ctx(struct gss_cl_ctx *ctx)
 
 	gss_delete_sec_context(&ctx->gc_gss_ctx);
 	kfree(ctx->gc_wire_ctx.data);
+	kfree(ctx->gc_acceptor.data);
 	kfree(ctx);
 }
 
@@ -1260,7 +1275,7 @@ gss_destroy_nullcred(struct rpc_cred *cred)
 {
 	struct gss_cred *gss_cred = container_of(cred, struct gss_cred, gc_base);
 	struct gss_auth *gss_auth = container_of(cred->cr_auth, struct gss_auth, rpc_auth);
-	struct gss_cl_ctx *ctx = gss_cred->gc_ctx;
+	struct gss_cl_ctx *ctx = rcu_dereference_protected(gss_cred->gc_ctx, 1);
 
 	RCU_INIT_POINTER(gss_cred->gc_ctx, NULL);
 	call_rcu(&cred->cr_rcu, gss_free_cred_callback);
@@ -1332,6 +1347,36 @@ gss_cred_init(struct rpc_auth *auth, struct rpc_cred *cred)
 	return err;
 }
 
+static char *
+gss_stringify_acceptor(struct rpc_cred *cred)
+{
+	char *string = NULL;
+	struct gss_cred *gss_cred = container_of(cred, struct gss_cred, gc_base);
+	struct gss_cl_ctx *ctx;
+	struct xdr_netobj *acceptor;
+
+	rcu_read_lock();
+	ctx = rcu_dereference(gss_cred->gc_ctx);
+	if (!ctx)
+		goto out;
+
+	acceptor = &ctx->gc_acceptor;
+
+	/* no point if there's no string */
+	if (!acceptor->len)
+		goto out;
+
+	string = kmalloc(acceptor->len + 1, GFP_KERNEL);
+	if (!string)
+		goto out;
+
+	memcpy(string, acceptor->data, acceptor->len);
+	string[acceptor->len] = '\0';
+out:
+	rcu_read_unlock();
+	return string;
+}
+
 /*
  * Returns -EACCES if GSS context is NULL or will expire within the
  * timeout (miliseconds)
@@ -1340,15 +1385,16 @@ static int
 gss_key_timeout(struct rpc_cred *rc)
 {
 	struct gss_cred *gss_cred = container_of(rc, struct gss_cred, gc_base);
+	struct gss_cl_ctx *ctx;
 	unsigned long now = jiffies;
 	unsigned long expire;
 
-	if (gss_cred->gc_ctx == NULL)
-		return -EACCES;
-
-	expire = gss_cred->gc_ctx->gc_expiry - (gss_key_expire_timeo * HZ);
-
-	if (time_after(now, expire))
+	rcu_read_lock();
+	ctx = rcu_dereference(gss_cred->gc_ctx);
+	if (ctx)
+		expire = ctx->gc_expiry - (gss_key_expire_timeo * HZ);
+	rcu_read_unlock();
+	if (!ctx || time_after(now, expire))
 		return -EACCES;
 	return 0;
 }
@@ -1357,13 +1403,19 @@ static int
 gss_match(struct auth_cred *acred, struct rpc_cred *rc, int flags)
 {
 	struct gss_cred *gss_cred = container_of(rc, struct gss_cred, gc_base);
+	struct gss_cl_ctx *ctx;
 	int ret;
 
 	if (test_bit(RPCAUTH_CRED_NEW, &rc->cr_flags))
 		goto out;
 	/* Don't match with creds that have expired. */
-	if (time_after(jiffies, gss_cred->gc_ctx->gc_expiry))
+	rcu_read_lock();
+	ctx = rcu_dereference(gss_cred->gc_ctx);
+	if (!ctx || time_after(jiffies, ctx->gc_expiry)) {
+		rcu_read_unlock();
 		return 0;
+	}
+	rcu_read_unlock();
 	if (!test_bit(RPCAUTH_CRED_UPTODATE, &rc->cr_flags))
 		return 0;
 out:
@@ -1909,29 +1961,31 @@ static const struct rpc_authops authgss_ops = {
 };
 
 static const struct rpc_credops gss_credops = {
-	.cr_name	= "AUTH_GSS",
-	.crdestroy	= gss_destroy_cred,
-	.cr_init	= gss_cred_init,
-	.crbind		= rpcauth_generic_bind_cred,
-	.crmatch	= gss_match,
-	.crmarshal	= gss_marshal,
-	.crrefresh	= gss_refresh,
-	.crvalidate	= gss_validate,
-	.crwrap_req	= gss_wrap_req,
-	.crunwrap_resp	= gss_unwrap_resp,
-	.crkey_timeout	= gss_key_timeout,
+	.cr_name		= "AUTH_GSS",
+	.crdestroy		= gss_destroy_cred,
+	.cr_init		= gss_cred_init,
+	.crbind			= rpcauth_generic_bind_cred,
+	.crmatch		= gss_match,
+	.crmarshal		= gss_marshal,
+	.crrefresh		= gss_refresh,
+	.crvalidate		= gss_validate,
+	.crwrap_req		= gss_wrap_req,
+	.crunwrap_resp		= gss_unwrap_resp,
+	.crkey_timeout		= gss_key_timeout,
+	.crstringify_acceptor	= gss_stringify_acceptor,
 };
 
 static const struct rpc_credops gss_nullops = {
-	.cr_name	= "AUTH_GSS",
-	.crdestroy	= gss_destroy_nullcred,
-	.crbind		= rpcauth_generic_bind_cred,
-	.crmatch	= gss_match,
-	.crmarshal	= gss_marshal,
-	.crrefresh	= gss_refresh_null,
-	.crvalidate	= gss_validate,
-	.crwrap_req	= gss_wrap_req,
-	.crunwrap_resp	= gss_unwrap_resp,
+	.cr_name		= "AUTH_GSS",
+	.crdestroy		= gss_destroy_nullcred,
+	.crbind			= rpcauth_generic_bind_cred,
+	.crmatch		= gss_match,
+	.crmarshal		= gss_marshal,
+	.crrefresh		= gss_refresh_null,
+	.crvalidate		= gss_validate,
+	.crwrap_req		= gss_wrap_req,
+	.crunwrap_resp		= gss_unwrap_resp,
+	.crstringify_acceptor	= gss_stringify_acceptor,
 };
 
 static const struct rpc_pipe_ops gss_upcall_ops_v0 = {

net/sunrpc/auth_gss/gss_krb5_crypto.c

@@ -641,7 +641,7 @@ out:
 
 u32
 gss_krb5_aes_encrypt(struct krb5_ctx *kctx, u32 offset,
-		     struct xdr_buf *buf, int ec, struct page **pages)
+		     struct xdr_buf *buf, struct page **pages)
 {
 	u32 err;
 	struct xdr_netobj hmac;
@@ -684,13 +684,8 @@ gss_krb5_aes_encrypt(struct krb5_ctx *kctx, u32 offset,
 		ecptr = buf->tail[0].iov_base;
 	}
 
-	memset(ecptr, 'X', ec);
-	buf->tail[0].iov_len += ec;
-	buf->len += ec;
-
 	/* copy plaintext gss token header after filler (if any) */
-	memcpy(ecptr + ec, buf->head[0].iov_base + offset,
-						GSS_KRB5_TOK_HDR_LEN);
+	memcpy(ecptr, buf->head[0].iov_base + offset, GSS_KRB5_TOK_HDR_LEN);
 	buf->tail[0].iov_len += GSS_KRB5_TOK_HDR_LEN;
 	buf->len += GSS_KRB5_TOK_HDR_LEN;
 

net/sunrpc/auth_gss/gss_krb5_seal.c

@@ -70,31 +70,37 @@
 
 DEFINE_SPINLOCK(krb5_seq_lock);
 
-static char *
+static void *
 setup_token(struct krb5_ctx *ctx, struct xdr_netobj *token)
 {
-	__be16 *ptr, *krb5_hdr;
+	u16 *ptr;
+	void *krb5_hdr;
 	int body_size = GSS_KRB5_TOK_HDR_LEN + ctx->gk5e->cksumlength;
 
 	token->len = g_token_size(&ctx->mech_used, body_size);
 
-	ptr = (__be16 *)token->data;
+	ptr = (u16 *)token->data;
 	g_make_token_header(&ctx->mech_used, body_size, (unsigned char **)&ptr);
 
 	/* ptr now at start of header described in rfc 1964, section 1.2.1: */
 	krb5_hdr = ptr;
 	*ptr++ = KG_TOK_MIC_MSG;
-	*ptr++ = cpu_to_le16(ctx->gk5e->signalg);
+	/*
+	 * signalg is stored as if it were converted from LE to host endian, even
+	 * though it's an opaque pair of bytes according to the RFC.
+	 */
+	*ptr++ = (__force u16)cpu_to_le16(ctx->gk5e->signalg);
 	*ptr++ = SEAL_ALG_NONE;
-	*ptr++ = 0xffff;
+	*ptr = 0xffff;
 
-	return (char *)krb5_hdr;
+	return krb5_hdr;
 }
 
 static void *
 setup_token_v2(struct krb5_ctx *ctx, struct xdr_netobj *token)
 {
-	__be16 *ptr, *krb5_hdr;
+	u16 *ptr;
+	void *krb5_hdr;
 	u8 *p, flags = 0x00;
 
 	if ((ctx->flags & KRB5_CTX_FLAG_INITIATOR) == 0)
@@ -104,15 +110,15 @@ setup_token_v2(struct krb5_ctx *ctx, struct xdr_netobj *token)
 
 	/* Per rfc 4121, sec 4.2.6.1, there is no header,
 	 * just start the token */
-	krb5_hdr = ptr = (__be16 *)token->data;
+	krb5_hdr = ptr = (u16 *)token->data;
 
 	*ptr++ = KG2_TOK_MIC;
 	p = (u8 *)ptr;
 	*p++ = flags;
 	*p++ = 0xff;
-	ptr = (__be16 *)p;
-	*ptr++ = 0xffff;
+	ptr = (u16 *)p;
 	*ptr++ = 0xffff;
+	*ptr = 0xffff;
 
 	token->len = GSS_KRB5_TOK_HDR_LEN + ctx->gk5e->cksumlength;
 	return krb5_hdr;
@@ -181,7 +187,7 @@ gss_get_mic_v2(struct krb5_ctx *ctx, struct xdr_buf *text,
 	spin_lock(&krb5_seq_lock);
 	seq_send = ctx->seq_send64++;
 	spin_unlock(&krb5_seq_lock);
-	*((u64 *)(krb5_hdr + 8)) = cpu_to_be64(seq_send);
+	*((__be64 *)(krb5_hdr + 8)) = cpu_to_be64(seq_send);
 
 	if (ctx->initiate) {
 		cksumkey = ctx->initiator_sign;

net/sunrpc/auth_gss/gss_krb5_wrap.c

@@ -201,9 +201,15 @@ gss_wrap_kerberos_v1(struct krb5_ctx *kctx, int offset,
 
 	msg_start = ptr + GSS_KRB5_TOK_HDR_LEN + kctx->gk5e->cksumlength;
 
-	*(__be16 *)(ptr + 2) = cpu_to_le16(kctx->gk5e->signalg);
-	memset(ptr + 4, 0xff, 4);
-	*(__be16 *)(ptr + 4) = cpu_to_le16(kctx->gk5e->sealalg);
+	/*
+	 * signalg and sealalg are stored as if they were converted from LE
+	 * to host endian, even though they're opaque pairs of bytes according
+	 * to the RFC.
+	 */
+	*(__le16 *)(ptr + 2) = cpu_to_le16(kctx->gk5e->signalg);
+	*(__le16 *)(ptr + 4) = cpu_to_le16(kctx->gk5e->sealalg);
+	ptr[6] = 0xff;
+	ptr[7] = 0xff;
 
 	gss_krb5_make_confounder(msg_start, conflen);
 
@@ -438,7 +444,7 @@ gss_wrap_kerberos_v2(struct krb5_ctx *kctx, u32 offset,
 	u8		*ptr, *plainhdr;
 	s32		now;
 	u8		flags = 0x00;
-	__be16		*be16ptr, ec = 0;
+	__be16		*be16ptr;
 	__be64		*be64ptr;
 	u32		err;
 
@@ -468,16 +474,16 @@ gss_wrap_kerberos_v2(struct krb5_ctx *kctx, u32 offset,
 	be16ptr = (__be16 *)ptr;
 
 	blocksize = crypto_blkcipher_blocksize(kctx->acceptor_enc);
-	*be16ptr++ = cpu_to_be16(ec);
+	*be16ptr++ = 0;
 	/* "inner" token header always uses 0 for RRC */
-	*be16ptr++ = cpu_to_be16(0);
+	*be16ptr++ = 0;
 
 	be64ptr = (__be64 *)be16ptr;
 	spin_lock(&krb5_seq_lock);
 	*be64ptr = cpu_to_be64(kctx->seq_send64++);
 	spin_unlock(&krb5_seq_lock);
 
-	err = (*kctx->gk5e->encrypt_v2)(kctx, offset, buf, ec, pages);
+	err = (*kctx->gk5e->encrypt_v2)(kctx, offset, buf, pages);
 	if (err)
 		return err;
 

net/sunrpc/auth_null.c

@@ -35,6 +35,8 @@ nul_destroy(struct rpc_auth *auth)
 static struct rpc_cred *
 nul_lookup_cred(struct rpc_auth *auth, struct auth_cred *acred, int flags)
 {
+	if (flags & RPCAUTH_LOOKUP_RCU)
+		return &null_cred;
 	return get_rpccred(&null_cred);
 }
 

net/sunrpc/clnt.c

@@ -1746,6 +1746,7 @@ call_bind_status(struct rpc_task *task)
 	case -EHOSTDOWN:
 	case -EHOSTUNREACH:
 	case -ENETUNREACH:
+	case -ENOBUFS:
 	case -EPIPE:
 		dprintk("RPC: %5u remote rpcbind unreachable: %d\n",
 				task->tk_pid, task->tk_status);
@@ -1812,6 +1813,8 @@ call_connect_status(struct rpc_task *task)
 	case -ECONNABORTED:
 	case -ENETUNREACH:
 	case -EHOSTUNREACH:
+	case -ENOBUFS:
+	case -EPIPE:
 		if (RPC_IS_SOFTCONN(task))
 			break;
 		/* retry with existing socket, after a delay */
@@ -1918,6 +1921,7 @@ call_transmit_status(struct rpc_task *task)
 	case -ECONNRESET:
 	case -ECONNABORTED:
 	case -ENOTCONN:
+	case -ENOBUFS:
 	case -EPIPE:
 		rpc_task_force_reencode(task);
 	}
@@ -2034,6 +2038,7 @@ call_status(struct rpc_task *task)
 	case -ECONNRESET:
 	case -ECONNABORTED:
 		rpc_force_rebind(clnt);
+	case -ENOBUFS:
 		rpc_delay(task, 3*HZ);
 	case -EPIPE:
 	case -ENOTCONN:

net/sunrpc/rpc_pipe.c

@@ -195,7 +195,7 @@ static struct inode *
 rpc_alloc_inode(struct super_block *sb)
 {
 	struct rpc_inode *rpci;
-	rpci = (struct rpc_inode *)kmem_cache_alloc(rpc_inode_cachep, GFP_KERNEL);
+	rpci = kmem_cache_alloc(rpc_inode_cachep, GFP_KERNEL);
 	if (!rpci)
 		return NULL;
 	return &rpci->vfs_inode;

net/sunrpc/xprt.c

@@ -744,6 +744,7 @@ static void xprt_connect_status(struct rpc_task *task)
 	case -ECONNABORTED:
 	case -ENETUNREACH:
 	case -EHOSTUNREACH:
+	case -EPIPE:
 	case -EAGAIN:
 		dprintk("RPC: %5u xprt_connect_status: retrying\n", task->tk_pid);
 		break;

net/sunrpc/xprtrdma/rpc_rdma.c

@@ -53,14 +53,6 @@
 # define RPCDBG_FACILITY	RPCDBG_TRANS
 #endif
 
-enum rpcrdma_chunktype {
-	rpcrdma_noch = 0,
-	rpcrdma_readch,
-	rpcrdma_areadch,
-	rpcrdma_writech,
-	rpcrdma_replych
-};
-
 #ifdef RPC_DEBUG
 static const char transfertypes[][12] = {
 	"pure inline",	/* no chunks */
@@ -279,12 +271,36 @@ rpcrdma_create_chunks(struct rpc_rqst *rqst, struct xdr_buf *target,
 	return (unsigned char *)iptr - (unsigned char *)headerp;
 
 out:
-	for (pos = 0; nchunks--;)
-		pos += rpcrdma_deregister_external(
-				&req->rl_segments[pos], r_xprt);
+	if (r_xprt->rx_ia.ri_memreg_strategy != RPCRDMA_FRMR) {
+		for (pos = 0; nchunks--;)
+			pos += rpcrdma_deregister_external(
+					&req->rl_segments[pos], r_xprt);
+	}
 	return n;
 }
 
+/*
+ * Marshal chunks. This routine returns the header length
+ * consumed by marshaling.
+ *
+ * Returns positive RPC/RDMA header size, or negative errno.
+ */
+
+ssize_t
+rpcrdma_marshal_chunks(struct rpc_rqst *rqst, ssize_t result)
+{
+	struct rpcrdma_req *req = rpcr_to_rdmar(rqst);
+	struct rpcrdma_msg *headerp = (struct rpcrdma_msg *)req->rl_base;
+
+	if (req->rl_rtype != rpcrdma_noch)
+		result = rpcrdma_create_chunks(rqst, &rqst->rq_snd_buf,
+					       headerp, req->rl_rtype);
+	else if (req->rl_wtype != rpcrdma_noch)
+		result = rpcrdma_create_chunks(rqst, &rqst->rq_rcv_buf,
+					       headerp, req->rl_wtype);
+	return result;
+}
+
 /*
  * Copy write data inline.
  * This function is used for "small" requests. Data which is passed
@@ -377,7 +393,6 @@ rpcrdma_marshal_req(struct rpc_rqst *rqst)
 	char *base;
 	size_t rpclen, padlen;
 	ssize_t hdrlen;
-	enum rpcrdma_chunktype rtype, wtype;
 	struct rpcrdma_msg *headerp;
 
 	/*
@@ -415,13 +430,13 @@ rpcrdma_marshal_req(struct rpc_rqst *rqst)
 	 * into pages; otherwise use reply chunks.
 	 */
 	if (rqst->rq_rcv_buf.buflen <= RPCRDMA_INLINE_READ_THRESHOLD(rqst))
-		wtype = rpcrdma_noch;
+		req->rl_wtype = rpcrdma_noch;
 	else if (rqst->rq_rcv_buf.page_len == 0)
-		wtype = rpcrdma_replych;
+		req->rl_wtype = rpcrdma_replych;
 	else if (rqst->rq_rcv_buf.flags & XDRBUF_READ)
-		wtype = rpcrdma_writech;
+		req->rl_wtype = rpcrdma_writech;
 	else
-		wtype = rpcrdma_replych;
+		req->rl_wtype = rpcrdma_replych;
 
 	/*
 	 * Chunks needed for arguments?
@@ -438,16 +453,16 @@ rpcrdma_marshal_req(struct rpc_rqst *rqst)
 	 * TBD check NFSv4 setacl
 	 */
 	if (rqst->rq_snd_buf.len <= RPCRDMA_INLINE_WRITE_THRESHOLD(rqst))
-		rtype = rpcrdma_noch;
+		req->rl_rtype = rpcrdma_noch;
 	else if (rqst->rq_snd_buf.page_len == 0)
-		rtype = rpcrdma_areadch;
+		req->rl_rtype = rpcrdma_areadch;
 	else
-		rtype = rpcrdma_readch;
+		req->rl_rtype = rpcrdma_readch;
 
 	/* The following simplification is not true forever */
-	if (rtype != rpcrdma_noch && wtype == rpcrdma_replych)
-		wtype = rpcrdma_noch;
-	if (rtype != rpcrdma_noch && wtype != rpcrdma_noch) {
+	if (req->rl_rtype != rpcrdma_noch && req->rl_wtype == rpcrdma_replych)
+		req->rl_wtype = rpcrdma_noch;
+	if (req->rl_rtype != rpcrdma_noch && req->rl_wtype != rpcrdma_noch) {
 		dprintk("RPC:       %s: cannot marshal multiple chunk lists\n",
 			__func__);
 		return -EIO;
@@ -461,7 +476,7 @@ rpcrdma_marshal_req(struct rpc_rqst *rqst)
 	 * When padding is in use and applies to the transfer, insert
 	 * it and change the message type.
 	 */
-	if (rtype == rpcrdma_noch) {
+	if (req->rl_rtype == rpcrdma_noch) {
 
 		padlen = rpcrdma_inline_pullup(rqst,
 						RPCRDMA_INLINE_PAD_VALUE(rqst));
@@ -476,7 +491,7 @@ rpcrdma_marshal_req(struct rpc_rqst *rqst)
 			headerp->rm_body.rm_padded.rm_pempty[1] = xdr_zero;
 			headerp->rm_body.rm_padded.rm_pempty[2] = xdr_zero;
 			hdrlen += 2 * sizeof(u32); /* extra words in padhdr */
-			if (wtype != rpcrdma_noch) {
+			if (req->rl_wtype != rpcrdma_noch) {
 				dprintk("RPC:       %s: invalid chunk list\n",
 					__func__);
 				return -EIO;
@@ -497,30 +512,18 @@ rpcrdma_marshal_req(struct rpc_rqst *rqst)
 			 * on receive. Therefore, we request a reply chunk
 			 * for non-writes wherever feasible and efficient.
 			 */
-			if (wtype == rpcrdma_noch)
-				wtype = rpcrdma_replych;
+			if (req->rl_wtype == rpcrdma_noch)
+				req->rl_wtype = rpcrdma_replych;
 		}
 	}
 
-	/*
-	 * Marshal chunks. This routine will return the header length
-	 * consumed by marshaling.
-	 */
-	if (rtype != rpcrdma_noch) {
-		hdrlen = rpcrdma_create_chunks(rqst,
-					&rqst->rq_snd_buf, headerp, rtype);
-		wtype = rtype;	/* simplify dprintk */
-
-	} else if (wtype != rpcrdma_noch) {
-		hdrlen = rpcrdma_create_chunks(rqst,
-					&rqst->rq_rcv_buf, headerp, wtype);
-	}
+	hdrlen = rpcrdma_marshal_chunks(rqst, hdrlen);
 	if (hdrlen < 0)
 		return hdrlen;
 
 	dprintk("RPC:       %s: %s: hdrlen %zd rpclen %zd padlen %zd"
 		" headerp 0x%p base 0x%p lkey 0x%x\n",
-		__func__, transfertypes[wtype], hdrlen, rpclen, padlen,
+		__func__, transfertypes[req->rl_wtype], hdrlen, rpclen, padlen,
 		headerp, base, req->rl_iov.lkey);
 
 	/*

net/sunrpc/xprtrdma/transport.c

@@ -296,7 +296,6 @@ xprt_setup_rdma(struct xprt_create *args)
 
 	xprt->resvport = 0;		/* privileged port not needed */
 	xprt->tsh_size = 0;		/* RPC-RDMA handles framing */
-	xprt->max_payload = RPCRDMA_MAX_DATA_SEGS * PAGE_SIZE;
 	xprt->ops = &xprt_rdma_procs;
 
 	/*
@@ -382,6 +381,9 @@ xprt_setup_rdma(struct xprt_create *args)
 	new_ep->rep_xprt = xprt;
 
 	xprt_rdma_format_addresses(xprt);
+	xprt->max_payload = rpcrdma_max_payload(new_xprt);
+	dprintk("RPC:       %s: transport data payload maximum: %zu bytes\n",
+		__func__, xprt->max_payload);
 
 	if (!try_module_get(THIS_MODULE))
 		goto out4;
@@ -412,7 +414,7 @@ xprt_rdma_close(struct rpc_xprt *xprt)
 	if (r_xprt->rx_ep.rep_connected > 0)
 		xprt->reestablish_timeout = 0;
 	xprt_disconnect_done(xprt);
-	(void) rpcrdma_ep_disconnect(&r_xprt->rx_ep, &r_xprt->rx_ia);
+	rpcrdma_ep_disconnect(&r_xprt->rx_ep, &r_xprt->rx_ia);
 }
 
 static void
@@ -595,13 +597,14 @@ xprt_rdma_send_request(struct rpc_task *task)
 	struct rpc_xprt *xprt = rqst->rq_xprt;
 	struct rpcrdma_req *req = rpcr_to_rdmar(rqst);
 	struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt);
-	int rc;
+	int rc = 0;
 
-	if (req->rl_niovs == 0) {
+	if (req->rl_niovs == 0)
 		rc = rpcrdma_marshal_req(rqst);
-		if (rc < 0)
-			goto failed_marshal;
-	}
+	else if (r_xprt->rx_ia.ri_memreg_strategy == RPCRDMA_FRMR)
+		rc = rpcrdma_marshal_chunks(rqst, 0);
+	if (rc < 0)
+		goto failed_marshal;
 
 	if (req->rl_reply == NULL) 		/* e.g. reconnection */
 		rpcrdma_recv_buffer_get(req);

net/sunrpc/xprtrdma/verbs.c

@@ -61,6 +61,8 @@
 # define RPCDBG_FACILITY	RPCDBG_TRANS
 #endif
 
+static void rpcrdma_reset_frmrs(struct rpcrdma_ia *);
+
 /*
  * internal functions
  */
@@ -103,17 +105,6 @@ rpcrdma_run_tasklet(unsigned long data)
 
 static DECLARE_TASKLET(rpcrdma_tasklet_g, rpcrdma_run_tasklet, 0UL);
 
-static inline void
-rpcrdma_schedule_tasklet(struct rpcrdma_rep *rep)
-{
-	unsigned long flags;
-
-	spin_lock_irqsave(&rpcrdma_tk_lock_g, flags);
-	list_add_tail(&rep->rr_list, &rpcrdma_tasklets_g);
-	spin_unlock_irqrestore(&rpcrdma_tk_lock_g, flags);
-	tasklet_schedule(&rpcrdma_tasklet_g);
-}
-
 static void
 rpcrdma_qp_async_error_upcall(struct ib_event *event, void *context)
 {
@@ -153,12 +144,7 @@ rpcrdma_sendcq_process_wc(struct ib_wc *wc)
 	if (wc->wr_id == 0ULL)
 		return;
 	if (wc->status != IB_WC_SUCCESS)
-		return;
-
-	if (wc->opcode == IB_WC_FAST_REG_MR)
-		frmr->r.frmr.state = FRMR_IS_VALID;
-	else if (wc->opcode == IB_WC_LOCAL_INV)
-		frmr->r.frmr.state = FRMR_IS_INVALID;
+		frmr->r.frmr.fr_state = FRMR_IS_STALE;
 }
 
 static int
@@ -217,7 +203,7 @@ rpcrdma_sendcq_upcall(struct ib_cq *cq, void *cq_context)
 }
 
 static void
-rpcrdma_recvcq_process_wc(struct ib_wc *wc)
+rpcrdma_recvcq_process_wc(struct ib_wc *wc, struct list_head *sched_list)
 {
 	struct rpcrdma_rep *rep =
 			(struct rpcrdma_rep *)(unsigned long)wc->wr_id;
@@ -248,28 +234,38 @@ rpcrdma_recvcq_process_wc(struct ib_wc *wc)
 	}
 
 out_schedule:
-	rpcrdma_schedule_tasklet(rep);
+	list_add_tail(&rep->rr_list, sched_list);
 }
 
 static int
 rpcrdma_recvcq_poll(struct ib_cq *cq, struct rpcrdma_ep *ep)
 {
+	struct list_head sched_list;
 	struct ib_wc *wcs;
 	int budget, count, rc;
+	unsigned long flags;
 
+	INIT_LIST_HEAD(&sched_list);
 	budget = RPCRDMA_WC_BUDGET / RPCRDMA_POLLSIZE;
 	do {
 		wcs = ep->rep_recv_wcs;
 
 		rc = ib_poll_cq(cq, RPCRDMA_POLLSIZE, wcs);
 		if (rc <= 0)
-			return rc;
+			goto out_schedule;
 
 		count = rc;
 		while (count-- > 0)
-			rpcrdma_recvcq_process_wc(wcs++);
+			rpcrdma_recvcq_process_wc(wcs++, &sched_list);
 	} while (rc == RPCRDMA_POLLSIZE && --budget);
-	return 0;
+	rc = 0;
+
+out_schedule:
+	spin_lock_irqsave(&rpcrdma_tk_lock_g, flags);
+	list_splice_tail(&sched_list, &rpcrdma_tasklets_g);
+	spin_unlock_irqrestore(&rpcrdma_tk_lock_g, flags);
+	tasklet_schedule(&rpcrdma_tasklet_g);
+	return rc;
 }
 
 /*
@@ -310,6 +306,13 @@ rpcrdma_recvcq_upcall(struct ib_cq *cq, void *cq_context)
 	rpcrdma_recvcq_poll(cq, ep);
 }
 
+static void
+rpcrdma_flush_cqs(struct rpcrdma_ep *ep)
+{
+	rpcrdma_recvcq_upcall(ep->rep_attr.recv_cq, ep);
+	rpcrdma_sendcq_upcall(ep->rep_attr.send_cq, ep);
+}
+
 #ifdef RPC_DEBUG
 static const char * const conn[] = {
 	"address resolved",
@@ -323,8 +326,16 @@ static const char * const conn[] = {
 	"rejected",
 	"established",
 	"disconnected",
-	"device removal"
+	"device removal",
+	"multicast join",
+	"multicast error",
+	"address change",
+	"timewait exit",
 };
+
+#define CONNECTION_MSG(status)						\
+	((status) < ARRAY_SIZE(conn) ?					\
+		conn[(status)] : "unrecognized connection error")
 #endif
 
 static int
@@ -382,23 +393,18 @@ rpcrdma_conn_upcall(struct rdma_cm_id *id, struct rdma_cm_event *event)
 	case RDMA_CM_EVENT_DEVICE_REMOVAL:
 		connstate = -ENODEV;
 connected:
-		dprintk("RPC:       %s: %s: %pI4:%u (ep 0x%p event 0x%x)\n",
-			__func__,
-			(event->event <= 11) ? conn[event->event] :
-						"unknown connection error",
-			&addr->sin_addr.s_addr,
-			ntohs(addr->sin_port),
-			ep, event->event);
 		atomic_set(&rpcx_to_rdmax(ep->rep_xprt)->rx_buf.rb_credits, 1);
 		dprintk("RPC:       %s: %sconnected\n",
 					__func__, connstate > 0 ? "" : "dis");
 		ep->rep_connected = connstate;
 		ep->rep_func(ep);
 		wake_up_all(&ep->rep_connect_wait);
-		break;
+		/*FALLTHROUGH*/
 	default:
-		dprintk("RPC:       %s: unexpected CM event %d\n",
-			__func__, event->event);
+		dprintk("RPC:       %s: %pI4:%u (ep 0x%p): %s\n",
+			__func__, &addr->sin_addr.s_addr,
+			ntohs(addr->sin_port), ep,
+			CONNECTION_MSG(event->event));
 		break;
 	}
 
@@ -558,12 +564,7 @@ rpcrdma_ia_open(struct rpcrdma_xprt *xprt, struct sockaddr *addr, int memreg)
 		if (!ia->ri_id->device->alloc_fmr) {
 			dprintk("RPC:       %s: MTHCAFMR registration "
 				"not supported by HCA\n", __func__);
-#if RPCRDMA_PERSISTENT_REGISTRATION
 			memreg = RPCRDMA_ALLPHYSICAL;
-#else
-			rc = -ENOMEM;
-			goto out2;
-#endif
 		}
 	}
 
@@ -578,20 +579,16 @@ rpcrdma_ia_open(struct rpcrdma_xprt *xprt, struct sockaddr *addr, int memreg)
 	switch (memreg) {
 	case RPCRDMA_FRMR:
 		break;
-#if RPCRDMA_PERSISTENT_REGISTRATION
 	case RPCRDMA_ALLPHYSICAL:
 		mem_priv = IB_ACCESS_LOCAL_WRITE |
 				IB_ACCESS_REMOTE_WRITE |
 				IB_ACCESS_REMOTE_READ;
 		goto register_setup;
-#endif
 	case RPCRDMA_MTHCAFMR:
 		if (ia->ri_have_dma_lkey)
 			break;
 		mem_priv = IB_ACCESS_LOCAL_WRITE;
-#if RPCRDMA_PERSISTENT_REGISTRATION
 	register_setup:
-#endif
 		ia->ri_bind_mem = ib_get_dma_mr(ia->ri_pd, mem_priv);
 		if (IS_ERR(ia->ri_bind_mem)) {
 			printk(KERN_ALERT "%s: ib_get_dma_mr for "
@@ -613,6 +610,7 @@ rpcrdma_ia_open(struct rpcrdma_xprt *xprt, struct sockaddr *addr, int memreg)
 	/* Else will do memory reg/dereg for each chunk */
 	ia->ri_memreg_strategy = memreg;
 
+	rwlock_init(&ia->ri_qplock);
 	return 0;
 out2:
 	rdma_destroy_id(ia->ri_id);
@@ -826,10 +824,7 @@ rpcrdma_ep_destroy(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia)
 	cancel_delayed_work_sync(&ep->rep_connect_worker);
 
 	if (ia->ri_id->qp) {
-		rc = rpcrdma_ep_disconnect(ep, ia);
-		if (rc)
-			dprintk("RPC:       %s: rpcrdma_ep_disconnect"
-				" returned %i\n", __func__, rc);
+		rpcrdma_ep_disconnect(ep, ia);
 		rdma_destroy_qp(ia->ri_id);
 		ia->ri_id->qp = NULL;
 	}
@@ -859,7 +854,7 @@ rpcrdma_ep_destroy(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia)
 int
 rpcrdma_ep_connect(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia)
 {
-	struct rdma_cm_id *id;
+	struct rdma_cm_id *id, *old;
 	int rc = 0;
 	int retry_count = 0;
 
@@ -867,13 +862,12 @@ rpcrdma_ep_connect(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia)
 		struct rpcrdma_xprt *xprt;
 retry:
 		dprintk("RPC:       %s: reconnecting...\n", __func__);
-		rc = rpcrdma_ep_disconnect(ep, ia);
-		if (rc && rc != -ENOTCONN)
-			dprintk("RPC:       %s: rpcrdma_ep_disconnect"
-				" status %i\n", __func__, rc);
 
-		rpcrdma_clean_cq(ep->rep_attr.recv_cq);
-		rpcrdma_clean_cq(ep->rep_attr.send_cq);
+		rpcrdma_ep_disconnect(ep, ia);
+		rpcrdma_flush_cqs(ep);
+
+		if (ia->ri_memreg_strategy == RPCRDMA_FRMR)
+			rpcrdma_reset_frmrs(ia);
 
 		xprt = container_of(ia, struct rpcrdma_xprt, rx_ia);
 		id = rpcrdma_create_id(xprt, ia,
@@ -905,9 +899,14 @@ retry:
 			rc = -ENETUNREACH;
 			goto out;
 		}
-		rdma_destroy_qp(ia->ri_id);
-		rdma_destroy_id(ia->ri_id);
+
+		write_lock(&ia->ri_qplock);
+		old = ia->ri_id;
 		ia->ri_id = id;
+		write_unlock(&ia->ri_qplock);
+
+		rdma_destroy_qp(old);
+		rdma_destroy_id(old);
 	} else {
 		dprintk("RPC:       %s: connecting...\n", __func__);
 		rc = rdma_create_qp(ia->ri_id, ia->ri_pd, &ep->rep_attr);
@@ -974,13 +973,12 @@ out:
  * This call is not reentrant, and must not be made in parallel
  * on the same endpoint.
  */
-int
+void
 rpcrdma_ep_disconnect(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia)
 {
 	int rc;
 
-	rpcrdma_clean_cq(ep->rep_attr.recv_cq);
-	rpcrdma_clean_cq(ep->rep_attr.send_cq);
+	rpcrdma_flush_cqs(ep);
 	rc = rdma_disconnect(ia->ri_id);
 	if (!rc) {
 		/* returns without wait if not connected */
@@ -992,12 +990,93 @@ rpcrdma_ep_disconnect(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia)
 		dprintk("RPC:       %s: rdma_disconnect %i\n", __func__, rc);
 		ep->rep_connected = rc;
 	}
+}
+
+static int
+rpcrdma_init_fmrs(struct rpcrdma_ia *ia, struct rpcrdma_buffer *buf)
+{
+	int mr_access_flags = IB_ACCESS_REMOTE_WRITE | IB_ACCESS_REMOTE_READ;
+	struct ib_fmr_attr fmr_attr = {
+		.max_pages	= RPCRDMA_MAX_DATA_SEGS,
+		.max_maps	= 1,
+		.page_shift	= PAGE_SHIFT
+	};
+	struct rpcrdma_mw *r;
+	int i, rc;
+
+	i = (buf->rb_max_requests + 1) * RPCRDMA_MAX_SEGS;
+	dprintk("RPC:       %s: initalizing %d FMRs\n", __func__, i);
+
+	while (i--) {
+		r = kzalloc(sizeof(*r), GFP_KERNEL);
+		if (r == NULL)
+			return -ENOMEM;
+
+		r->r.fmr = ib_alloc_fmr(ia->ri_pd, mr_access_flags, &fmr_attr);
+		if (IS_ERR(r->r.fmr)) {
+			rc = PTR_ERR(r->r.fmr);
+			dprintk("RPC:       %s: ib_alloc_fmr failed %i\n",
+				__func__, rc);
+			goto out_free;
+		}
+
+		list_add(&r->mw_list, &buf->rb_mws);
+		list_add(&r->mw_all, &buf->rb_all);
+	}
+	return 0;
+
+out_free:
+	kfree(r);
+	return rc;
+}
+
+static int
+rpcrdma_init_frmrs(struct rpcrdma_ia *ia, struct rpcrdma_buffer *buf)
+{
+	struct rpcrdma_frmr *f;
+	struct rpcrdma_mw *r;
+	int i, rc;
+
+	i = (buf->rb_max_requests + 1) * RPCRDMA_MAX_SEGS;
+	dprintk("RPC:       %s: initalizing %d FRMRs\n", __func__, i);
+
+	while (i--) {
+		r = kzalloc(sizeof(*r), GFP_KERNEL);
+		if (r == NULL)
+			return -ENOMEM;
+		f = &r->r.frmr;
+
+		f->fr_mr = ib_alloc_fast_reg_mr(ia->ri_pd,
+						ia->ri_max_frmr_depth);
+		if (IS_ERR(f->fr_mr)) {
+			rc = PTR_ERR(f->fr_mr);
+			dprintk("RPC:       %s: ib_alloc_fast_reg_mr "
+				"failed %i\n", __func__, rc);
+			goto out_free;
+		}
+
+		f->fr_pgl = ib_alloc_fast_reg_page_list(ia->ri_id->device,
+							ia->ri_max_frmr_depth);
+		if (IS_ERR(f->fr_pgl)) {
+			rc = PTR_ERR(f->fr_pgl);
+			dprintk("RPC:       %s: ib_alloc_fast_reg_page_list "
+				"failed %i\n", __func__, rc);
+
+			ib_dereg_mr(f->fr_mr);
+			goto out_free;
+		}
+
+		list_add(&r->mw_list, &buf->rb_mws);
+		list_add(&r->mw_all, &buf->rb_all);
+	}
+
+	return 0;
+
+out_free:
+	kfree(r);
 	return rc;
 }
 
-/*
- * Initialize buffer memory
- */
 int
 rpcrdma_buffer_create(struct rpcrdma_buffer *buf, struct rpcrdma_ep *ep,
 	struct rpcrdma_ia *ia, struct rpcrdma_create_data_internal *cdata)
@@ -1005,7 +1084,6 @@ rpcrdma_buffer_create(struct rpcrdma_buffer *buf, struct rpcrdma_ep *ep,
 	char *p;
 	size_t len, rlen, wlen;
 	int i, rc;
-	struct rpcrdma_mw *r;
 
 	buf->rb_max_requests = cdata->max_requests;
 	spin_lock_init(&buf->rb_lock);
@@ -1016,28 +1094,12 @@ rpcrdma_buffer_create(struct rpcrdma_buffer *buf, struct rpcrdma_ep *ep,
 	 *   2.  arrays of struct rpcrdma_req to fill in pointers
 	 *   3.  array of struct rpcrdma_rep for replies
 	 *   4.  padding, if any
-	 *   5.  mw's, fmr's or frmr's, if any
 	 * Send/recv buffers in req/rep need to be registered
 	 */
-
 	len = buf->rb_max_requests *
 		(sizeof(struct rpcrdma_req *) + sizeof(struct rpcrdma_rep *));
 	len += cdata->padding;
-	switch (ia->ri_memreg_strategy) {
-	case RPCRDMA_FRMR:
-		len += buf->rb_max_requests * RPCRDMA_MAX_SEGS *
-				sizeof(struct rpcrdma_mw);
-		break;
-	case RPCRDMA_MTHCAFMR:
-		/* TBD we are perhaps overallocating here */
-		len += (buf->rb_max_requests + 1) * RPCRDMA_MAX_SEGS *
-				sizeof(struct rpcrdma_mw);
-		break;
-	default:
-		break;
-	}
 
-	/* allocate 1, 4 and 5 in one shot */
 	p = kzalloc(len, GFP_KERNEL);
 	if (p == NULL) {
 		dprintk("RPC:       %s: req_t/rep_t/pad kzalloc(%zd) failed\n",
@@ -1064,51 +1126,17 @@ rpcrdma_buffer_create(struct rpcrdma_buffer *buf, struct rpcrdma_ep *ep,
 	p += cdata->padding;
 
 	INIT_LIST_HEAD(&buf->rb_mws);
-	r = (struct rpcrdma_mw *)p;
+	INIT_LIST_HEAD(&buf->rb_all);
 	switch (ia->ri_memreg_strategy) {
 	case RPCRDMA_FRMR:
-		for (i = buf->rb_max_requests * RPCRDMA_MAX_SEGS; i; i--) {
-			r->r.frmr.fr_mr = ib_alloc_fast_reg_mr(ia->ri_pd,
-						ia->ri_max_frmr_depth);
-			if (IS_ERR(r->r.frmr.fr_mr)) {
-				rc = PTR_ERR(r->r.frmr.fr_mr);
-				dprintk("RPC:       %s: ib_alloc_fast_reg_mr"
-					" failed %i\n", __func__, rc);
-				goto out;
-			}
-			r->r.frmr.fr_pgl = ib_alloc_fast_reg_page_list(
-						ia->ri_id->device,
-						ia->ri_max_frmr_depth);
-			if (IS_ERR(r->r.frmr.fr_pgl)) {
-				rc = PTR_ERR(r->r.frmr.fr_pgl);
-				dprintk("RPC:       %s: "
-					"ib_alloc_fast_reg_page_list "
-					"failed %i\n", __func__, rc);
-
-				ib_dereg_mr(r->r.frmr.fr_mr);
-				goto out;
-			}
-			list_add(&r->mw_list, &buf->rb_mws);
-			++r;
-		}
+		rc = rpcrdma_init_frmrs(ia, buf);
+		if (rc)
+			goto out;
 		break;
 	case RPCRDMA_MTHCAFMR:
-		/* TBD we are perhaps overallocating here */
-		for (i = (buf->rb_max_requests+1) * RPCRDMA_MAX_SEGS; i; i--) {
-			static struct ib_fmr_attr fa =
-				{ RPCRDMA_MAX_DATA_SEGS, 1, PAGE_SHIFT };
-			r->r.fmr = ib_alloc_fmr(ia->ri_pd,
-				IB_ACCESS_REMOTE_WRITE | IB_ACCESS_REMOTE_READ,
-				&fa);
-			if (IS_ERR(r->r.fmr)) {
-				rc = PTR_ERR(r->r.fmr);
-				dprintk("RPC:       %s: ib_alloc_fmr"
-					" failed %i\n", __func__, rc);
-				goto out;
-			}
-			list_add(&r->mw_list, &buf->rb_mws);
-			++r;
-		}
+		rc = rpcrdma_init_fmrs(ia, buf);
+		if (rc)
+			goto out;
 		break;
 	default:
 		break;
@@ -1176,24 +1204,57 @@ out:
 	return rc;
 }
 
-/*
- * Unregister and destroy buffer memory. Need to deal with
- * partial initialization, so it's callable from failed create.
- * Must be called before destroying endpoint, as registrations
- * reference it.
- */
+static void
+rpcrdma_destroy_fmrs(struct rpcrdma_buffer *buf)
+{
+	struct rpcrdma_mw *r;
+	int rc;
+
+	while (!list_empty(&buf->rb_all)) {
+		r = list_entry(buf->rb_all.next, struct rpcrdma_mw, mw_all);
+		list_del(&r->mw_all);
+		list_del(&r->mw_list);
+
+		rc = ib_dealloc_fmr(r->r.fmr);
+		if (rc)
+			dprintk("RPC:       %s: ib_dealloc_fmr failed %i\n",
+				__func__, rc);
+
+		kfree(r);
+	}
+}
+
+static void
+rpcrdma_destroy_frmrs(struct rpcrdma_buffer *buf)
+{
+	struct rpcrdma_mw *r;
+	int rc;
+
+	while (!list_empty(&buf->rb_all)) {
+		r = list_entry(buf->rb_all.next, struct rpcrdma_mw, mw_all);
+		list_del(&r->mw_all);
+		list_del(&r->mw_list);
+
+		rc = ib_dereg_mr(r->r.frmr.fr_mr);
+		if (rc)
+			dprintk("RPC:       %s: ib_dereg_mr failed %i\n",
+				__func__, rc);
+		ib_free_fast_reg_page_list(r->r.frmr.fr_pgl);
+
+		kfree(r);
+	}
+}
+
 void
 rpcrdma_buffer_destroy(struct rpcrdma_buffer *buf)
 {
-	int rc, i;
 	struct rpcrdma_ia *ia = rdmab_to_ia(buf);
-	struct rpcrdma_mw *r;
+	int i;
 
 	/* clean up in reverse order from create
 	 *   1.  recv mr memory (mr free, then kfree)
 	 *   2.  send mr memory (mr free, then kfree)
-	 *   3.  padding (if any) [moved to rpcrdma_ep_destroy]
-	 *   4.  arrays
+	 *   3.  MWs
 	 */
 	dprintk("RPC:       %s: entering\n", __func__);
 
@@ -1212,34 +1273,217 @@ rpcrdma_buffer_destroy(struct rpcrdma_buffer *buf)
 		}
 	}
 
+	switch (ia->ri_memreg_strategy) {
+	case RPCRDMA_FRMR:
+		rpcrdma_destroy_frmrs(buf);
+		break;
+	case RPCRDMA_MTHCAFMR:
+		rpcrdma_destroy_fmrs(buf);
+		break;
+	default:
+		break;
+	}
+
+	kfree(buf->rb_pool);
+}
+
+/* After a disconnect, a flushed FAST_REG_MR can leave an FRMR in
+ * an unusable state. Find FRMRs in this state and dereg / reg
+ * each.  FRMRs that are VALID and attached to an rpcrdma_req are
+ * also torn down.
+ *
+ * This gives all in-use FRMRs a fresh rkey and leaves them INVALID.
+ *
+ * This is invoked only in the transport connect worker in order
+ * to serialize with rpcrdma_register_frmr_external().
+ */
+static void
+rpcrdma_reset_frmrs(struct rpcrdma_ia *ia)
+{
+	struct rpcrdma_xprt *r_xprt =
+				container_of(ia, struct rpcrdma_xprt, rx_ia);
+	struct rpcrdma_buffer *buf = &r_xprt->rx_buf;
+	struct list_head *pos;
+	struct rpcrdma_mw *r;
+	int rc;
+
+	list_for_each(pos, &buf->rb_all) {
+		r = list_entry(pos, struct rpcrdma_mw, mw_all);
+
+		if (r->r.frmr.fr_state == FRMR_IS_INVALID)
+			continue;
+
+		rc = ib_dereg_mr(r->r.frmr.fr_mr);
+		if (rc)
+			dprintk("RPC:       %s: ib_dereg_mr failed %i\n",
+				__func__, rc);
+		ib_free_fast_reg_page_list(r->r.frmr.fr_pgl);
+
+		r->r.frmr.fr_mr = ib_alloc_fast_reg_mr(ia->ri_pd,
+					ia->ri_max_frmr_depth);
+		if (IS_ERR(r->r.frmr.fr_mr)) {
+			rc = PTR_ERR(r->r.frmr.fr_mr);
+			dprintk("RPC:       %s: ib_alloc_fast_reg_mr"
+				" failed %i\n", __func__, rc);
+			continue;
+		}
+		r->r.frmr.fr_pgl = ib_alloc_fast_reg_page_list(
+					ia->ri_id->device,
+					ia->ri_max_frmr_depth);
+		if (IS_ERR(r->r.frmr.fr_pgl)) {
+			rc = PTR_ERR(r->r.frmr.fr_pgl);
+			dprintk("RPC:       %s: "
+				"ib_alloc_fast_reg_page_list "
+				"failed %i\n", __func__, rc);
+
+			ib_dereg_mr(r->r.frmr.fr_mr);
+			continue;
+		}
+		r->r.frmr.fr_state = FRMR_IS_INVALID;
+	}
+}
+
+/* "*mw" can be NULL when rpcrdma_buffer_get_mrs() fails, leaving
+ * some req segments uninitialized.
+ */
+static void
+rpcrdma_buffer_put_mr(struct rpcrdma_mw **mw, struct rpcrdma_buffer *buf)
+{
+	if (*mw) {
+		list_add_tail(&(*mw)->mw_list, &buf->rb_mws);
+		*mw = NULL;
+	}
+}
+
+/* Cycle mw's back in reverse order, and "spin" them.
+ * This delays and scrambles reuse as much as possible.
+ */
+static void
+rpcrdma_buffer_put_mrs(struct rpcrdma_req *req, struct rpcrdma_buffer *buf)
+{
+	struct rpcrdma_mr_seg *seg = req->rl_segments;
+	struct rpcrdma_mr_seg *seg1 = seg;
+	int i;
+
+	for (i = 1, seg++; i < RPCRDMA_MAX_SEGS; seg++, i++)
+		rpcrdma_buffer_put_mr(&seg->mr_chunk.rl_mw, buf);
+	rpcrdma_buffer_put_mr(&seg1->mr_chunk.rl_mw, buf);
+}
+
+static void
+rpcrdma_buffer_put_sendbuf(struct rpcrdma_req *req, struct rpcrdma_buffer *buf)
+{
+	buf->rb_send_bufs[--buf->rb_send_index] = req;
+	req->rl_niovs = 0;
+	if (req->rl_reply) {
+		buf->rb_recv_bufs[--buf->rb_recv_index] = req->rl_reply;
+		req->rl_reply->rr_func = NULL;
+		req->rl_reply = NULL;
+	}
+}
+
+/* rpcrdma_unmap_one() was already done by rpcrdma_deregister_frmr_external().
+ * Redo only the ib_post_send().
+ */
+static void
+rpcrdma_retry_local_inv(struct rpcrdma_mw *r, struct rpcrdma_ia *ia)
+{
+	struct rpcrdma_xprt *r_xprt =
+				container_of(ia, struct rpcrdma_xprt, rx_ia);
+	struct ib_send_wr invalidate_wr, *bad_wr;
+	int rc;
+
+	dprintk("RPC:       %s: FRMR %p is stale\n", __func__, r);
+
+	/* When this FRMR is re-inserted into rb_mws, it is no longer stale */
+	r->r.frmr.fr_state = FRMR_IS_INVALID;
+
+	memset(&invalidate_wr, 0, sizeof(invalidate_wr));
+	invalidate_wr.wr_id = (unsigned long)(void *)r;
+	invalidate_wr.opcode = IB_WR_LOCAL_INV;
+	invalidate_wr.ex.invalidate_rkey = r->r.frmr.fr_mr->rkey;
+	DECR_CQCOUNT(&r_xprt->rx_ep);
+
+	dprintk("RPC:       %s: frmr %p invalidating rkey %08x\n",
+		__func__, r, r->r.frmr.fr_mr->rkey);
+
+	read_lock(&ia->ri_qplock);
+	rc = ib_post_send(ia->ri_id->qp, &invalidate_wr, &bad_wr);
+	read_unlock(&ia->ri_qplock);
+	if (rc) {
+		/* Force rpcrdma_buffer_get() to retry */
+		r->r.frmr.fr_state = FRMR_IS_STALE;
+		dprintk("RPC:       %s: ib_post_send failed, %i\n",
+			__func__, rc);
+	}
+}
+
+static void
+rpcrdma_retry_flushed_linv(struct list_head *stale,
+			   struct rpcrdma_buffer *buf)
+{
+	struct rpcrdma_ia *ia = rdmab_to_ia(buf);
+	struct list_head *pos;
+	struct rpcrdma_mw *r;
+	unsigned long flags;
+
+	list_for_each(pos, stale) {
+		r = list_entry(pos, struct rpcrdma_mw, mw_list);
+		rpcrdma_retry_local_inv(r, ia);
+	}
+
+	spin_lock_irqsave(&buf->rb_lock, flags);
+	list_splice_tail(stale, &buf->rb_mws);
+	spin_unlock_irqrestore(&buf->rb_lock, flags);
+}
+
+static struct rpcrdma_req *
+rpcrdma_buffer_get_frmrs(struct rpcrdma_req *req, struct rpcrdma_buffer *buf,
+			 struct list_head *stale)
+{
+	struct rpcrdma_mw *r;
+	int i;
+
+	i = RPCRDMA_MAX_SEGS - 1;
 	while (!list_empty(&buf->rb_mws)) {
 		r = list_entry(buf->rb_mws.next,
-			struct rpcrdma_mw, mw_list);
+			       struct rpcrdma_mw, mw_list);
 		list_del(&r->mw_list);
-		switch (ia->ri_memreg_strategy) {
-		case RPCRDMA_FRMR:
-			rc = ib_dereg_mr(r->r.frmr.fr_mr);
-			if (rc)
-				dprintk("RPC:       %s:"
-					" ib_dereg_mr"
-					" failed %i\n",
-					__func__, rc);
-			ib_free_fast_reg_page_list(r->r.frmr.fr_pgl);
-			break;
-		case RPCRDMA_MTHCAFMR:
-			rc = ib_dealloc_fmr(r->r.fmr);
-			if (rc)
-				dprintk("RPC:       %s:"
-					" ib_dealloc_fmr"
-					" failed %i\n",
-					__func__, rc);
-			break;
-		default:
-			break;
+		if (r->r.frmr.fr_state == FRMR_IS_STALE) {
+			list_add(&r->mw_list, stale);
+			continue;
 		}
+		req->rl_segments[i].mr_chunk.rl_mw = r;
+		if (unlikely(i-- == 0))
+			return req;	/* Success */
 	}
 
-	kfree(buf->rb_pool);
+	/* Not enough entries on rb_mws for this req */
+	rpcrdma_buffer_put_sendbuf(req, buf);
+	rpcrdma_buffer_put_mrs(req, buf);
+	return NULL;
+}
+
+static struct rpcrdma_req *
+rpcrdma_buffer_get_fmrs(struct rpcrdma_req *req, struct rpcrdma_buffer *buf)
+{
+	struct rpcrdma_mw *r;
+	int i;
+
+	i = RPCRDMA_MAX_SEGS - 1;
+	while (!list_empty(&buf->rb_mws)) {
+		r = list_entry(buf->rb_mws.next,
+			       struct rpcrdma_mw, mw_list);
+		list_del(&r->mw_list);
+		req->rl_segments[i].mr_chunk.rl_mw = r;
+		if (unlikely(i-- == 0))
+			return req;	/* Success */
+	}
+
+	/* Not enough entries on rb_mws for this req */
+	rpcrdma_buffer_put_sendbuf(req, buf);
+	rpcrdma_buffer_put_mrs(req, buf);
+	return NULL;
 }
 
 /*
@@ -1254,10 +1498,10 @@ rpcrdma_buffer_destroy(struct rpcrdma_buffer *buf)
 struct rpcrdma_req *
 rpcrdma_buffer_get(struct rpcrdma_buffer *buffers)
 {
+	struct rpcrdma_ia *ia = rdmab_to_ia(buffers);
+	struct list_head stale;
 	struct rpcrdma_req *req;
 	unsigned long flags;
-	int i;
-	struct rpcrdma_mw *r;
 
 	spin_lock_irqsave(&buffers->rb_lock, flags);
 	if (buffers->rb_send_index == buffers->rb_max_requests) {
@@ -1277,16 +1521,21 @@ rpcrdma_buffer_get(struct rpcrdma_buffer *buffers)
 		buffers->rb_recv_bufs[buffers->rb_recv_index++] = NULL;
 	}
 	buffers->rb_send_bufs[buffers->rb_send_index++] = NULL;
-	if (!list_empty(&buffers->rb_mws)) {
-		i = RPCRDMA_MAX_SEGS - 1;
-		do {
-			r = list_entry(buffers->rb_mws.next,
-					struct rpcrdma_mw, mw_list);
-			list_del(&r->mw_list);
-			req->rl_segments[i].mr_chunk.rl_mw = r;
-		} while (--i >= 0);
+
+	INIT_LIST_HEAD(&stale);
+	switch (ia->ri_memreg_strategy) {
+	case RPCRDMA_FRMR:
+		req = rpcrdma_buffer_get_frmrs(req, buffers, &stale);
+		break;
+	case RPCRDMA_MTHCAFMR:
+		req = rpcrdma_buffer_get_fmrs(req, buffers);
+		break;
+	default:
+		break;
 	}
 	spin_unlock_irqrestore(&buffers->rb_lock, flags);
+	if (!list_empty(&stale))
+		rpcrdma_retry_flushed_linv(&stale, buffers);
 	return req;
 }
 
@@ -1299,34 +1548,14 @@ rpcrdma_buffer_put(struct rpcrdma_req *req)
 {
 	struct rpcrdma_buffer *buffers = req->rl_buffer;
 	struct rpcrdma_ia *ia = rdmab_to_ia(buffers);
-	int i;
 	unsigned long flags;
 
 	spin_lock_irqsave(&buffers->rb_lock, flags);
-	buffers->rb_send_bufs[--buffers->rb_send_index] = req;
-	req->rl_niovs = 0;
-	if (req->rl_reply) {
-		buffers->rb_recv_bufs[--buffers->rb_recv_index] = req->rl_reply;
-		req->rl_reply->rr_func = NULL;
-		req->rl_reply = NULL;
-	}
+	rpcrdma_buffer_put_sendbuf(req, buffers);
 	switch (ia->ri_memreg_strategy) {
 	case RPCRDMA_FRMR:
 	case RPCRDMA_MTHCAFMR:
-		/*
-		 * Cycle mw's back in reverse order, and "spin" them.
-		 * This delays and scrambles reuse as much as possible.
-		 */
-		i = 1;
-		do {
-			struct rpcrdma_mw **mw;
-			mw = &req->rl_segments[i].mr_chunk.rl_mw;
-			list_add_tail(&(*mw)->mw_list, &buffers->rb_mws);
-			*mw = NULL;
-		} while (++i < RPCRDMA_MAX_SEGS);
-		list_add_tail(&req->rl_segments[0].mr_chunk.rl_mw->mw_list,
-					&buffers->rb_mws);
-		req->rl_segments[0].mr_chunk.rl_mw = NULL;
+		rpcrdma_buffer_put_mrs(req, buffers);
 		break;
 	default:
 		break;
@@ -1388,6 +1617,9 @@ rpcrdma_register_internal(struct rpcrdma_ia *ia, void *va, int len,
 	 */
 	iov->addr = ib_dma_map_single(ia->ri_id->device,
 			va, len, DMA_BIDIRECTIONAL);
+	if (ib_dma_mapping_error(ia->ri_id->device, iov->addr))
+		return -ENOMEM;
+
 	iov->length = len;
 
 	if (ia->ri_have_dma_lkey) {
@@ -1483,8 +1715,10 @@ rpcrdma_register_frmr_external(struct rpcrdma_mr_seg *seg,
 			struct rpcrdma_xprt *r_xprt)
 {
 	struct rpcrdma_mr_seg *seg1 = seg;
-	struct ib_send_wr invalidate_wr, frmr_wr, *bad_wr, *post_wr;
-
+	struct rpcrdma_mw *mw = seg1->mr_chunk.rl_mw;
+	struct rpcrdma_frmr *frmr = &mw->r.frmr;
+	struct ib_mr *mr = frmr->fr_mr;
+	struct ib_send_wr fastreg_wr, *bad_wr;
 	u8 key;
 	int len, pageoff;
 	int i, rc;
@@ -1502,8 +1736,7 @@ rpcrdma_register_frmr_external(struct rpcrdma_mr_seg *seg,
 		rpcrdma_map_one(ia, seg, writing);
 		pa = seg->mr_dma;
 		for (seg_len = seg->mr_len; seg_len > 0; seg_len -= PAGE_SIZE) {
-			seg1->mr_chunk.rl_mw->r.frmr.fr_pgl->
-				page_list[page_no++] = pa;
+			frmr->fr_pgl->page_list[page_no++] = pa;
 			pa += PAGE_SIZE;
 		}
 		len += seg->mr_len;
@@ -1515,65 +1748,51 @@ rpcrdma_register_frmr_external(struct rpcrdma_mr_seg *seg,
 			break;
 	}
 	dprintk("RPC:       %s: Using frmr %p to map %d segments\n",
-		__func__, seg1->mr_chunk.rl_mw, i);
-
-	if (unlikely(seg1->mr_chunk.rl_mw->r.frmr.state == FRMR_IS_VALID)) {
-		dprintk("RPC:       %s: frmr %x left valid, posting invalidate.\n",
-			__func__,
-			seg1->mr_chunk.rl_mw->r.frmr.fr_mr->rkey);
-		/* Invalidate before using. */
-		memset(&invalidate_wr, 0, sizeof invalidate_wr);
-		invalidate_wr.wr_id = (unsigned long)(void *)seg1->mr_chunk.rl_mw;
-		invalidate_wr.next = &frmr_wr;
-		invalidate_wr.opcode = IB_WR_LOCAL_INV;
-		invalidate_wr.send_flags = IB_SEND_SIGNALED;
-		invalidate_wr.ex.invalidate_rkey =
-			seg1->mr_chunk.rl_mw->r.frmr.fr_mr->rkey;
-		DECR_CQCOUNT(&r_xprt->rx_ep);
-		post_wr = &invalidate_wr;
-	} else
-		post_wr = &frmr_wr;
-
-	/* Prepare FRMR WR */
-	memset(&frmr_wr, 0, sizeof frmr_wr);
-	frmr_wr.wr_id = (unsigned long)(void *)seg1->mr_chunk.rl_mw;
-	frmr_wr.opcode = IB_WR_FAST_REG_MR;
-	frmr_wr.send_flags = IB_SEND_SIGNALED;
-	frmr_wr.wr.fast_reg.iova_start = seg1->mr_dma;
-	frmr_wr.wr.fast_reg.page_list = seg1->mr_chunk.rl_mw->r.frmr.fr_pgl;
-	frmr_wr.wr.fast_reg.page_list_len = page_no;
-	frmr_wr.wr.fast_reg.page_shift = PAGE_SHIFT;
-	frmr_wr.wr.fast_reg.length = page_no << PAGE_SHIFT;
-	if (frmr_wr.wr.fast_reg.length < len) {
-		while (seg1->mr_nsegs--)
-			rpcrdma_unmap_one(ia, seg++);
-		return -EIO;
+		__func__, mw, i);
+
+	frmr->fr_state = FRMR_IS_VALID;
+
+	memset(&fastreg_wr, 0, sizeof(fastreg_wr));
+	fastreg_wr.wr_id = (unsigned long)(void *)mw;
+	fastreg_wr.opcode = IB_WR_FAST_REG_MR;
+	fastreg_wr.wr.fast_reg.iova_start = seg1->mr_dma;
+	fastreg_wr.wr.fast_reg.page_list = frmr->fr_pgl;
+	fastreg_wr.wr.fast_reg.page_list_len = page_no;
+	fastreg_wr.wr.fast_reg.page_shift = PAGE_SHIFT;
+	fastreg_wr.wr.fast_reg.length = page_no << PAGE_SHIFT;
+	if (fastreg_wr.wr.fast_reg.length < len) {
+		rc = -EIO;
+		goto out_err;
 	}
 
 	/* Bump the key */
-	key = (u8)(seg1->mr_chunk.rl_mw->r.frmr.fr_mr->rkey & 0x000000FF);
-	ib_update_fast_reg_key(seg1->mr_chunk.rl_mw->r.frmr.fr_mr, ++key);
+	key = (u8)(mr->rkey & 0x000000FF);
+	ib_update_fast_reg_key(mr, ++key);
 
-	frmr_wr.wr.fast_reg.access_flags = (writing ?
+	fastreg_wr.wr.fast_reg.access_flags = (writing ?
 				IB_ACCESS_REMOTE_WRITE | IB_ACCESS_LOCAL_WRITE :
 				IB_ACCESS_REMOTE_READ);
-	frmr_wr.wr.fast_reg.rkey = seg1->mr_chunk.rl_mw->r.frmr.fr_mr->rkey;
+	fastreg_wr.wr.fast_reg.rkey = mr->rkey;
 	DECR_CQCOUNT(&r_xprt->rx_ep);
 
-	rc = ib_post_send(ia->ri_id->qp, post_wr, &bad_wr);
-
+	rc = ib_post_send(ia->ri_id->qp, &fastreg_wr, &bad_wr);
 	if (rc) {
 		dprintk("RPC:       %s: failed ib_post_send for register,"
 			" status %i\n", __func__, rc);
-		while (i--)
-			rpcrdma_unmap_one(ia, --seg);
+		ib_update_fast_reg_key(mr, --key);
+		goto out_err;
 	} else {
-		seg1->mr_rkey = seg1->mr_chunk.rl_mw->r.frmr.fr_mr->rkey;
+		seg1->mr_rkey = mr->rkey;
 		seg1->mr_base = seg1->mr_dma + pageoff;
 		seg1->mr_nsegs = i;
 		seg1->mr_len = len;
 	}
 	*nsegs = i;
+	return 0;
+out_err:
+	frmr->fr_state = FRMR_IS_INVALID;
+	while (i--)
+		rpcrdma_unmap_one(ia, --seg);
 	return rc;
 }
 
@@ -1585,20 +1804,25 @@ rpcrdma_deregister_frmr_external(struct rpcrdma_mr_seg *seg,
 	struct ib_send_wr invalidate_wr, *bad_wr;
 	int rc;
 
-	while (seg1->mr_nsegs--)
-		rpcrdma_unmap_one(ia, seg++);
+	seg1->mr_chunk.rl_mw->r.frmr.fr_state = FRMR_IS_INVALID;
 
 	memset(&invalidate_wr, 0, sizeof invalidate_wr);
 	invalidate_wr.wr_id = (unsigned long)(void *)seg1->mr_chunk.rl_mw;
 	invalidate_wr.opcode = IB_WR_LOCAL_INV;
-	invalidate_wr.send_flags = IB_SEND_SIGNALED;
 	invalidate_wr.ex.invalidate_rkey = seg1->mr_chunk.rl_mw->r.frmr.fr_mr->rkey;
 	DECR_CQCOUNT(&r_xprt->rx_ep);
 
+	read_lock(&ia->ri_qplock);
+	while (seg1->mr_nsegs--)
+		rpcrdma_unmap_one(ia, seg++);
 	rc = ib_post_send(ia->ri_id->qp, &invalidate_wr, &bad_wr);
-	if (rc)
+	read_unlock(&ia->ri_qplock);
+	if (rc) {
+		/* Force rpcrdma_buffer_get() to retry */
+		seg1->mr_chunk.rl_mw->r.frmr.fr_state = FRMR_IS_STALE;
 		dprintk("RPC:       %s: failed ib_post_send for invalidate,"
 			" status %i\n", __func__, rc);
+	}
 	return rc;
 }
 
@@ -1656,8 +1880,10 @@ rpcrdma_deregister_fmr_external(struct rpcrdma_mr_seg *seg,
 
 	list_add(&seg1->mr_chunk.rl_mw->r.fmr->list, &l);
 	rc = ib_unmap_fmr(&l);
+	read_lock(&ia->ri_qplock);
 	while (seg1->mr_nsegs--)
 		rpcrdma_unmap_one(ia, seg++);
+	read_unlock(&ia->ri_qplock);
 	if (rc)
 		dprintk("RPC:       %s: failed ib_unmap_fmr,"
 			" status %i\n", __func__, rc);
@@ -1673,7 +1899,6 @@ rpcrdma_register_external(struct rpcrdma_mr_seg *seg,
 
 	switch (ia->ri_memreg_strategy) {
 
-#if RPCRDMA_PERSISTENT_REGISTRATION
 	case RPCRDMA_ALLPHYSICAL:
 		rpcrdma_map_one(ia, seg, writing);
 		seg->mr_rkey = ia->ri_bind_mem->rkey;
@@ -1681,7 +1906,6 @@ rpcrdma_register_external(struct rpcrdma_mr_seg *seg,
 		seg->mr_nsegs = 1;
 		nsegs = 1;
 		break;
-#endif
 
 	/* Registration using frmr registration */
 	case RPCRDMA_FRMR:
@@ -1711,11 +1935,11 @@ rpcrdma_deregister_external(struct rpcrdma_mr_seg *seg,
 
 	switch (ia->ri_memreg_strategy) {
 
-#if RPCRDMA_PERSISTENT_REGISTRATION
 	case RPCRDMA_ALLPHYSICAL:
+		read_lock(&ia->ri_qplock);
 		rpcrdma_unmap_one(ia, seg);
+		read_unlock(&ia->ri_qplock);
 		break;
-#endif
 
 	case RPCRDMA_FRMR:
 		rc = rpcrdma_deregister_frmr_external(seg, ia, r_xprt);
@@ -1809,3 +2033,44 @@ rpcrdma_ep_post_recv(struct rpcrdma_ia *ia,
 			rc);
 	return rc;
 }
+
+/* Physical mapping means one Read/Write list entry per-page.
+ * All list entries must fit within an inline buffer
+ *
+ * NB: The server must return a Write list for NFS READ,
+ *     which has the same constraint. Factor in the inline
+ *     rsize as well.
+ */
+static size_t
+rpcrdma_physical_max_payload(struct rpcrdma_xprt *r_xprt)
+{
+	struct rpcrdma_create_data_internal *cdata = &r_xprt->rx_data;
+	unsigned int inline_size, pages;
+
+	inline_size = min_t(unsigned int,
+			    cdata->inline_wsize, cdata->inline_rsize);
+	inline_size -= RPCRDMA_HDRLEN_MIN;
+	pages = inline_size / sizeof(struct rpcrdma_segment);
+	return pages << PAGE_SHIFT;
+}
+
+static size_t
+rpcrdma_mr_max_payload(struct rpcrdma_xprt *r_xprt)
+{
+	return RPCRDMA_MAX_DATA_SEGS << PAGE_SHIFT;
+}
+
+size_t
+rpcrdma_max_payload(struct rpcrdma_xprt *r_xprt)
+{
+	size_t result;
+
+	switch (r_xprt->rx_ia.ri_memreg_strategy) {
+	case RPCRDMA_ALLPHYSICAL:
+		result = rpcrdma_physical_max_payload(r_xprt);
+		break;
+	default:
+		result = rpcrdma_mr_max_payload(r_xprt);
+	}
+	return result;
+}

net/sunrpc/xprtrdma/xprt_rdma.h

@@ -59,6 +59,7 @@
  * Interface Adapter -- one per transport instance
  */
 struct rpcrdma_ia {
+	rwlock_t		ri_qplock;
 	struct rdma_cm_id 	*ri_id;
 	struct ib_pd		*ri_pd;
 	struct ib_mr		*ri_bind_mem;
@@ -98,6 +99,14 @@ struct rpcrdma_ep {
 #define INIT_CQCOUNT(ep) atomic_set(&(ep)->rep_cqcount, (ep)->rep_cqinit)
 #define DECR_CQCOUNT(ep) atomic_sub_return(1, &(ep)->rep_cqcount)
 
+enum rpcrdma_chunktype {
+	rpcrdma_noch = 0,
+	rpcrdma_readch,
+	rpcrdma_areadch,
+	rpcrdma_writech,
+	rpcrdma_replych
+};
+
 /*
  * struct rpcrdma_rep -- this structure encapsulates state required to recv
  * and complete a reply, asychronously. It needs several pieces of
@@ -136,6 +145,40 @@ struct rpcrdma_rep {
 	char	rr_base[MAX_RPCRDMAHDR]; /* minimal inline receive buffer */
 };
 
+/*
+ * struct rpcrdma_mw - external memory region metadata
+ *
+ * An external memory region is any buffer or page that is registered
+ * on the fly (ie, not pre-registered).
+ *
+ * Each rpcrdma_buffer has a list of free MWs anchored in rb_mws. During
+ * call_allocate, rpcrdma_buffer_get() assigns one to each segment in
+ * an rpcrdma_req. Then rpcrdma_register_external() grabs these to keep
+ * track of registration metadata while each RPC is pending.
+ * rpcrdma_deregister_external() uses this metadata to unmap and
+ * release these resources when an RPC is complete.
+ */
+enum rpcrdma_frmr_state {
+	FRMR_IS_INVALID,	/* ready to be used */
+	FRMR_IS_VALID,		/* in use */
+	FRMR_IS_STALE,		/* failed completion */
+};
+
+struct rpcrdma_frmr {
+	struct ib_fast_reg_page_list	*fr_pgl;
+	struct ib_mr			*fr_mr;
+	enum rpcrdma_frmr_state		fr_state;
+};
+
+struct rpcrdma_mw {
+	union {
+		struct ib_fmr		*fmr;
+		struct rpcrdma_frmr	frmr;
+	} r;
+	struct list_head	mw_list;
+	struct list_head	mw_all;
+};
+
 /*
  * struct rpcrdma_req -- structure central to the request/reply sequence.
  *
@@ -163,17 +206,7 @@ struct rpcrdma_rep {
 struct rpcrdma_mr_seg {		/* chunk descriptors */
 	union {				/* chunk memory handles */
 		struct ib_mr	*rl_mr;		/* if registered directly */
-		struct rpcrdma_mw {		/* if registered from region */
-			union {
-				struct ib_fmr	*fmr;
-				struct {
-					struct ib_fast_reg_page_list *fr_pgl;
-					struct ib_mr *fr_mr;
-					enum { FRMR_IS_INVALID, FRMR_IS_VALID  } state;
-				} frmr;
-			} r;
-			struct list_head mw_list;
-		} *rl_mw;
+		struct rpcrdma_mw *rl_mw;	/* if registered from region */
 	} mr_chunk;
 	u64		mr_base;	/* registration result */
 	u32		mr_rkey;	/* registration result */
@@ -191,6 +224,7 @@ struct rpcrdma_req {
 	unsigned int	rl_niovs;	/* 0, 2 or 4 */
 	unsigned int	rl_nchunks;	/* non-zero if chunks */
 	unsigned int	rl_connect_cookie;	/* retry detection */
+	enum rpcrdma_chunktype	rl_rtype, rl_wtype;
 	struct rpcrdma_buffer *rl_buffer; /* home base for this structure */
 	struct rpcrdma_rep	*rl_reply;/* holder for reply buffer */
 	struct rpcrdma_mr_seg rl_segments[RPCRDMA_MAX_SEGS];/* chunk segments */
@@ -214,6 +248,7 @@ struct rpcrdma_buffer {
 	atomic_t	rb_credits;	/* most recent server credits */
 	int		rb_max_requests;/* client max requests */
 	struct list_head rb_mws;	/* optional memory windows/fmrs/frmrs */
+	struct list_head rb_all;
 	int		rb_send_index;
 	struct rpcrdma_req	**rb_send_bufs;
 	int		rb_recv_index;
@@ -306,7 +341,7 @@ int rpcrdma_ep_create(struct rpcrdma_ep *, struct rpcrdma_ia *,
 				struct rpcrdma_create_data_internal *);
 void rpcrdma_ep_destroy(struct rpcrdma_ep *, struct rpcrdma_ia *);
 int rpcrdma_ep_connect(struct rpcrdma_ep *, struct rpcrdma_ia *);
-int rpcrdma_ep_disconnect(struct rpcrdma_ep *, struct rpcrdma_ia *);
+void rpcrdma_ep_disconnect(struct rpcrdma_ep *, struct rpcrdma_ia *);
 
 int rpcrdma_ep_post(struct rpcrdma_ia *, struct rpcrdma_ep *,
 				struct rpcrdma_req *);
@@ -346,7 +381,9 @@ void rpcrdma_reply_handler(struct rpcrdma_rep *);
 /*
  * RPC/RDMA protocol calls - xprtrdma/rpc_rdma.c
  */
+ssize_t rpcrdma_marshal_chunks(struct rpc_rqst *, ssize_t);
 int rpcrdma_marshal_req(struct rpc_rqst *);
+size_t rpcrdma_max_payload(struct rpcrdma_xprt *);
 
 /* Temporary NFS request map cache. Created in svc_rdma.c  */
 extern struct kmem_cache *svc_rdma_map_cachep;

net/sunrpc/xprtsock.c

@@ -594,6 +594,7 @@ static int xs_local_send_request(struct rpc_task *task)
 	}
 
 	switch (status) {
+	case -ENOBUFS:
 	case -EAGAIN:
 		status = xs_nospace(task);
 		break;
@@ -661,6 +662,7 @@ static int xs_udp_send_request(struct rpc_task *task)
 		dprintk("RPC:       sendmsg returned unrecognized error %d\n",
 			-status);
 	case -ENETUNREACH:
+	case -ENOBUFS:
 	case -EPIPE:
 	case -ECONNREFUSED:
 		/* When the server has died, an ICMP port unreachable message
@@ -758,6 +760,7 @@ static int xs_tcp_send_request(struct rpc_task *task)
 		status = -ENOTCONN;
 		/* Should we call xs_close() here? */
 		break;
+	case -ENOBUFS:
 	case -EAGAIN:
 		status = xs_nospace(task);
 		break;
@@ -1946,6 +1949,7 @@ static int xs_local_setup_socket(struct sock_xprt *transport)
 		dprintk("RPC:       xprt %p connected to %s\n",
 				xprt, xprt->address_strings[RPC_DISPLAY_ADDR]);
 		xprt_set_connected(xprt);
+	case -ENOBUFS:
 		break;
 	case -ENOENT:
 		dprintk("RPC:       xprt %p: socket %s does not exist\n",
@@ -2281,6 +2285,7 @@ static void xs_tcp_setup_socket(struct work_struct *work)
 	case -ECONNREFUSED:
 	case -ECONNRESET:
 	case -ENETUNREACH:
+	case -ENOBUFS:
 		/* retry with existing socket, after a delay */
 		goto out;
 	}
@@ -3054,12 +3059,12 @@ static int param_set_uint_minmax(const char *val,
 		const struct kernel_param *kp,
 		unsigned int min, unsigned int max)
 {
-	unsigned long num;
+	unsigned int num;
 	int ret;
 
 	if (!val)
 		return -EINVAL;
-	ret = strict_strtoul(val, 0, &num);
+	ret = kstrtouint(val, 0, &num);
 	if (ret == -EINVAL || num < min || num > max)
 		return -EINVAL;
 	*((unsigned int *)kp->arg) = num;