Merge branch 'writeback'

17 files changed, 437 insertions, 280 deletions

diff --git a/fs/nfs/Makefile b/fs/nfs/Makefile
index 8664417955a2..6abdda209642 100644
--- a/fs/nfs/Makefile
+++ b/fs/nfs/Makefile
@@ -6,7 +6,7 @@ obj-$(CONFIG_NFS_FS) += nfs.o
 
 CFLAGS_nfstrace.o += -I$(src)
 nfs-y 			:= client.o dir.o file.o getroot.o inode.o super.o \
-			   direct.o pagelist.o read.o symlink.o unlink.o \
+			   io.o direct.o pagelist.o read.o symlink.o unlink.o \
 			   write.o namespace.o mount_clnt.o nfstrace.o
 nfs-$(CONFIG_ROOT_NFS)	+= nfsroot.o
 nfs-$(CONFIG_SYSCTL)	+= sysctl.o
diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c
index d8015a03db4c..e2d606abc9e8 100644
--- a/fs/nfs/dir.c
+++ b/fs/nfs/dir.c
@@ -2231,21 +2231,37 @@ static struct nfs_access_entry *nfs_access_search_rbtree(struct inode *inode, st
 	return NULL;
 }
 
-static int nfs_access_get_cached(struct inode *inode, struct rpc_cred *cred, struct nfs_access_entry *res)
+static int nfs_access_get_cached(struct inode *inode, struct rpc_cred *cred, struct nfs_access_entry *res, bool may_block)
 {
 	struct nfs_inode *nfsi = NFS_I(inode);
 	struct nfs_access_entry *cache;
-	int err = -ENOENT;
+	bool retry = true;
+	int err;
 
 	spin_lock(&inode->i_lock);
-	if (nfsi->cache_validity & NFS_INO_INVALID_ACCESS)
-		goto out_zap;
-	cache = nfs_access_search_rbtree(inode, cred);
-	if (cache == NULL)
-		goto out;
-	if (!nfs_have_delegated_attributes(inode) &&
-	    !time_in_range_open(jiffies, cache->jiffies, cache->jiffies + nfsi->attrtimeo))
-		goto out_stale;
+	for(;;) {
+		if (nfsi->cache_validity & NFS_INO_INVALID_ACCESS)
+			goto out_zap;
+		cache = nfs_access_search_rbtree(inode, cred);
+		err = -ENOENT;
+		if (cache == NULL)
+			goto out;
+		/* Found an entry, is our attribute cache valid? */
+		if (!nfs_attribute_cache_expired(inode) &&
+		    !(nfsi->cache_validity & NFS_INO_INVALID_ATTR))
+			break;
+		err = -ECHILD;
+		if (!may_block)
+			goto out;
+		if (!retry)
+			goto out_zap;
+		spin_unlock(&inode->i_lock);
+		err = __nfs_revalidate_inode(NFS_SERVER(inode), inode);
+		if (err)
+			return err;
+		spin_lock(&inode->i_lock);
+		retry = false;
+	}
 	res->jiffies = cache->jiffies;
 	res->cred = cache->cred;
 	res->mask = cache->mask;
@@ -2254,12 +2270,6 @@ static int nfs_access_get_cached(struct inode *inode, struct rpc_cred *cred, str
 out:
 	spin_unlock(&inode->i_lock);
 	return err;
-out_stale:
-	rb_erase(&cache->rb_node, &nfsi->access_cache);
-	list_del(&cache->lru);
-	spin_unlock(&inode->i_lock);
-	nfs_access_free_entry(cache);
-	return -ENOENT;
 out_zap:
 	spin_unlock(&inode->i_lock);
 	nfs_access_zap_cache(inode);
@@ -2286,13 +2296,12 @@ static int nfs_access_get_cached_rcu(struct inode *inode, struct rpc_cred *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))
+	err = nfs_revalidate_inode_rcu(NFS_SERVER(inode), inode);
+	if (err)
 		goto out;
 	res->jiffies = cache->jiffies;
 	res->cred = cache->cred;
 	res->mask = cache->mask;
-	err = 0;
 out:
 	rcu_read_unlock();
 	return err;
@@ -2381,18 +2390,19 @@ EXPORT_SYMBOL_GPL(nfs_access_set_mask);
 static int nfs_do_access(struct inode *inode, struct rpc_cred *cred, int mask)
 {
 	struct nfs_access_entry cache;
+	bool may_block = (mask & MAY_NOT_BLOCK) == 0;
 	int status;
 
 	trace_nfs_access_enter(inode);
 
 	status = nfs_access_get_cached_rcu(inode, cred, &cache);
 	if (status != 0)
-		status = nfs_access_get_cached(inode, cred, &cache);
+		status = nfs_access_get_cached(inode, cred, &cache, may_block);
 	if (status == 0)
 		goto out_cached;
 
 	status = -ECHILD;
-	if (mask & MAY_NOT_BLOCK)
+	if (!may_block)
 		goto out;
 
 	/* Be clever: ask server to check for all possible rights */
diff --git a/fs/nfs/direct.c b/fs/nfs/direct.c
index c7326c2af2c3..f0f4b8d3d83f 100644
--- a/fs/nfs/direct.c
+++ b/fs/nfs/direct.c
@@ -196,6 +196,12 @@ static void nfs_direct_set_hdr_verf(struct nfs_direct_req *dreq,
 	WARN_ON_ONCE(verfp->committed < 0);
 }
 
+static int nfs_direct_cmp_verf(const struct nfs_writeverf *v1,
+		const struct nfs_writeverf *v2)
+{
+	return nfs_write_verifier_cmp(&v1->verifier, &v2->verifier);
+}
+
 /*
  * nfs_direct_cmp_hdr_verf - compare verifier for pgio header
  * @dreq - direct request possibly spanning multiple servers
@@ -215,7 +221,7 @@ static int nfs_direct_set_or_cmp_hdr_verf(struct nfs_direct_req *dreq,
 		nfs_direct_set_hdr_verf(dreq, hdr);
 		return 0;
 	}
-	return memcmp(verfp, &hdr->verf, sizeof(struct nfs_writeverf));
+	return nfs_direct_cmp_verf(verfp, &hdr->verf);
 }
 
 /*
@@ -238,7 +244,7 @@ static int nfs_direct_cmp_commit_data_verf(struct nfs_direct_req *dreq,
 	if (verfp->committed < 0)
 		return 1;
 
-	return memcmp(verfp, &data->verf, sizeof(struct nfs_writeverf));
+	return nfs_direct_cmp_verf(verfp, &data->verf);
 }
 
 /**
@@ -368,22 +374,10 @@ out:
  * Synchronous I/O uses a stack-allocated iocb.  Thus we can't trust
  * the iocb is still valid here if this is a synchronous request.
  */
-static void nfs_direct_complete(struct nfs_direct_req *dreq, bool write)
+static void nfs_direct_complete(struct nfs_direct_req *dreq)
 {
 	struct inode *inode = dreq->inode;
 
-	if (dreq->iocb && write) {
-		loff_t pos = dreq->iocb->ki_pos + dreq->count;
-
-		spin_lock(&inode->i_lock);
-		if (i_size_read(inode) < pos)
-			i_size_write(inode, pos);
-		spin_unlock(&inode->i_lock);
-	}
-
-	if (write)
-		nfs_zap_mapping(inode, inode->i_mapping);
-
 	inode_dio_end(inode);
 
 	if (dreq->iocb) {
@@ -438,7 +432,7 @@ static void nfs_direct_read_completion(struct nfs_pgio_header *hdr)
 	}
 out_put:
 	if (put_dreq(dreq))
-		nfs_direct_complete(dreq, false);
+		nfs_direct_complete(dreq);
 	hdr->release(hdr);
 }
 
@@ -544,7 +538,7 @@ static ssize_t nfs_direct_read_schedule_iovec(struct nfs_direct_req *dreq,
 	}
 
 	if (put_dreq(dreq))
-		nfs_direct_complete(dreq, false);
+		nfs_direct_complete(dreq);
 	return 0;
 }
 
@@ -585,17 +579,12 @@ ssize_t nfs_file_direct_read(struct kiocb *iocb, struct iov_iter *iter)
 	if (!count)
 		goto out;
 
-	inode_lock(inode);
-	result = nfs_sync_mapping(mapping);
-	if (result)
-		goto out_unlock;
-
 	task_io_account_read(count);
 
 	result = -ENOMEM;
 	dreq = nfs_direct_req_alloc();
 	if (dreq == NULL)
-		goto out_unlock;
+		goto out;
 
 	dreq->inode = inode;
 	dreq->bytes_left = dreq->max_count = count;
@@ -610,10 +599,12 @@ ssize_t nfs_file_direct_read(struct kiocb *iocb, struct iov_iter *iter)
 	if (!is_sync_kiocb(iocb))
 		dreq->iocb = iocb;
 
+	nfs_start_io_direct(inode);
+
 	NFS_I(inode)->read_io += count;
 	result = nfs_direct_read_schedule_iovec(dreq, iter, iocb->ki_pos);
 
-	inode_unlock(inode);
+	nfs_end_io_direct(inode);
 
 	if (!result) {
 		result = nfs_direct_wait(dreq);
@@ -621,13 +612,8 @@ ssize_t nfs_file_direct_read(struct kiocb *iocb, struct iov_iter *iter)
 			iocb->ki_pos += result;
 	}
 
-	nfs_direct_req_release(dreq);
-	return result;
-
 out_release:
 	nfs_direct_req_release(dreq);
-out_unlock:
-	inode_unlock(inode);
 out:
 	return result;
 }
@@ -659,6 +645,8 @@ static void nfs_direct_write_reschedule(struct nfs_direct_req *dreq)
 	nfs_direct_write_scan_commit_list(dreq->inode, &reqs, &cinfo);
 
 	dreq->count = 0;
+	dreq->verf.committed = NFS_INVALID_STABLE_HOW;
+	nfs_clear_pnfs_ds_commit_verifiers(&dreq->ds_cinfo);
 	for (i = 0; i < dreq->mirror_count; i++)
 		dreq->mirrors[i].count = 0;
 	get_dreq(dreq);
@@ -777,7 +765,8 @@ static void nfs_direct_write_schedule_work(struct work_struct *work)
 			nfs_direct_write_reschedule(dreq);
 			break;
 		default:
-			nfs_direct_complete(dreq, true);
+			nfs_zap_mapping(dreq->inode, dreq->inode->i_mapping);
+			nfs_direct_complete(dreq);
 	}
 }
 
@@ -993,6 +982,7 @@ static ssize_t nfs_direct_write_schedule_iovec(struct nfs_direct_req *dreq,
 ssize_t nfs_file_direct_write(struct kiocb *iocb, struct iov_iter *iter)
 {
 	ssize_t result = -EINVAL;
+	size_t count;
 	struct file *file = iocb->ki_filp;
 	struct address_space *mapping = file->f_mapping;
 	struct inode *inode = mapping->host;
@@ -1003,34 +993,24 @@ ssize_t nfs_file_direct_write(struct kiocb *iocb, struct iov_iter *iter)
 	dfprintk(FILE, "NFS: direct write(%pD2, %zd@%Ld)\n",
 		file, iov_iter_count(iter), (long long) iocb->ki_pos);
 
-	nfs_add_stats(mapping->host, NFSIOS_DIRECTWRITTENBYTES,
-		      iov_iter_count(iter));
+	result = generic_write_checks(iocb, iter);
+	if (result <= 0)
+		return result;
+	count = result;
+	nfs_add_stats(mapping->host, NFSIOS_DIRECTWRITTENBYTES, count);
 
 	pos = iocb->ki_pos;
 	end = (pos + iov_iter_count(iter) - 1) >> PAGE_SHIFT;
 
-	inode_lock(inode);
-
-	result = nfs_sync_mapping(mapping);
-	if (result)
-		goto out_unlock;
-
-	if (mapping->nrpages) {
-		result = invalidate_inode_pages2_range(mapping,
-					pos >> PAGE_SHIFT, end);
-		if (result)
-			goto out_unlock;
-	}
-
-	task_io_account_write(iov_iter_count(iter));
+	task_io_account_write(count);
 
 	result = -ENOMEM;
 	dreq = nfs_direct_req_alloc();
 	if (!dreq)
-		goto out_unlock;
+		goto out;
 
 	dreq->inode = inode;
-	dreq->bytes_left = dreq->max_count = iov_iter_count(iter);
+	dreq->bytes_left = dreq->max_count = count;
 	dreq->io_start = pos;
 	dreq->ctx = get_nfs_open_context(nfs_file_open_context(iocb->ki_filp));
 	l_ctx = nfs_get_lock_context(dreq->ctx);
@@ -1042,6 +1022,8 @@ ssize_t nfs_file_direct_write(struct kiocb *iocb, struct iov_iter *iter)
 	if (!is_sync_kiocb(iocb))
 		dreq->iocb = iocb;
 
+	nfs_start_io_direct(inode);
+
 	result = nfs_direct_write_schedule_iovec(dreq, iter, pos);
 
 	if (mapping->nrpages) {
@@ -1049,30 +1031,19 @@ ssize_t nfs_file_direct_write(struct kiocb *iocb, struct iov_iter *iter)
 					      pos >> PAGE_SHIFT, end);
 	}
 
-	inode_unlock(inode);
+	nfs_end_io_direct(inode);
 
 	if (!result) {
 		result = nfs_direct_wait(dreq);
 		if (result > 0) {
-			struct inode *inode = mapping->host;
-
 			iocb->ki_pos = pos + result;
-			spin_lock(&inode->i_lock);
-			if (i_size_read(inode) < iocb->ki_pos)
-				i_size_write(inode, iocb->ki_pos);
-			spin_unlock(&inode->i_lock);
-
 			/* XXX: should check the generic_write_sync retval */
 			generic_write_sync(iocb, result);
 		}
 	}
-	nfs_direct_req_release(dreq);
-	return result;
-
 out_release:
 	nfs_direct_req_release(dreq);
-out_unlock:
-	inode_unlock(inode);
+out:
 	return result;
 }
 
diff --git a/fs/nfs/file.c b/fs/nfs/file.c
index 6bcd8913e8a9..7d620970f2e1 100644
--- a/fs/nfs/file.c
+++ b/fs/nfs/file.c
@@ -170,12 +170,14 @@ nfs_file_read(struct kiocb *iocb, struct iov_iter *to)
 		iocb->ki_filp,
 		iov_iter_count(to), (unsigned long) iocb->ki_pos);
 
-	result = nfs_revalidate_mapping_protected(inode, iocb->ki_filp->f_mapping);
+	nfs_start_io_read(inode);
+	result = nfs_revalidate_mapping(inode, iocb->ki_filp->f_mapping);
 	if (!result) {
 		result = generic_file_read_iter(iocb, to);
 		if (result > 0)
 			nfs_add_stats(inode, NFSIOS_NORMALREADBYTES, result);
 	}
+	nfs_end_io_read(inode);
 	return result;
 }
 EXPORT_SYMBOL_GPL(nfs_file_read);
@@ -191,12 +193,14 @@ nfs_file_splice_read(struct file *filp, loff_t *ppos,
 	dprintk("NFS: splice_read(%pD2, %lu@%Lu)\n",
 		filp, (unsigned long) count, (unsigned long long) *ppos);
 
-	res = nfs_revalidate_mapping_protected(inode, filp->f_mapping);
+	nfs_start_io_read(inode);
+	res = nfs_revalidate_mapping(inode, filp->f_mapping);
 	if (!res) {
 		res = generic_file_splice_read(filp, ppos, pipe, count, flags);
 		if (res > 0)
 			nfs_add_stats(inode, NFSIOS_NORMALREADBYTES, res);
 	}
+	nfs_end_io_read(inode);
 	return res;
 }
 EXPORT_SYMBOL_GPL(nfs_file_splice_read);
@@ -272,16 +276,13 @@ nfs_file_fsync(struct file *file, loff_t start, loff_t end, int datasync)
 
 	trace_nfs_fsync_enter(inode);
 
-	inode_dio_wait(inode);
 	do {
 		ret = filemap_write_and_wait_range(inode->i_mapping, start, end);
 		if (ret != 0)
 			break;
-		inode_lock(inode);
 		ret = nfs_file_fsync_commit(file, start, end, datasync);
 		if (!ret)
 			ret = pnfs_sync_inode(inode, !!datasync);
-		inode_unlock(inode);
 		/*
 		 * If nfs_file_fsync_commit detected a server reboot, then
 		 * resend all dirty pages that might have been covered by
@@ -359,19 +360,6 @@ static int nfs_write_begin(struct file *file, struct address_space *mapping,
 		file, mapping->host->i_ino, len, (long long) pos);
 
 start:
-	/*
-	 * Prevent starvation issues if someone is doing a consistency
-	 * sync-to-disk
-	 */
-	ret = wait_on_bit_action(&NFS_I(mapping->host)->flags, NFS_INO_FLUSHING,
-				 nfs_wait_bit_killable, TASK_KILLABLE);
-	if (ret)
-		return ret;
-	/*
-	 * Wait for O_DIRECT to complete
-	 */
-	inode_dio_wait(mapping->host);
-
 	page = grab_cache_page_write_begin(mapping, index, flags);
 	if (!page)
 		return -ENOMEM;
@@ -470,31 +458,8 @@ static void nfs_invalidate_page(struct page *page, unsigned int offset,
  */
 static int nfs_release_page(struct page *page, gfp_t gfp)
 {
-	struct address_space *mapping = page->mapping;
-
 	dfprintk(PAGECACHE, "NFS: release_page(%p)\n", page);
 
-	/* Always try to initiate a 'commit' if relevant, but only
-	 * wait for it if the caller allows blocking.  Even then,
-	 * only wait 1 second and only if the 'bdi' is not congested.
-	 * Waiting indefinitely can cause deadlocks when the NFS
-	 * server is on this machine, when a new TCP connection is
-	 * needed and in other rare cases.  There is no particular
-	 * need to wait extensively here.  A short wait has the
-	 * benefit that someone else can worry about the freezer.
-	 */
-	if (mapping) {
-		struct nfs_server *nfss = NFS_SERVER(mapping->host);
-		nfs_commit_inode(mapping->host, 0);
-		if (gfpflags_allow_blocking(gfp) &&
-		    !bdi_write_congested(&nfss->backing_dev_info)) {
-			wait_on_page_bit_killable_timeout(page, PG_private,
-							  HZ);
-			if (PagePrivate(page))
-				set_bdi_congested(&nfss->backing_dev_info,
-						  BLK_RW_ASYNC);
-		}
-	}
 	/* If PagePrivate() is set, then the page is not freeable */
 	if (PagePrivate(page))
 		return 0;
@@ -604,6 +569,8 @@ static int nfs_vm_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
 		filp, filp->f_mapping->host->i_ino,
 		(long long)page_offset(page));
 
+	sb_start_pagefault(inode->i_sb);
+
 	/* make sure the cache has finished storing the page */
 	nfs_fscache_wait_on_page_write(NFS_I(inode), page);
 
@@ -630,6 +597,7 @@ static int nfs_vm_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
 out_unlock:
 	unlock_page(page);
 out:
+	sb_end_pagefault(inode->i_sb);
 	return ret;
 }
 
@@ -656,23 +624,17 @@ ssize_t nfs_file_write(struct kiocb *iocb, struct iov_iter *from)
 	struct inode *inode = file_inode(file);
 	unsigned long written = 0;
 	ssize_t result;
-	size_t count = iov_iter_count(from);
 
 	result = nfs_key_timeout_notify(file, inode);
 	if (result)
 		return result;
 
-	if (iocb->ki_flags & IOCB_DIRECT) {
-		result = generic_write_checks(iocb, from);
-		if (result <= 0)
-			return result;
+	if (iocb->ki_flags & IOCB_DIRECT)
 		return nfs_file_direct_write(iocb, from);
-	}
 
 	dprintk("NFS: write(%pD2, %zu@%Ld)\n",
-		file, count, (long long) iocb->ki_pos);
+		file, iov_iter_count(from), (long long) iocb->ki_pos);
 
-	result = -EBUSY;
 	if (IS_SWAPFILE(inode))
 		goto out_swapfile;
 	/*
@@ -684,28 +646,33 @@ ssize_t nfs_file_write(struct kiocb *iocb, struct iov_iter *from)
 			goto out;
 	}
 
-	result = count;
-	if (!count)
+	nfs_start_io_write(inode);
+	result = generic_write_checks(iocb, from);
+	if (result > 0) {
+		current->backing_dev_info = inode_to_bdi(inode);
+		result = generic_perform_write(file, from, iocb->ki_pos);
+		current->backing_dev_info = NULL;
+	}
+	nfs_end_io_write(inode);
+	if (result <= 0)
 		goto out;
 
-	result = generic_file_write_iter(iocb, from);
-	if (result > 0)
-		written = result;
+	written = generic_write_sync(iocb, result);
+	iocb->ki_pos += written;
 
 	/* Return error values */
-	if (result >= 0 && nfs_need_check_write(file, inode)) {
+	if (nfs_need_check_write(file, inode)) {
 		int err = vfs_fsync(file, 0);
 		if (err < 0)
 			result = err;
 	}
-	if (result > 0)
-		nfs_add_stats(inode, NFSIOS_NORMALWRITTENBYTES, written);
+	nfs_add_stats(inode, NFSIOS_NORMALWRITTENBYTES, written);
 out:
 	return result;
 
 out_swapfile:
 	printk(KERN_INFO "NFS: attempt to write to active swap file!\n");
-	goto out;
+	return -EBUSY;
 }
 EXPORT_SYMBOL_GPL(nfs_file_write);
 
@@ -780,11 +747,6 @@ do_unlk(struct file *filp, int cmd, struct file_lock *fl, int is_local)
 }
 
 static int
-is_time_granular(struct timespec *ts) {
-	return ((ts->tv_sec == 0) && (ts->tv_nsec <= 1000));
-}
-
-static int
 do_setlk(struct file *filp, int cmd, struct file_lock *fl, int is_local)
 {
 	struct inode *inode = filp->f_mapping->host;
@@ -817,12 +779,8 @@ do_setlk(struct file *filp, int cmd, struct file_lock *fl, int is_local)
 	 * This makes locking act as a cache coherency point.
 	 */
 	nfs_sync_mapping(filp->f_mapping);
-	if (!NFS_PROTO(inode)->have_delegation(inode, FMODE_READ)) {
-		if (is_time_granular(&NFS_SERVER(inode)->time_delta))
-			__nfs_revalidate_inode(NFS_SERVER(inode), inode);
-		else
-			nfs_zap_caches(inode);
-	}
+	if (!NFS_PROTO(inode)->have_delegation(inode, FMODE_READ))
+		nfs_zap_mapping(inode, filp->f_mapping);
 out:
 	return status;
 }
diff --git a/fs/nfs/filelayout/filelayout.c b/fs/nfs/filelayout/filelayout.c
index aa59757389dc..a3fc48ba4931 100644
--- a/fs/nfs/filelayout/filelayout.c
+++ b/fs/nfs/filelayout/filelayout.c
@@ -255,13 +255,16 @@ static int filelayout_read_done_cb(struct rpc_task *task,
 static void
 filelayout_set_layoutcommit(struct nfs_pgio_header *hdr)
 {
+	loff_t end_offs = 0;
 
 	if (FILELAYOUT_LSEG(hdr->lseg)->commit_through_mds ||
-	    hdr->res.verf->committed != NFS_DATA_SYNC)
+	    hdr->res.verf->committed == NFS_FILE_SYNC)
 		return;
+	if (hdr->res.verf->committed == NFS_DATA_SYNC)
+		end_offs = hdr->mds_offset + (loff_t)hdr->res.count;
 
-	pnfs_set_layoutcommit(hdr->inode, hdr->lseg,
-			hdr->mds_offset + hdr->res.count);
+	/* Note: if the write is unstable, don't set end_offs until commit */
+	pnfs_set_layoutcommit(hdr->inode, hdr->lseg, end_offs);
 	dprintk("%s inode %lu pls_end_pos %lu\n", __func__, hdr->inode->i_ino,
 		(unsigned long) NFS_I(hdr->inode)->layout->plh_lwb);
 }
@@ -354,6 +357,12 @@ static int filelayout_write_done_cb(struct rpc_task *task,
 	}
 
 	filelayout_set_layoutcommit(hdr);
+
+	/* zero out the fattr */
+	hdr->fattr.valid = 0;
+	if (task->tk_status >= 0)
+		nfs_writeback_update_inode(hdr);
+
 	return 0;
 }
 
@@ -375,8 +384,7 @@ static int filelayout_commit_done_cb(struct rpc_task *task,
 		return -EAGAIN;
 	}
 
-	if (data->verf.committed == NFS_UNSTABLE)
-		pnfs_set_layoutcommit(data->inode, data->lseg, data->lwb);
+	pnfs_set_layoutcommit(data->inode, data->lseg, data->lwb);
 
 	return 0;
 }
diff --git a/fs/nfs/flexfilelayout/flexfilelayout.c b/fs/nfs/flexfilelayout/flexfilelayout.c
index 0e8018bc9880..e6206eaf2bdf 100644
--- a/fs/nfs/flexfilelayout/flexfilelayout.c
+++ b/fs/nfs/flexfilelayout/flexfilelayout.c
@@ -1325,15 +1325,16 @@ ff_layout_need_layoutcommit(struct pnfs_layout_segment *lseg)
  * we always send layoutcommit after DS writes.
  */
 static void
-ff_layout_set_layoutcommit(struct nfs_pgio_header *hdr)
+ff_layout_set_layoutcommit(struct inode *inode,
+		struct pnfs_layout_segment *lseg,
+		loff_t end_offset)
 {
-	if (!ff_layout_need_layoutcommit(hdr->lseg))
+	if (!ff_layout_need_layoutcommit(lseg))
 		return;
 
-	pnfs_set_layoutcommit(hdr->inode, hdr->lseg,
-			hdr->mds_offset + hdr->res.count);
-	dprintk("%s inode %lu pls_end_pos %lu\n", __func__, hdr->inode->i_ino,
-		(unsigned long) NFS_I(hdr->inode)->layout->plh_lwb);
+	pnfs_set_layoutcommit(inode, lseg, end_offset);
+	dprintk("%s inode %lu pls_end_pos %llu\n", __func__, inode->i_ino,
+		(unsigned long long) NFS_I(inode)->layout->plh_lwb);
 }
 
 static bool
@@ -1469,6 +1470,7 @@ static void ff_layout_read_release(void *data)
 static int ff_layout_write_done_cb(struct rpc_task *task,
 				struct nfs_pgio_header *hdr)
 {
+	loff_t end_offs = 0;
 	int err;
 
 	trace_nfs4_pnfs_write(hdr, task->tk_status);
@@ -1494,7 +1496,10 @@ static int ff_layout_write_done_cb(struct rpc_task *task,
 
 	if (hdr->res.verf->committed == NFS_FILE_SYNC ||
 	    hdr->res.verf->committed == NFS_DATA_SYNC)
-		ff_layout_set_layoutcommit(hdr);
+		end_offs = hdr->mds_offset + (loff_t)hdr->res.count;
+
+	/* Note: if the write is unstable, don't set end_offs until commit */
+	ff_layout_set_layoutcommit(hdr->inode, hdr->lseg, end_offs);
 
 	/* zero out fattr since we don't care DS attr at all */
 	hdr->fattr.valid = 0;
@@ -1530,9 +1535,7 @@ static int ff_layout_commit_done_cb(struct rpc_task *task,
 		return -EAGAIN;
 	}
 
-	if (data->verf.committed == NFS_UNSTABLE
-	    && ff_layout_need_layoutcommit(data->lseg))
-		pnfs_set_layoutcommit(data->inode, data->lseg, data->lwb);
+	ff_layout_set_layoutcommit(data->inode, data->lseg, data->lwb);
 
 	return 0;
 }
diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index dda689d7a8a7..f108d58101f8 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -662,9 +662,7 @@ int nfs_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat)
 	trace_nfs_getattr_enter(inode);
 	/* Flush out writes to the server in order to update c/mtime.  */
 	if (S_ISREG(inode->i_mode)) {
-		inode_lock(inode);
-		err = nfs_sync_inode(inode);
-		inode_unlock(inode);
+		err = filemap_write_and_wait(inode->i_mapping);
 		if (err)
 			goto out;
 	}
@@ -879,7 +877,10 @@ void nfs_inode_attach_open_context(struct nfs_open_context *ctx)
 	struct nfs_inode *nfsi = NFS_I(inode);
 
 	spin_lock(&inode->i_lock);
-	list_add(&ctx->list, &nfsi->open_files);
+	if (ctx->mode & FMODE_WRITE)
+		list_add(&ctx->list, &nfsi->open_files);
+	else
+		list_add_tail(&ctx->list, &nfsi->open_files);
 	spin_unlock(&inode->i_lock);
 }
 EXPORT_SYMBOL_GPL(nfs_inode_attach_open_context);
@@ -972,6 +973,13 @@ __nfs_revalidate_inode(struct nfs_server *server, struct inode *inode)
 	if (NFS_STALE(inode))
 		goto out;
 
+	/* pNFS: Attributes aren't updated until we layoutcommit */
+	if (S_ISREG(inode->i_mode)) {
+		status = pnfs_sync_inode(inode, false);
+		if (status)
+			goto out;
+	}
+
 	status = -ENOMEM;
 	fattr = nfs_alloc_fattr();
 	if (fattr == NULL)
@@ -1122,14 +1130,12 @@ out:
 }
 
 /**
- * __nfs_revalidate_mapping - Revalidate the pagecache
+ * nfs_revalidate_mapping - Revalidate the pagecache
  * @inode - pointer to host inode
  * @mapping - pointer to mapping
- * @may_lock - take inode->i_mutex?
  */
-static int __nfs_revalidate_mapping(struct inode *inode,
-		struct address_space *mapping,
-		bool may_lock)
+int nfs_revalidate_mapping(struct inode *inode,
+		struct address_space *mapping)
 {
 	struct nfs_inode *nfsi = NFS_I(inode);
 	unsigned long *bitlock = &nfsi->flags;
@@ -1178,12 +1184,7 @@ static int __nfs_revalidate_mapping(struct inode *inode,
 	nfsi->cache_validity &= ~NFS_INO_INVALID_DATA;
 	spin_unlock(&inode->i_lock);
 	trace_nfs_invalidate_mapping_enter(inode);
-	if (may_lock) {
-		inode_lock(inode);
-		ret = nfs_invalidate_mapping(inode, mapping);
-		inode_unlock(inode);
-	} else
-		ret = nfs_invalidate_mapping(inode, mapping);
+	ret = nfs_invalidate_mapping(inode, mapping);
 	trace_nfs_invalidate_mapping_exit(inode, ret);
 
 	clear_bit_unlock(NFS_INO_INVALIDATING, bitlock);
@@ -1193,27 +1194,28 @@ out:
 	return ret;
 }
 
-/**
- * nfs_revalidate_mapping - Revalidate the pagecache
- * @inode - pointer to host inode
- * @mapping - pointer to mapping
- */
-int nfs_revalidate_mapping(struct inode *inode, struct address_space *mapping)
+static bool nfs_file_has_writers(struct nfs_inode *nfsi)
 {
-	return __nfs_revalidate_mapping(inode, mapping, false);
+	struct inode *inode = &nfsi->vfs_inode;
+
+	assert_spin_locked(&inode->i_lock);
+
+	if (!S_ISREG(inode->i_mode))
+		return false;
+	if (list_empty(&nfsi->open_files))
+		return false;
+	/* Note: This relies on nfsi->open_files being ordered with writers
+	 *       being placed at the head of the list.
+	 *       See nfs_inode_attach_open_context()
+	 */
+	return (list_first_entry(&nfsi->open_files,
+			struct nfs_open_context,
+			list)->mode & FMODE_WRITE) == FMODE_WRITE;
 }
 
-/**
- * nfs_revalidate_mapping_protected - Revalidate the pagecache
- * @inode - pointer to host inode
- * @mapping - pointer to mapping
- *
- * Differs from nfs_revalidate_mapping() in that it grabs the inode->i_mutex
- * while invalidating the mapping.
- */
-int nfs_revalidate_mapping_protected(struct inode *inode, struct address_space *mapping)
+static bool nfs_file_has_buffered_writers(struct nfs_inode *nfsi)
 {
-	return __nfs_revalidate_mapping(inode, mapping, true);
+	return nfs_file_has_writers(nfsi) && nfs_file_io_is_buffered(nfsi);
 }
 
 static unsigned long nfs_wcc_update_inode(struct inode *inode, struct nfs_fattr *fattr)
@@ -1280,22 +1282,24 @@ static int nfs_check_inode_attributes(struct inode *inode, struct nfs_fattr *fat
 	if ((fattr->valid & NFS_ATTR_FATTR_TYPE) && (inode->i_mode & S_IFMT) != (fattr->mode & S_IFMT))
 		return -EIO;
 
-	if ((fattr->valid & NFS_ATTR_FATTR_CHANGE) != 0 &&
-			inode->i_version != fattr->change_attr)
-		invalid |= NFS_INO_INVALID_ATTR|NFS_INO_REVAL_PAGECACHE;
+	if (!nfs_file_has_buffered_writers(nfsi)) {
+		/* Verify a few of the more important attributes */
+		if ((fattr->valid & NFS_ATTR_FATTR_CHANGE) != 0 && inode->i_version != fattr->change_attr)
+			invalid |= NFS_INO_INVALID_ATTR | NFS_INO_REVAL_PAGECACHE;
 
-	/* Verify a few of the more important attributes */
-	if ((fattr->valid & NFS_ATTR_FATTR_MTIME) && !timespec_equal(&inode->i_mtime, &fattr->mtime))
-		invalid |= NFS_INO_INVALID_ATTR;
+		if ((fattr->valid & NFS_ATTR_FATTR_MTIME) && !timespec_equal(&inode->i_mtime, &fattr->mtime))
+			invalid |= NFS_INO_INVALID_ATTR;
 
-	if (fattr->valid & NFS_ATTR_FATTR_SIZE) {
-		cur_size = i_size_read(inode);
-		new_isize = nfs_size_to_loff_t(fattr->size);
-		if (cur_size != new_isize)
-			invalid |= NFS_INO_INVALID_ATTR|NFS_INO_REVAL_PAGECACHE;
+		if ((fattr->valid & NFS_ATTR_FATTR_CTIME) && !timespec_equal(&inode->i_ctime, &fattr->ctime))
+			invalid |= NFS_INO_INVALID_ATTR;
+
+		if (fattr->valid & NFS_ATTR_FATTR_SIZE) {
+			cur_size = i_size_read(inode);
+			new_isize = nfs_size_to_loff_t(fattr->size);
+			if (cur_size != new_isize)
+				invalid |= NFS_INO_INVALID_ATTR|NFS_INO_REVAL_PAGECACHE;
+		}
 	}
-	if (nfsi->nrequests != 0)
-		invalid &= ~NFS_INO_REVAL_PAGECACHE;
 
 	/* Have any file permissions changed? */
 	if ((fattr->valid & NFS_ATTR_FATTR_MODE) && (inode->i_mode & S_IALLUGO) != (fattr->mode & S_IALLUGO))
@@ -1470,28 +1474,12 @@ static int nfs_inode_attrs_need_update(const struct inode *inode, const struct n
 		((long)nfsi->attr_gencount - (long)nfs_read_attr_generation_counter() > 0);
 }
 
-/*
- * Don't trust the change_attribute, mtime, ctime or size if
- * a pnfs LAYOUTCOMMIT is outstanding
- */
-static void nfs_inode_attrs_handle_layoutcommit(struct inode *inode,
-		struct nfs_fattr *fattr)
-{
-	if (pnfs_layoutcommit_outstanding(inode))
-		fattr->valid &= ~(NFS_ATTR_FATTR_CHANGE |
-				NFS_ATTR_FATTR_MTIME |
-				NFS_ATTR_FATTR_CTIME |
-				NFS_ATTR_FATTR_SIZE);
-}
-
 static int nfs_refresh_inode_locked(struct inode *inode, struct nfs_fattr *fattr)
 {
 	int ret;
 
 	trace_nfs_refresh_inode_enter(inode);
 
-	nfs_inode_attrs_handle_layoutcommit(inode, fattr);
-
 	if (nfs_inode_attrs_need_update(inode, fattr))
 		ret = nfs_update_inode(inode, fattr);
 	else
@@ -1527,7 +1515,7 @@ EXPORT_SYMBOL_GPL(nfs_refresh_inode);
 
 static int nfs_post_op_update_inode_locked(struct inode *inode, struct nfs_fattr *fattr)
 {
-	unsigned long invalid = NFS_INO_INVALID_ATTR|NFS_INO_REVAL_PAGECACHE;
+	unsigned long invalid = NFS_INO_INVALID_ATTR;
 
 	/*
 	 * Don't revalidate the pagecache if we hold a delegation, but do
@@ -1676,7 +1664,8 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
 	unsigned long invalid = 0;
 	unsigned long now = jiffies;
 	unsigned long save_cache_validity;
-	bool cache_revalidated = true;
+	bool have_writers = nfs_file_has_buffered_writers(nfsi);
+	bool cache_revalidated;
 
 	dfprintk(VFS, "NFS: %s(%s/%lu fh_crc=0x%08x ct=%d info=0x%x)\n",
 			__func__, inode->i_sb->s_id, inode->i_ino,
@@ -1725,17 +1714,23 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
 	/* Do atomic weak cache consistency updates */
 	invalid |= nfs_wcc_update_inode(inode, fattr);
 
+
+	cache_revalidated = !pnfs_layoutcommit_outstanding(inode);
+
 	/* More cache consistency checks */
 	if (fattr->valid & NFS_ATTR_FATTR_CHANGE) {
 		if (inode->i_version != fattr->change_attr) {
 			dprintk("NFS: change_attr change on server for file %s/%ld\n",
 					inode->i_sb->s_id, inode->i_ino);
-			invalid |= NFS_INO_INVALID_ATTR
-				| NFS_INO_INVALID_DATA
-				| NFS_INO_INVALID_ACCESS
-				| NFS_INO_INVALID_ACL;
-			if (S_ISDIR(inode->i_mode))
-				nfs_force_lookup_revalidate(inode);
+			/* Could it be a race with writeback? */
+			if (!have_writers) {
+				invalid |= NFS_INO_INVALID_ATTR
+					| NFS_INO_INVALID_DATA
+					| NFS_INO_INVALID_ACCESS
+					| NFS_INO_INVALID_ACL;
+				if (S_ISDIR(inode->i_mode))
+					nfs_force_lookup_revalidate(inode);
+			}
 			inode->i_version = fattr->change_attr;
 		}
 	} else {
@@ -1768,9 +1763,10 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
 		if (new_isize != cur_isize) {
 			/* Do we perhaps have any outstanding writes, or has
 			 * the file grown beyond our last write? */
-			if ((nfsi->nrequests == 0) || new_isize > cur_isize) {
+			if (nfsi->nrequests == 0 || new_isize > cur_isize) {
 				i_size_write(inode, new_isize);
-				invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_DATA;
+				if (!have_writers)
+					invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_DATA;
 			}
 			dprintk("NFS: isize change on server for file %s/%ld "
 					"(%Ld to %Ld)\n",
diff --git a/fs/nfs/internal.h b/fs/nfs/internal.h
index f7e33a5984d8..8de509b65e8d 100644
--- a/fs/nfs/internal.h
+++ b/fs/nfs/internal.h
@@ -412,6 +412,19 @@ extern void __exit unregister_nfs_fs(void);
 extern bool nfs_sb_active(struct super_block *sb);
 extern void nfs_sb_deactive(struct super_block *sb);
 
+/* io.c */
+extern void nfs_start_io_read(struct inode *inode);
+extern void nfs_end_io_read(struct inode *inode);
+extern void nfs_start_io_write(struct inode *inode);
+extern void nfs_end_io_write(struct inode *inode);
+extern void nfs_start_io_direct(struct inode *inode);
+extern void nfs_end_io_direct(struct inode *inode);
+
+static inline bool nfs_file_io_is_buffered(struct nfs_inode *nfsi)
+{
+	return test_bit(NFS_INO_ODIRECT, &nfsi->flags) == 0;
+}
+
 /* namespace.c */
 #define NFS_PATH_CANONICAL 1
 extern char *nfs_path(char **p, struct dentry *dentry,
@@ -500,6 +513,26 @@ int nfs_key_timeout_notify(struct file *filp, struct inode *inode);
 bool nfs_ctx_key_to_expire(struct nfs_open_context *ctx, struct inode *inode);
 void nfs_pageio_stop_mirroring(struct nfs_pageio_descriptor *pgio);
 
+int nfs_filemap_write_and_wait_range(struct address_space *mapping,
+		loff_t lstart, loff_t lend);
+
+#ifdef CONFIG_NFS_V4_1
+static inline
+void nfs_clear_pnfs_ds_commit_verifiers(struct pnfs_ds_commit_info *cinfo)
+{
+	int i;
+
+	for (i = 0; i < cinfo->nbuckets; i++)
+		cinfo->buckets[i].direct_verf.committed = NFS_INVALID_STABLE_HOW;
+}
+#else
+static inline
+void nfs_clear_pnfs_ds_commit_verifiers(struct pnfs_ds_commit_info *cinfo)
+{
+}
+#endif
+
+
 #ifdef CONFIG_MIGRATION
 extern int nfs_migrate_page(struct address_space *,
 		struct page *, struct page *, enum migrate_mode);
@@ -507,6 +540,13 @@ extern int nfs_migrate_page(struct address_space *,
 #define nfs_migrate_page NULL
 #endif
 
+static inline int
+nfs_write_verifier_cmp(const struct nfs_write_verifier *v1,
+		const struct nfs_write_verifier *v2)
+{
+	return memcmp(v1->data, v2->data, sizeof(v1->data));
+}
+
 /* unlink.c */
 extern struct rpc_task *
 nfs_async_rename(struct inode *old_dir, struct inode *new_dir,
diff --git a/fs/nfs/io.c b/fs/nfs/io.c
new file mode 100644
index 000000000000..1fc5d1ce327e
--- /dev/null
+++ b/fs/nfs/io.c
@@ -0,0 +1,147 @@
+/*
+ * Copyright (c) 2016 Trond Myklebust
+ *
+ * I/O and data path helper functionality.
+ */
+
+#include <linux/types.h>
+#include <linux/kernel.h>
+#include <linux/bitops.h>
+#include <linux/rwsem.h>
+#include <linux/fs.h>
+#include <linux/nfs_fs.h>
+
+#include "internal.h"
+
+/* Call with exclusively locked inode->i_rwsem */
+static void nfs_block_o_direct(struct nfs_inode *nfsi, struct inode *inode)
+{
+	if (test_bit(NFS_INO_ODIRECT, &nfsi->flags)) {
+		clear_bit(NFS_INO_ODIRECT, &nfsi->flags);
+		inode_dio_wait(inode);
+	}
+}
+
+/**
+ * nfs_start_io_read - declare the file is being used for buffered reads
+ * @inode - file inode
+ *
+ * Declare that a buffered read operation is about to start, and ensure
+ * that we block all direct I/O.
+ * On exit, the function ensures that the NFS_INO_ODIRECT flag is unset,
+ * and holds a shared lock on inode->i_rwsem to ensure that the flag
+ * cannot be changed.
+ * In practice, this means that buffered read operations are allowed to
+ * execute in parallel, thanks to the shared lock, whereas direct I/O
+ * operations need to wait to grab an exclusive lock in order to set
+ * NFS_INO_ODIRECT.
+ * Note that buffered writes and truncates both take a write lock on
+ * inode->i_rwsem, meaning that those are serialised w.r.t. the reads.
+ */
+void
+nfs_start_io_read(struct inode *inode)
+{
+	struct nfs_inode *nfsi = NFS_I(inode);
+	/* Be an optimist! */
+	down_read(&inode->i_rwsem);
+	if (test_bit(NFS_INO_ODIRECT, &nfsi->flags) == 0)
+		return;
+	up_read(&inode->i_rwsem);
+	/* Slow path.... */
+	down_write(&inode->i_rwsem);
+	nfs_block_o_direct(nfsi, inode);
+	downgrade_write(&inode->i_rwsem);
+}
+
+/**
+ * nfs_end_io_read - declare that the buffered read operation is done
+ * @inode - file inode
+ *
+ * Declare that a buffered read operation is done, and release the shared
+ * lock on inode->i_rwsem.
+ */
+void
+nfs_end_io_read(struct inode *inode)
+{
+	up_read(&inode->i_rwsem);
+}
+
+/**
+ * nfs_start_io_write - declare the file is being used for buffered writes
+ * @inode - file inode
+ *
+ * Declare that a buffered read operation is about to start, and ensure
+ * that we block all direct I/O.
+ */
+void
+nfs_start_io_write(struct inode *inode)
+{
+	down_write(&inode->i_rwsem);
+	nfs_block_o_direct(NFS_I(inode), inode);
+}
+
+/**
+ * nfs_end_io_write - declare that the buffered write operation is done
+ * @inode - file inode
+ *
+ * Declare that a buffered write operation is done, and release the
+ * lock on inode->i_rwsem.
+ */
+void
+nfs_end_io_write(struct inode *inode)
+{
+	up_write(&inode->i_rwsem);
+}
+
+/* Call with exclusively locked inode->i_rwsem */
+static void nfs_block_buffered(struct nfs_inode *nfsi, struct inode *inode)
+{
+	if (!test_bit(NFS_INO_ODIRECT, &nfsi->flags)) {
+		set_bit(NFS_INO_ODIRECT, &nfsi->flags);
+		nfs_wb_all(inode);
+	}
+}
+
+/**
+ * nfs_end_io_direct - declare the file is being used for direct i/o
+ * @inode - file inode
+ *
+ * Declare that a direct I/O operation is about to start, and ensure
+ * that we block all buffered I/O.
+ * On exit, the function ensures that the NFS_INO_ODIRECT flag is set,
+ * and holds a shared lock on inode->i_rwsem to ensure that the flag
+ * cannot be changed.
+ * In practice, this means that direct I/O operations are allowed to
+ * execute in parallel, thanks to the shared lock, whereas buffered I/O
+ * operations need to wait to grab an exclusive lock in order to clear
+ * NFS_INO_ODIRECT.
+ * Note that buffered writes and truncates both take a write lock on
+ * inode->i_rwsem, meaning that those are serialised w.r.t. O_DIRECT.
+ */
+void
+nfs_start_io_direct(struct inode *inode)
+{
+	struct nfs_inode *nfsi = NFS_I(inode);
+	/* Be an optimist! */
+	down_read(&inode->i_rwsem);
+	if (test_bit(NFS_INO_ODIRECT, &nfsi->flags) != 0)
+		return;
+	up_read(&inode->i_rwsem);
+	/* Slow path.... */
+	down_write(&inode->i_rwsem);
+	nfs_block_buffered(nfsi, inode);
+	downgrade_write(&inode->i_rwsem);
+}
+
+/**
+ * nfs_end_io_direct - declare that the direct i/o operation is done
+ * @inode - file inode
+ *
+ * Declare that a direct I/O operation is done, and release the shared
+ * lock on inode->i_rwsem.
+ */
+void
+nfs_end_io_direct(struct inode *inode)
+{
+	up_read(&inode->i_rwsem);
+}
diff --git a/fs/nfs/nfs42proc.c b/fs/nfs/nfs42proc.c
index aa03ed09ba06..616dc254b38b 100644
--- a/fs/nfs/nfs42proc.c
+++ b/fs/nfs/nfs42proc.c
@@ -113,15 +113,17 @@ int nfs42_proc_deallocate(struct file *filep, loff_t offset, loff_t len)
 	if (!nfs_server_capable(inode, NFS_CAP_DEALLOCATE))
 		return -EOPNOTSUPP;
 
-	nfs_wb_all(inode);
 	inode_lock(inode);
+	err = nfs_sync_inode(inode);
+	if (err)
+		goto out_unlock;
 
 	err = nfs42_proc_fallocate(&msg, filep, offset, len);
 	if (err == 0)
 		truncate_pagecache_range(inode, offset, (offset + len) -1);
 	if (err == -EOPNOTSUPP)
 		NFS_SERVER(inode)->caps &= ~NFS_CAP_DEALLOCATE;
-
+out_unlock:
 	inode_unlock(inode);
 	return err;
 }
@@ -154,11 +156,20 @@ static ssize_t _nfs42_proc_copy(struct file *src, loff_t pos_src,
 	if (status)
 		return status;
 
+	status = nfs_filemap_write_and_wait_range(file_inode(src)->i_mapping,
+			pos_src, pos_src + (loff_t)count - 1);
+	if (status)
+		return status;
+
 	status = nfs4_set_rw_stateid(&args.dst_stateid, dst_lock->open_context,
 				     dst_lock, FMODE_WRITE);
 	if (status)
 		return status;
 
+	status = nfs_sync_inode(dst_inode);
+	if (status)
+		return status;
+
 	status = nfs4_call_sync(server->client, server, &msg,
 				&args.seq_args, &res.seq_res, 0);
 	if (status == -ENOTSUPP)
@@ -258,7 +269,11 @@ static loff_t _nfs42_proc_llseek(struct file *filep,
 	if (status)
 		return status;
 
-	nfs_wb_all(inode);
+	status = nfs_filemap_write_and_wait_range(inode->i_mapping,
+			offset, LLONG_MAX);
+	if (status)
+		return status;
+
 	status = nfs4_call_sync(server->client, server, &msg,
 				&args.seq_args, &res.seq_res, 0);
 	if (status == -ENOTSUPP)
diff --git a/fs/nfs/nfs4file.c b/fs/nfs/nfs4file.c
index 014b0e41ace5..d085ad794884 100644
--- a/fs/nfs/nfs4file.c
+++ b/fs/nfs/nfs4file.c
@@ -66,7 +66,7 @@ nfs4_file_open(struct inode *inode, struct file *filp)
 	if (openflags & O_TRUNC) {
 		attr.ia_valid |= ATTR_SIZE;
 		attr.ia_size = 0;
-		nfs_sync_inode(inode);
+		filemap_write_and_wait(inode->i_mapping);
 	}
 
 	inode = NFS_PROTO(dir)->open_context(dir, ctx, openflags, &attr, NULL);
@@ -133,21 +133,9 @@ static ssize_t nfs4_copy_file_range(struct file *file_in, loff_t pos_in,
 				    struct file *file_out, loff_t pos_out,
 				    size_t count, unsigned int flags)
 {
-	struct inode *in_inode = file_inode(file_in);
-	struct inode *out_inode = file_inode(file_out);
-	int ret;
-
-	if (in_inode == out_inode)
+	if (file_inode(file_in) == file_inode(file_out))
 		return -EINVAL;
 
-	/* flush any pending writes */
-	ret = nfs_sync_inode(in_inode);
-	if (ret)
-		return ret;
-	ret = nfs_sync_inode(out_inode);
-	if (ret)
-		return ret;
-
 	return nfs42_proc_copy(file_in, pos_in, file_out, pos_out, count);
 }
 
diff --git a/fs/nfs/nfs4xdr.c b/fs/nfs/nfs4xdr.c
index 661e753fe1c9..7bd3a5c09d31 100644
--- a/fs/nfs/nfs4xdr.c
+++ b/fs/nfs/nfs4xdr.c
@@ -1985,9 +1985,14 @@ encode_layoutcommit(struct xdr_stream *xdr,
 	p = xdr_encode_hyper(p, args->lastbytewritten + 1);	/* length */
 	*p = cpu_to_be32(0); /* reclaim */
 	encode_nfs4_stateid(xdr, &args->stateid);
-	p = reserve_space(xdr, 20);
-	*p++ = cpu_to_be32(1); /* newoffset = TRUE */
-	p = xdr_encode_hyper(p, args->lastbytewritten);
+	if (args->lastbytewritten != U64_MAX) {
+		p = reserve_space(xdr, 20);
+		*p++ = cpu_to_be32(1); /* newoffset = TRUE */
+		p = xdr_encode_hyper(p, args->lastbytewritten);
+	} else {
+		p = reserve_space(xdr, 12);
+		*p++ = cpu_to_be32(0); /* newoffset = FALSE */
+	}
 	*p++ = cpu_to_be32(0); /* Never send time_modify_changed */
 	*p++ = cpu_to_be32(NFS_SERVER(args->inode)->pnfs_curr_ld->id);/* type */
 
diff --git a/fs/nfs/nfstrace.h b/fs/nfs/nfstrace.h
index 0b9e5cc9a747..fe80a1c26340 100644
--- a/fs/nfs/nfstrace.h
+++ b/fs/nfs/nfstrace.h
@@ -37,7 +37,6 @@
 			{ 1 << NFS_INO_ADVISE_RDPLUS, "ADVISE_RDPLUS" }, \
 			{ 1 << NFS_INO_STALE, "STALE" }, \
 			{ 1 << NFS_INO_INVALIDATING, "INVALIDATING" }, \
-			{ 1 << NFS_INO_FLUSHING, "FLUSHING" }, \
 			{ 1 << NFS_INO_FSCACHE, "FSCACHE" }, \
 			{ 1 << NFS_INO_LAYOUTCOMMIT, "NEED_LAYOUTCOMMIT" }, \
 			{ 1 << NFS_INO_LAYOUTCOMMITTING, "LAYOUTCOMMIT" })
diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
index 7d992362ff04..4110c1dc8f68 100644
--- a/fs/nfs/pnfs.c
+++ b/fs/nfs/pnfs.c
@@ -2393,7 +2393,10 @@ pnfs_layoutcommit_inode(struct inode *inode, bool sync)
 	nfs_fattr_init(&data->fattr);
 	data->args.bitmask = NFS_SERVER(inode)->cache_consistency_bitmask;
 	data->res.fattr = &data->fattr;
-	data->args.lastbytewritten = end_pos - 1;
+	if (end_pos != 0)
+		data->args.lastbytewritten = end_pos - 1;
+	else
+		data->args.lastbytewritten = U64_MAX;
 	data->res.server = NFS_SERVER(inode);
 
 	if (ld->prepare_layoutcommit) {
diff --git a/fs/nfs/pnfs.h b/fs/nfs/pnfs.h
index b21bd0bee784..181283c4ebc3 100644
--- a/fs/nfs/pnfs.h
+++ b/fs/nfs/pnfs.h
@@ -629,6 +629,13 @@ pnfs_sync_inode(struct inode *inode, bool datasync)
 }
 
 static inline bool
+pnfs_layoutcommit_outstanding(struct inode *inode)
+{
+	return false;
+}
+
+
+static inline bool
 pnfs_roc(struct inode *ino)
 {
 	return false;
@@ -716,13 +723,6 @@ pnfs_use_threshold(struct nfs4_threshold **dst, struct nfs4_threshold *src,
 	return false;
 }
 
-static inline bool
-pnfs_layoutcommit_outstanding(struct inode *inode)
-{
-	return false;
-}
-
-
 static inline struct nfs4_threshold *pnfs_mdsthreshold_alloc(void)
 {
 	return NULL;
diff --git a/fs/nfs/pnfs_nfs.c b/fs/nfs/pnfs_nfs.c
index fe183fbc4b90..f3468b57a32a 100644
--- a/fs/nfs/pnfs_nfs.c
+++ b/fs/nfs/pnfs_nfs.c
@@ -940,6 +940,13 @@ EXPORT_SYMBOL_GPL(pnfs_layout_mark_request_commit);
 int
 pnfs_nfs_generic_sync(struct inode *inode, bool datasync)
 {
+	int ret;
+
+	if (!pnfs_layoutcommit_outstanding(inode))
+		return 0;
+	ret = nfs_commit_inode(inode, FLUSH_SYNC);
+	if (ret < 0)
+		return ret;
 	if (datasync)
 		return 0;
 	return pnfs_layoutcommit_inode(inode, true);
diff --git a/fs/nfs/write.c b/fs/nfs/write.c
index b5f3da346f1d..510c5befdf9b 100644
--- a/fs/nfs/write.c
+++ b/fs/nfs/write.c
@@ -625,7 +625,7 @@ static int nfs_writepage_locked(struct page *page,
 	int err;
 
 	nfs_inc_stats(inode, NFSIOS_VFSWRITEPAGE);
-	nfs_pageio_init_write(&pgio, inode, wb_priority(wbc),
+	nfs_pageio_init_write(&pgio, inode, 0,
 				false, &nfs_async_write_completion_ops);
 	err = nfs_do_writepage(page, wbc, &pgio, launder);
 	nfs_pageio_complete(&pgio);
@@ -657,16 +657,9 @@ static int nfs_writepages_callback(struct page *page, struct writeback_control *
 int nfs_writepages(struct address_space *mapping, struct writeback_control *wbc)
 {
 	struct inode *inode = mapping->host;
-	unsigned long *bitlock = &NFS_I(inode)->flags;
 	struct nfs_pageio_descriptor pgio;
 	int err;
 
-	/* Stop dirtying of new pages while we sync */
-	err = wait_on_bit_lock_action(bitlock, NFS_INO_FLUSHING,
-			nfs_wait_bit_killable, TASK_KILLABLE);
-	if (err)
-		goto out_err;
-
 	nfs_inc_stats(inode, NFSIOS_VFSWRITEPAGES);
 
 	nfs_pageio_init_write(&pgio, inode, wb_priority(wbc), false,
@@ -674,10 +667,6 @@ int nfs_writepages(struct address_space *mapping, struct writeback_control *wbc)
 	err = write_cache_pages(mapping, wbc, nfs_writepages_callback, &pgio);
 	nfs_pageio_complete(&pgio);
 
-	clear_bit_unlock(NFS_INO_FLUSHING, bitlock);
-	smp_mb__after_atomic();
-	wake_up_bit(bitlock, NFS_INO_FLUSHING);
-
 	if (err < 0)
 		goto out_err;
 	err = pgio.pg_error;
@@ -1805,7 +1794,7 @@ static void nfs_commit_release_pages(struct nfs_commit_data *data)
 
 		/* Okay, COMMIT succeeded, apparently. Check the verifier
 		 * returned by the server against all stored verfs. */
-		if (!memcmp(&req->wb_verf, &data->verf.verifier, sizeof(req->wb_verf))) {
+		if (!nfs_write_verifier_cmp(&req->wb_verf, &data->verf.verifier)) {
 			/* We have a match */
 			nfs_inode_remove_request(req);
 			dprintk(" OK\n");
@@ -1929,6 +1918,24 @@ out_mark_dirty:
 EXPORT_SYMBOL_GPL(nfs_write_inode);
 
 /*
+ * Wrapper for filemap_write_and_wait_range()
+ *
+ * Needed for pNFS in order to ensure data becomes visible to the
+ * client.
+ */
+int nfs_filemap_write_and_wait_range(struct address_space *mapping,
+		loff_t lstart, loff_t lend)
+{
+	int ret;
+
+	ret = filemap_write_and_wait_range(mapping, lstart, lend);
+	if (ret == 0)
+		ret = pnfs_sync_inode(mapping->host, true);
+	return ret;
+}
+EXPORT_SYMBOL_GPL(nfs_filemap_write_and_wait_range);
+
+/*
  * flush the inode to disk.
  */
 int nfs_wb_all(struct inode *inode)