1 files changed, 948 insertions, 749 deletions

diff --git a/fs/nfs/write.c b/fs/nfs/write.c
index 834f0fe96f8..5e2f1030454 100644
--- a/fs/nfs/write.c
+++ b/fs/nfs/write.c
@@ -31,6 +31,8 @@
 #include "fscache.h"
 #include "pnfs.h"
 
+#include "nfstrace.h"
+
 #define NFSDBG_FACILITY		NFSDBG_PAGECACHE
 
 #define MIN_POOL_WRITE		(32)
@@ -39,20 +41,21 @@
 /*
  * Local function declarations
  */
-static void nfs_pageio_init_write(struct nfs_pageio_descriptor *desc,
-				  struct inode *inode, int ioflags);
 static void nfs_redirty_request(struct nfs_page *req);
-static const struct rpc_call_ops nfs_write_partial_ops;
-static const struct rpc_call_ops nfs_write_full_ops;
 static const struct rpc_call_ops nfs_commit_ops;
+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 struct kmem_cache *nfs_wdata_cachep;
 static mempool_t *nfs_wdata_mempool;
+static struct kmem_cache *nfs_cdata_cachep;
 static mempool_t *nfs_commit_mempool;
 
-struct nfs_write_data *nfs_commitdata_alloc(void)
+struct nfs_commit_data *nfs_commitdata_alloc(void)
 {
-	struct nfs_write_data *p = mempool_alloc(nfs_commit_mempool, GFP_NOFS);
+	struct nfs_commit_data *p = mempool_alloc(nfs_commit_mempool, GFP_NOIO);
 
 	if (p) {
 		memset(p, 0, sizeof(*p));
@@ -62,47 +65,24 @@ struct nfs_write_data *nfs_commitdata_alloc(void)
 }
 EXPORT_SYMBOL_GPL(nfs_commitdata_alloc);
 
-void nfs_commit_free(struct nfs_write_data *p)
+void nfs_commit_free(struct nfs_commit_data *p)
 {
-	if (p && (p->pagevec != &p->page_array[0]))
-		kfree(p->pagevec);
 	mempool_free(p, nfs_commit_mempool);
 }
 EXPORT_SYMBOL_GPL(nfs_commit_free);
 
-struct nfs_write_data *nfs_writedata_alloc(unsigned int pagecount)
+static struct nfs_rw_header *nfs_writehdr_alloc(void)
 {
-	struct nfs_write_data *p = mempool_alloc(nfs_wdata_mempool, GFP_NOFS);
+	struct nfs_rw_header *p = mempool_alloc(nfs_wdata_mempool, GFP_NOIO);
 
-	if (p) {
+	if (p)
 		memset(p, 0, sizeof(*p));
-		INIT_LIST_HEAD(&p->pages);
-		p->npages = pagecount;
-		if (pagecount <= ARRAY_SIZE(p->page_array))
-			p->pagevec = p->page_array;
-		else {
-			p->pagevec = kcalloc(pagecount, sizeof(struct page *), GFP_NOFS);
-			if (!p->pagevec) {
-				mempool_free(p, nfs_wdata_mempool);
-				p = NULL;
-			}
-		}
-	}
 	return p;
 }
 
-void nfs_writedata_free(struct nfs_write_data *p)
-{
-	if (p && (p->pagevec != &p->page_array[0]))
-		kfree(p->pagevec);
-	mempool_free(p, nfs_wdata_mempool);
-}
-
-void nfs_writedata_release(struct nfs_write_data *wdata)
+static void nfs_writehdr_free(struct nfs_rw_header *whdr)
 {
-	put_lseg(wdata->lseg);
-	put_nfs_open_context(wdata->args.context);
-	nfs_writedata_free(wdata);
+	mempool_free(whdr, nfs_wdata_mempool);
 }
 
 static void nfs_context_set_write_error(struct nfs_open_context *ctx, int error)
@@ -112,25 +92,53 @@ static void nfs_context_set_write_error(struct nfs_open_context *ctx, int error)
 	set_bit(NFS_CONTEXT_ERROR_WRITE, &ctx->flags);
 }
 
-static struct nfs_page *nfs_page_find_request_locked(struct page *page)
+/*
+ * nfs_page_find_head_request_locked - find head request associated with @page
+ *
+ * must be called while holding the inode lock.
+ *
+ * returns matching head request with reference held, or NULL if not found.
+ */
+static struct nfs_page *
+nfs_page_find_head_request_locked(struct nfs_inode *nfsi, struct page *page)
 {
 	struct nfs_page *req = NULL;
 
-	if (PagePrivate(page)) {
+	if (PagePrivate(page))
 		req = (struct nfs_page *)page_private(page);
-		if (req != NULL)
-			kref_get(&req->wb_kref);
+	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;
+			}
+		}
 	}
+
+	if (req) {
+		WARN_ON_ONCE(req->wb_head != req);
+
+		kref_get(&req->wb_kref);
+	}
+
 	return req;
 }
 
-static struct nfs_page *nfs_page_find_request(struct page *page)
+/*
+ * nfs_page_find_head_request - find head request associated with @page
+ *
+ * returns matching head request with reference held, or NULL if not found.
+ */
+static struct nfs_page *nfs_page_find_head_request(struct page *page)
 {
-	struct inode *inode = page->mapping->host;
+	struct inode *inode = page_file_mapping(page)->host;
 	struct nfs_page *req = NULL;
 
 	spin_lock(&inode->i_lock);
-	req = nfs_page_find_request_locked(page);
+	req = nfs_page_find_head_request_locked(NFS_I(inode), page);
 	spin_unlock(&inode->i_lock);
 	return req;
 }
@@ -138,16 +146,16 @@ static struct nfs_page *nfs_page_find_request(struct page *page)
 /* Adjust the file length if we're writing beyond the end */
 static void nfs_grow_file(struct page *page, unsigned int offset, unsigned int count)
 {
-	struct inode *inode = page->mapping->host;
+	struct inode *inode = page_file_mapping(page)->host;
 	loff_t end, i_size;
 	pgoff_t end_index;
 
 	spin_lock(&inode->i_lock);
 	i_size = i_size_read(inode);
 	end_index = (i_size - 1) >> PAGE_CACHE_SHIFT;
-	if (i_size > 0 && page->index < end_index)
+	if (i_size > 0 && page_file_index(page) < end_index)
 		goto out;
-	end = ((loff_t)page->index << PAGE_CACHE_SHIFT) + ((loff_t)offset+count);
+	end = page_file_offset(page) + ((loff_t)offset+count);
 	if (i_size >= end)
 		goto out;
 	i_size_write(inode, end);
@@ -159,22 +167,81 @@ out:
 /* A writeback failed: mark the page as bad, and invalidate the page cache */
 static void nfs_set_pageerror(struct page *page)
 {
-	SetPageError(page);
-	nfs_zap_mapping(page->mapping->host, page->mapping);
+	nfs_zap_mapping(page_file_mapping(page)->host, page_file_mapping(page));
+}
+
+/*
+ * nfs_page_group_search_locked
+ * @head - head request of page group
+ * @page_offset - offset into page
+ *
+ * Search page group with head @head to find a request that contains the
+ * page offset @page_offset.
+ *
+ * Returns a pointer to the first matching nfs request, or NULL if no
+ * match is found.
+ *
+ * Must be called with the page group lock held
+ */
+static struct nfs_page *
+nfs_page_group_search_locked(struct nfs_page *head, unsigned int page_offset)
+{
+	struct nfs_page *req;
+
+	WARN_ON_ONCE(head != head->wb_head);
+	WARN_ON_ONCE(!test_bit(PG_HEADLOCK, &head->wb_head->wb_flags));
+
+	req = head;
+	do {
+		if (page_offset >= req->wb_pgbase &&
+		    page_offset < (req->wb_pgbase + req->wb_bytes))
+			return req;
+
+		req = req->wb_this_page;
+	} while (req != head);
+
+	return NULL;
+}
+
+/*
+ * nfs_page_group_covers_page
+ * @head - head request of page group
+ *
+ * Return true if the page group with head @head covers the whole page,
+ * returns false otherwise
+ */
+static bool nfs_page_group_covers_page(struct nfs_page *req)
+{
+	struct nfs_page *tmp;
+	unsigned int pos = 0;
+	unsigned int len = nfs_page_length(req->wb_page);
+
+	nfs_page_group_lock(req);
+
+	do {
+		tmp = nfs_page_group_search_locked(req->wb_head, pos);
+		if (tmp) {
+			/* no way this should happen */
+			WARN_ON_ONCE(tmp->wb_pgbase != pos);
+			pos += tmp->wb_bytes - (pos - tmp->wb_pgbase);
+		}
+	} while (tmp && pos < len);
+
+	nfs_page_group_unlock(req);
+	WARN_ON_ONCE(pos > len);
+	return pos == len;
 }
 
 /* We can set the PG_uptodate flag if we see that a write request
  * covers the full page.
  */
-static void nfs_mark_uptodate(struct page *page, unsigned int base, unsigned int count)
+static void nfs_mark_uptodate(struct nfs_page *req)
 {
-	if (PageUptodate(page))
-		return;
-	if (base != 0)
+	if (PageUptodate(req->wb_page))
 		return;
-	if (count != nfs_page_length(page))
+	if (!nfs_page_group_covers_page(req))
 		return;
-	SetPageUptodate(page);
+	SetPageUptodate(req->wb_page);
 }
 
 static int wb_priority(struct writeback_control *wbc)
@@ -196,65 +263,273 @@ int nfs_congestion_kb;
 #define NFS_CONGESTION_OFF_THRESH	\
 	(NFS_CONGESTION_ON_THRESH - (NFS_CONGESTION_ON_THRESH >> 2))
 
-static int nfs_set_page_writeback(struct page *page)
+static void nfs_set_page_writeback(struct page *page)
 {
+	struct nfs_server *nfss = NFS_SERVER(page_file_mapping(page)->host);
 	int ret = test_set_page_writeback(page);
 
-	if (!ret) {
-		struct inode *inode = page->mapping->host;
-		struct nfs_server *nfss = NFS_SERVER(inode);
+	WARN_ON_ONCE(ret != 0);
 
-		page_cache_get(page);
-		if (atomic_long_inc_return(&nfss->writeback) >
-				NFS_CONGESTION_ON_THRESH) {
-			set_bdi_congested(&nfss->backing_dev_info,
-						BLK_RW_ASYNC);
-		}
+	if (atomic_long_inc_return(&nfss->writeback) >
+			NFS_CONGESTION_ON_THRESH) {
+		set_bdi_congested(&nfss->backing_dev_info,
+					BLK_RW_ASYNC);
 	}
-	return ret;
 }
 
-static void nfs_end_page_writeback(struct page *page)
+static void nfs_end_page_writeback(struct nfs_page *req)
 {
-	struct inode *inode = page->mapping->host;
+	struct inode *inode = page_file_mapping(req->wb_page)->host;
 	struct nfs_server *nfss = NFS_SERVER(inode);
 
-	end_page_writeback(page);
-	page_cache_release(page);
+	if (!nfs_page_group_sync_on_bit(req, PG_WB_END))
+		return;
+
+	end_page_writeback(req->wb_page);
 	if (atomic_long_dec_return(&nfss->writeback) < NFS_CONGESTION_OFF_THRESH)
 		clear_bdi_congested(&nfss->backing_dev_info, BLK_RW_ASYNC);
 }
 
-static struct nfs_page *nfs_find_and_lock_request(struct page *page, bool nonblock)
+
+/* nfs_page_group_clear_bits
+ *   @req - an nfs request
+ * clears all page group related bits from @req
+ */
+static void
+nfs_page_group_clear_bits(struct nfs_page *req)
 {
-	struct inode *inode = page->mapping->host;
-	struct nfs_page *req;
+	clear_bit(PG_TEARDOWN, &req->wb_flags);
+	clear_bit(PG_UNLOCKPAGE, &req->wb_flags);
+	clear_bit(PG_UPTODATE, &req->wb_flags);
+	clear_bit(PG_WB_END, &req->wb_flags);
+	clear_bit(PG_REMOVE, &req->wb_flags);
+}
+
+
+/*
+ * nfs_unroll_locks_and_wait -  unlock all newly locked reqs and wait on @req
+ *
+ * this is a helper function for nfs_lock_and_join_requests
+ *
+ * @inode - inode associated with request page group, must be holding inode lock
+ * @head  - head request of page group, must be holding head lock
+ * @req   - request that couldn't lock and needs to wait on the req bit lock
+ * @nonblock - if true, don't actually wait
+ *
+ * NOTE: this must be called holding page_group bit lock and inode spin lock
+ *       and BOTH will be released before returning.
+ *
+ * returns 0 on success, < 0 on error.
+ */
+static int
+nfs_unroll_locks_and_wait(struct inode *inode, struct nfs_page *head,
+			  struct nfs_page *req, bool nonblock)
+	__releases(&inode->i_lock)
+{
+	struct nfs_page *tmp;
+	int ret;
+
+	/* relinquish all the locks successfully grabbed this run */
+	for (tmp = head ; tmp != req; tmp = tmp->wb_this_page)
+		nfs_unlock_request(tmp);
+
+	WARN_ON_ONCE(test_bit(PG_TEARDOWN, &req->wb_flags));
+
+	/* grab a ref on the request that will be waited on */
+	kref_get(&req->wb_kref);
+
+	nfs_page_group_unlock(head);
+	spin_unlock(&inode->i_lock);
+
+	/* release ref from nfs_page_find_head_request_locked */
+	nfs_release_request(head);
+
+	if (!nonblock)
+		ret = nfs_wait_on_request(req);
+	else
+		ret = -EAGAIN;
+	nfs_release_request(req);
+
+	return ret;
+}
+
+/*
+ * nfs_destroy_unlinked_subrequests - destroy recently unlinked subrequests
+ *
+ * @destroy_list - request list (using wb_this_page) terminated by @old_head
+ * @old_head - the old head of the list
+ *
+ * All subrequests must be locked and removed from all lists, so at this point
+ * they are only "active" in this function, and possibly in nfs_wait_on_request
+ * with a reference held by some other context.
+ */
+static void
+nfs_destroy_unlinked_subrequests(struct nfs_page *destroy_list,
+				 struct nfs_page *old_head)
+{
+	while (destroy_list) {
+		struct nfs_page *subreq = destroy_list;
+
+		destroy_list = (subreq->wb_this_page == old_head) ?
+				   NULL : subreq->wb_this_page;
+
+		WARN_ON_ONCE(old_head != subreq->wb_head);
+
+		/* make sure old group is not used */
+		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);
+
+		if (!test_bit(PG_TEARDOWN, &subreq->wb_flags)) {
+			/* release ref on old head request */
+			nfs_release_request(old_head);
+
+			nfs_page_group_clear_bits(subreq);
+
+			/* release the PG_INODE_REF reference */
+			if (test_and_clear_bit(PG_INODE_REF, &subreq->wb_flags))
+				nfs_release_request(subreq);
+			else
+				WARN_ON_ONCE(1);
+		} else {
+			WARN_ON_ONCE(test_bit(PG_CLEAN, &subreq->wb_flags));
+			/* zombie requests have already released the last
+			 * reference and were waiting on the rest of the
+			 * group to complete. Since it's no longer part of a
+			 * group, simply free the request */
+			nfs_page_group_clear_bits(subreq);
+			nfs_free_request(subreq);
+		}
+	}
+}
+
+/*
+ * nfs_lock_and_join_requests - join all subreqs to the head req and return
+ *                              a locked reference, cancelling any pending
+ *                              operations for this page.
+ *
+ * @page - the page used to lookup the "page group" of nfs_page structures
+ * @nonblock - if true, don't block waiting for request locks
+ *
+ * This function joins all sub requests to the head request by first
+ * locking all requests in the group, cancelling any pending operations
+ * and finally updating the head request to cover the whole range covered by
+ * the (former) group.  All subrequests are removed from any write or commit
+ * lists, unlinked from the group and destroyed.
+ *
+ * Returns a locked, referenced pointer to the head request - which after
+ * this call is guaranteed to be the only request associated with the page.
+ * Returns NULL if no requests are found for @page, or a ERR_PTR if an
+ * error was encountered.
+ */
+static struct nfs_page *
+nfs_lock_and_join_requests(struct page *page, bool nonblock)
+{
+	struct inode *inode = page_file_mapping(page)->host;
+	struct nfs_page *head, *subreq;
+	struct nfs_page *destroy_list = NULL;
+	unsigned int total_bytes;
 	int ret;
 
+try_again:
+	total_bytes = 0;
+
+	WARN_ON_ONCE(destroy_list);
+
 	spin_lock(&inode->i_lock);
-	for (;;) {
-		req = nfs_page_find_request_locked(page);
-		if (req == NULL)
-			break;
-		if (nfs_set_page_tag_locked(req))
-			break;
-		/* Note: If we hold the page lock, as is the case in nfs_writepage,
-		 *	 then the call to nfs_set_page_tag_locked() will always
-		 *	 succeed provided that someone hasn't already marked the
-		 *	 request as dirty (in which case we don't care).
-		 */
+
+	/*
+	 * A reference is taken only on the head request which acts as a
+	 * reference to the whole page group - the group will not be destroyed
+	 * until the head reference is released.
+	 */
+	head = nfs_page_find_head_request_locked(NFS_I(inode), page);
+
+	if (!head) {
 		spin_unlock(&inode->i_lock);
-		if (!nonblock)
-			ret = nfs_wait_on_request(req);
-		else
-			ret = -EAGAIN;
-		nfs_release_request(req);
-		if (ret != 0)
+		return NULL;
+	}
+
+	/* lock each request in the page group */
+	nfs_page_group_lock(head);
+	subreq = head;
+	do {
+		/*
+		 * Subrequests are always contiguous, non overlapping
+		 * and in order. If not, it's a programming error.
+		 */
+		WARN_ON_ONCE(subreq->wb_offset !=
+		     (head->wb_offset + total_bytes));
+
+		/* keep track of how many bytes this group covers */
+		total_bytes += subreq->wb_bytes;
+
+		if (!nfs_lock_request(subreq)) {
+			/* releases page group bit lock and
+			 * inode spin lock and all references */
+			ret = nfs_unroll_locks_and_wait(inode, head,
+				subreq, nonblock);
+
+			if (ret == 0)
+				goto try_again;
+
 			return ERR_PTR(ret);
-		spin_lock(&inode->i_lock);
+		}
+
+		subreq = subreq->wb_this_page;
+	} while (subreq != head);
+
+	/* Now that all requests are locked, make sure they aren't on any list.
+	 * Commit list removal accounting is done after locks are dropped */
+	subreq = head;
+	do {
+		nfs_list_remove_request(subreq);
+		subreq = subreq->wb_this_page;
+	} while (subreq != head);
+
+	/* unlink subrequests from head, destroy them later */
+	if (head->wb_this_page != head) {
+		/* destroy list will be terminated by head */
+		destroy_list = head->wb_this_page;
+		head->wb_this_page = head;
+
+		/* change head request to cover whole range that
+		 * the former page group covered */
+		head->wb_bytes = total_bytes;
 	}
+
+	/*
+	 * prepare head request to be added to new pgio descriptor
+	 */
+	nfs_page_group_clear_bits(head);
+
+	/*
+	 * some part of the group was still on the inode list - otherwise
+	 * the group wouldn't be involved in async write.
+	 * grab a reference for the head request, iff it needs one.
+	 */
+	if (!test_and_set_bit(PG_INODE_REF, &head->wb_flags))
+		kref_get(&head->wb_kref);
+
+	nfs_page_group_unlock(head);
+
+	/* drop lock to clear_request_commit the head req and clean up
+	 * requests on destroy list */
 	spin_unlock(&inode->i_lock);
-	return req;
+
+	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;
 }
 
 /*
@@ -267,17 +542,17 @@ static int nfs_page_async_flush(struct nfs_pageio_descriptor *pgio,
 	struct nfs_page *req;
 	int ret = 0;
 
-	req = nfs_find_and_lock_request(page, nonblock);
+	req = nfs_lock_and_join_requests(page, nonblock);
 	if (!req)
 		goto out;
 	ret = PTR_ERR(req);
 	if (IS_ERR(req))
 		goto out;
 
-	ret = nfs_set_page_writeback(page);
-	BUG_ON(ret != 0);
-	BUG_ON(test_bit(PG_CLEAN, &req->wb_flags));
+	nfs_set_page_writeback(page);
+	WARN_ON_ONCE(test_bit(PG_CLEAN, &req->wb_flags));
 
+	ret = 0;
 	if (!nfs_pageio_add_request(pgio, req)) {
 		nfs_redirty_request(req);
 		ret = pgio->pg_error;
@@ -288,13 +563,13 @@ out:
 
 static int nfs_do_writepage(struct page *page, struct writeback_control *wbc, struct nfs_pageio_descriptor *pgio)
 {
-	struct inode *inode = page->mapping->host;
+	struct inode *inode = page_file_mapping(page)->host;
 	int ret;
 
 	nfs_inc_stats(inode, NFSIOS_VFSWRITEPAGE);
 	nfs_add_stats(inode, NFSIOS_WRITEPAGES, 1);
 
-	nfs_pageio_cond_complete(pgio, page->index);
+	nfs_pageio_cond_complete(pgio, page_file_index(page));
 	ret = nfs_page_async_flush(pgio, page, wbc->sync_mode == WB_SYNC_NONE);
 	if (ret == -EAGAIN) {
 		redirty_page_for_writepage(wbc, page);
@@ -311,7 +586,8 @@ static int nfs_writepage_locked(struct page *page, struct writeback_control *wbc
 	struct nfs_pageio_descriptor pgio;
 	int err;
 
-	nfs_pageio_init_write(&pgio, page->mapping->host, wb_priority(wbc));
+	nfs_pageio_init_write(&pgio, page->mapping->host, wb_priority(wbc),
+				false, &nfs_async_write_completion_ops);
 	err = nfs_do_writepage(page, wbc, &pgio);
 	nfs_pageio_complete(&pgio);
 	if (err < 0)
@@ -354,12 +630,13 @@ int nfs_writepages(struct address_space *mapping, struct writeback_control *wbc)
 
 	nfs_inc_stats(inode, NFSIOS_VFSWRITEPAGES);
 
-	nfs_pageio_init_write(&pgio, inode, wb_priority(wbc));
+	nfs_pageio_init_write(&pgio, inode, wb_priority(wbc), false,
+				&nfs_async_write_completion_ops);
 	err = write_cache_pages(mapping, wbc, nfs_writepages_callback, &pgio);
 	nfs_pageio_complete(&pgio);
 
 	clear_bit_unlock(NFS_INO_FLUSHING, bitlock);
-	smp_mb__after_clear_bit();
+	smp_mb__after_atomic();
 	wake_up_bit(bitlock, NFS_INO_FLUSHING);
 
 	if (err < 0)
@@ -375,34 +652,33 @@ out_err:
 /*
  * Insert a write request into an inode
  */
-static int nfs_inode_add_request(struct inode *inode, struct nfs_page *req)
+static void nfs_inode_add_request(struct inode *inode, struct nfs_page *req)
 {
 	struct nfs_inode *nfsi = NFS_I(inode);
-	int error;
 
-	error = radix_tree_preload(GFP_NOFS);
-	if (error != 0)
-		goto out;
+	WARN_ON_ONCE(req->wb_this_page != req);
 
 	/* Lock the request! */
-	nfs_lock_request_dontget(req);
+	nfs_lock_request(req);
 
 	spin_lock(&inode->i_lock);
-	error = radix_tree_insert(&nfsi->nfs_page_tree, req->wb_index, req);
-	BUG_ON(error);
-	if (!nfsi->npages && nfs_have_delegation(inode, FMODE_WRITE))
+	if (!nfsi->npages && NFS_PROTO(inode)->have_delegation(inode, FMODE_WRITE))
 		inode->i_version++;
-	set_bit(PG_MAPPED, &req->wb_flags);
-	SetPagePrivate(req->wb_page);
-	set_page_private(req->wb_page, (unsigned long)req);
+	/*
+	 * Swap-space should not get truncated. Hence no need to plug the race
+	 * with invalidate/truncate.
+	 */
+	if (likely(!PageSwapCache(req->wb_page))) {
+		set_bit(PG_MAPPED, &req->wb_flags);
+		SetPagePrivate(req->wb_page);
+		set_page_private(req->wb_page, (unsigned long)req);
+	}
 	nfsi->npages++;
+	/* this a head request for a page group - mark it as having an
+	 * extra reference so sub groups can follow suit */
+	WARN_ON(test_and_set_bit(PG_INODE_REF, &req->wb_flags));
 	kref_get(&req->wb_kref);
-	radix_tree_tag_set(&nfsi->nfs_page_tree, req->wb_index,
-				NFS_PAGE_TAG_LOCKED);
 	spin_unlock(&inode->i_lock);
-	radix_tree_preload_end();
-out:
-	return error;
 }
 
 /*
@@ -412,17 +688,23 @@ static void nfs_inode_remove_request(struct nfs_page *req)
 {
 	struct inode *inode = req->wb_context->dentry->d_inode;
 	struct nfs_inode *nfsi = NFS_I(inode);
+	struct nfs_page *head;
 
-	BUG_ON (!NFS_WBACK_BUSY(req));
+	if (nfs_page_group_sync_on_bit(req, PG_REMOVE)) {
+		head = req->wb_head;
 
-	spin_lock(&inode->i_lock);
-	set_page_private(req->wb_page, 0);
-	ClearPagePrivate(req->wb_page);
-	clear_bit(PG_MAPPED, &req->wb_flags);
-	radix_tree_delete(&nfsi->nfs_page_tree, req->wb_index);
-	nfsi->npages--;
-	spin_unlock(&inode->i_lock);
-	nfs_release_request(req);
+		spin_lock(&inode->i_lock);
+		if (likely(!PageSwapCache(head->wb_page))) {
+			set_page_private(head->wb_page, 0);
+			ClearPagePrivate(head->wb_page);
+			clear_bit(PG_MAPPED, &head->wb_flags);
+		}
+		nfsi->npages--;
+		spin_unlock(&inode->i_lock);
+	}
+
+	if (test_and_clear_bit(PG_INODE_REF, &req->wb_flags))
+		nfs_release_request(req);
 }
 
 static void
@@ -431,135 +713,261 @@ nfs_mark_request_dirty(struct nfs_page *req)
 	__set_page_dirty_nobuffers(req->wb_page);
 }
 
-#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
+#if IS_ENABLED(CONFIG_NFS_V3) || IS_ENABLED(CONFIG_NFS_V4)
+/**
+ * nfs_request_add_commit_list - add request to a commit list
+ * @req: pointer to a struct nfs_page
+ * @dst: commit list head
+ * @cinfo: holds list lock and accounting info
+ *
+ * This sets the PG_CLEAN bit, updates the cinfo count of
+ * number of outstanding requests requiring a commit as well as
+ * the MM page stats.
+ *
+ * The caller must _not_ hold the cinfo->lock, but must be
+ * holding the nfs_page lock.
+ */
+void
+nfs_request_add_commit_list(struct nfs_page *req, struct list_head *dst,
+			    struct nfs_commit_info *cinfo)
+{
+	set_bit(PG_CLEAN, &(req)->wb_flags);
+	spin_lock(cinfo->lock);
+	nfs_list_add_request(req, dst);
+	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);
+	}
+}
+EXPORT_SYMBOL_GPL(nfs_request_add_commit_list);
+
+/**
+ * nfs_request_remove_commit_list - Remove request from a commit list
+ * @req: pointer to a nfs_page
+ * @cinfo: holds list lock and accounting info
+ *
+ * This clears the PG_CLEAN bit, and updates the cinfo's count of
+ * number of outstanding requests requiring a commit
+ * It does not update the MM page stats.
+ *
+ * The caller _must_ hold the cinfo->lock and the nfs_page lock.
+ */
+void
+nfs_request_remove_commit_list(struct nfs_page *req,
+			       struct nfs_commit_info *cinfo)
+{
+	if (!test_and_clear_bit(PG_CLEAN, &(req)->wb_flags))
+		return;
+	nfs_list_remove_request(req);
+	cinfo->mds->ncommit--;
+}
+EXPORT_SYMBOL_GPL(nfs_request_remove_commit_list);
+
+static void nfs_init_cinfo_from_inode(struct nfs_commit_info *cinfo,
+				      struct inode *inode)
+{
+	cinfo->lock = &inode->i_lock;
+	cinfo->mds = &NFS_I(inode)->commit_info;
+	cinfo->ds = pnfs_get_ds_info(inode);
+	cinfo->dreq = NULL;
+	cinfo->completion_ops = &nfs_commit_completion_ops;
+}
+
+void nfs_init_cinfo(struct nfs_commit_info *cinfo,
+		    struct inode *inode,
+		    struct nfs_direct_req *dreq)
+{
+	if (dreq)
+		nfs_init_cinfo_from_dreq(cinfo, dreq);
+	else
+		nfs_init_cinfo_from_inode(cinfo, inode);
+}
+EXPORT_SYMBOL_GPL(nfs_init_cinfo);
+
 /*
  * Add a request to the inode's commit list.
  */
-static void
-nfs_mark_request_commit(struct nfs_page *req, struct pnfs_layout_segment *lseg)
+void
+nfs_mark_request_commit(struct nfs_page *req, struct pnfs_layout_segment *lseg,
+			struct nfs_commit_info *cinfo)
 {
-	struct inode *inode = req->wb_context->dentry->d_inode;
-	struct nfs_inode *nfsi = NFS_I(inode);
+	if (pnfs_mark_request_commit(req, lseg, cinfo))
+		return;
+	nfs_request_add_commit_list(req, &cinfo->mds->list, cinfo);
+}
 
-	spin_lock(&inode->i_lock);
-	set_bit(PG_CLEAN, &(req)->wb_flags);
-	radix_tree_tag_set(&nfsi->nfs_page_tree,
-			req->wb_index,
-			NFS_PAGE_TAG_COMMIT);
-	nfsi->ncommit++;
-	spin_unlock(&inode->i_lock);
-	pnfs_mark_request_commit(req, lseg);
-	inc_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
-	inc_bdi_stat(req->wb_page->mapping->backing_dev_info, BDI_RECLAIMABLE);
-	__mark_inode_dirty(inode, I_DIRTY_DATASYNC);
+static void
+nfs_clear_page_commit(struct page *page)
+{
+	dec_zone_page_state(page, NR_UNSTABLE_NFS);
+	dec_bdi_stat(page_file_mapping(page)->backing_dev_info, BDI_RECLAIMABLE);
 }
 
-static int
+static void
 nfs_clear_request_commit(struct nfs_page *req)
 {
-	struct page *page = req->wb_page;
+	if (test_bit(PG_CLEAN, &req->wb_flags)) {
+		struct inode *inode = req->wb_context->dentry->d_inode;
+		struct nfs_commit_info cinfo;
 
-	if (test_and_clear_bit(PG_CLEAN, &(req)->wb_flags)) {
-		dec_zone_page_state(page, NR_UNSTABLE_NFS);
-		dec_bdi_stat(page->mapping->backing_dev_info, BDI_RECLAIMABLE);
-		return 1;
+		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);
 	}
-	return 0;
 }
 
 static inline
-int nfs_write_need_commit(struct nfs_write_data *data)
+int nfs_write_need_commit(struct nfs_pgio_data *data)
 {
 	if (data->verf.committed == NFS_DATA_SYNC)
-		return data->lseg == NULL;
-	else
-		return data->verf.committed != NFS_FILE_SYNC;
+		return data->header->lseg == NULL;
+	return data->verf.committed != NFS_FILE_SYNC;
 }
 
-static inline
-int nfs_reschedule_unstable_write(struct nfs_page *req,
-				  struct nfs_write_data *data)
+#else
+static void nfs_init_cinfo_from_inode(struct nfs_commit_info *cinfo,
+				      struct inode *inode)
 {
-	if (test_and_clear_bit(PG_NEED_COMMIT, &req->wb_flags)) {
-		nfs_mark_request_commit(req, data->lseg);
-		return 1;
-	}
-	if (test_and_clear_bit(PG_NEED_RESCHED, &req->wb_flags)) {
-		nfs_mark_request_dirty(req);
-		return 1;
-	}
-	return 0;
 }
-#else
-static inline void
-nfs_mark_request_commit(struct nfs_page *req, struct pnfs_layout_segment *lseg)
+
+void nfs_init_cinfo(struct nfs_commit_info *cinfo,
+		    struct inode *inode,
+		    struct nfs_direct_req *dreq)
 {
 }
 
-static inline int
-nfs_clear_request_commit(struct nfs_page *req)
+void
+nfs_mark_request_commit(struct nfs_page *req, struct pnfs_layout_segment *lseg,
+			struct nfs_commit_info *cinfo)
 {
-	return 0;
 }
 
-static inline
-int nfs_write_need_commit(struct nfs_write_data *data)
+static void
+nfs_clear_request_commit(struct nfs_page *req)
 {
-	return 0;
 }
 
 static inline
-int nfs_reschedule_unstable_write(struct nfs_page *req,
-				  struct nfs_write_data *data)
+int nfs_write_need_commit(struct nfs_pgio_data *data)
 {
 	return 0;
 }
+
 #endif
 
-#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
-static int
-nfs_need_commit(struct nfs_inode *nfsi)
+static void nfs_write_completion(struct nfs_pgio_header *hdr)
+{
+	struct nfs_commit_info cinfo;
+	unsigned long bytes = 0;
+
+	if (test_bit(NFS_IOHDR_REDO, &hdr->flags))
+		goto out;
+	nfs_init_cinfo_from_inode(&cinfo, hdr->inode);
+	while (!list_empty(&hdr->pages)) {
+		struct nfs_page *req = nfs_list_entry(hdr->pages.next);
+
+		bytes += req->wb_bytes;
+		nfs_list_remove_request(req);
+		if (test_bit(NFS_IOHDR_ERROR, &hdr->flags) &&
+		    (hdr->good_bytes < bytes)) {
+			nfs_set_pageerror(req->wb_page);
+			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)) {
+			memcpy(&req->wb_verf, &hdr->verf.verifier, sizeof(req->wb_verf));
+			nfs_mark_request_commit(req, hdr->lseg, &cinfo);
+			goto next;
+		}
+remove_req:
+		nfs_inode_remove_request(req);
+next:
+		nfs_unlock_request(req);
+		nfs_end_page_writeback(req);
+		nfs_release_request(req);
+	}
+out:
+	hdr->release(hdr);
+}
+
+#if  IS_ENABLED(CONFIG_NFS_V3) || IS_ENABLED(CONFIG_NFS_V4)
+unsigned long
+nfs_reqs_to_commit(struct nfs_commit_info *cinfo)
+{
+	return cinfo->mds->ncommit;
+}
+
+/* cinfo->lock held by caller */
+int
+nfs_scan_commit_list(struct list_head *src, struct list_head *dst,
+		     struct nfs_commit_info *cinfo, int max)
 {
-	return radix_tree_tagged(&nfsi->nfs_page_tree, NFS_PAGE_TAG_COMMIT);
+	struct nfs_page *req, *tmp;
+	int ret = 0;
+
+	list_for_each_entry_safe(req, tmp, src, wb_list) {
+		if (!nfs_lock_request(req))
+			continue;
+		kref_get(&req->wb_kref);
+		if (cond_resched_lock(cinfo->lock))
+			list_safe_reset_next(req, tmp, wb_list);
+		nfs_request_remove_commit_list(req, cinfo);
+		nfs_list_add_request(req, dst);
+		ret++;
+		if ((ret == max) && !cinfo->dreq)
+			break;
+	}
+	return ret;
 }
 
 /*
  * nfs_scan_commit - Scan an inode for commit requests
  * @inode: NFS inode to scan
- * @dst: destination list
- * @idx_start: lower bound of page->index to scan.
- * @npages: idx_start + npages sets the upper bound to scan.
+ * @dst: mds destination list
+ * @cinfo: mds and ds lists of reqs ready to commit
  *
  * Moves requests from the inode's 'commit' request list.
  * The requests are *not* checked to ensure that they form a contiguous set.
  */
-static int
-nfs_scan_commit(struct inode *inode, struct list_head *dst, pgoff_t idx_start, unsigned int npages)
+int
+nfs_scan_commit(struct inode *inode, struct list_head *dst,
+		struct nfs_commit_info *cinfo)
 {
-	struct nfs_inode *nfsi = NFS_I(inode);
-	int ret;
-
-	if (!nfs_need_commit(nfsi))
-		return 0;
-
-	spin_lock(&inode->i_lock);
-	ret = nfs_scan_list(nfsi, dst, idx_start, npages, NFS_PAGE_TAG_COMMIT);
-	if (ret > 0)
-		nfsi->ncommit -= ret;
-	spin_unlock(&inode->i_lock);
+	int ret = 0;
 
-	if (nfs_need_commit(NFS_I(inode)))
-		__mark_inode_dirty(inode, I_DIRTY_DATASYNC);
+	spin_lock(cinfo->lock);
+	if (cinfo->mds->ncommit > 0) {
+		const int max = INT_MAX;
 
+		ret = nfs_scan_commit_list(&cinfo->mds->list, dst,
+					   cinfo, max);
+		ret += pnfs_scan_commit_lists(inode, cinfo, max - ret);
+	}
+	spin_unlock(cinfo->lock);
 	return ret;
 }
+
 #else
-static inline int nfs_need_commit(struct nfs_inode *nfsi)
+unsigned long nfs_reqs_to_commit(struct nfs_commit_info *cinfo)
 {
 	return 0;
 }
 
-static inline int nfs_scan_commit(struct inode *inode, struct list_head *dst, pgoff_t idx_start, unsigned int npages)
+int nfs_scan_commit(struct inode *inode, struct list_head *dst,
+		    struct nfs_commit_info *cinfo)
 {
 	return 0;
 }
@@ -589,10 +997,14 @@ static struct nfs_page *nfs_try_to_update_request(struct inode *inode,
 	spin_lock(&inode->i_lock);
 
 	for (;;) {
-		req = nfs_page_find_request_locked(page);
+		req = nfs_page_find_head_request_locked(NFS_I(inode), page);
 		if (req == NULL)
 			goto out_unlock;
 
+		/* should be handled by nfs_flush_incompatible */
+		WARN_ON_ONCE(req->wb_head != req);
+		WARN_ON_ONCE(req->wb_this_page != req);
+
 		rqend = req->wb_offset + req->wb_bytes;
 		/*
 		 * Tell the caller to flush out the request if
@@ -604,7 +1016,7 @@ static struct nfs_page *nfs_try_to_update_request(struct inode *inode,
 		    || end < req->wb_offset)
 			goto out_flushme;
 
-		if (nfs_set_page_tag_locked(req))
+		if (nfs_lock_request(req))
 			break;
 
 		/* The request is locked, so wait and then retry */
@@ -616,13 +1028,6 @@ static struct nfs_page *nfs_try_to_update_request(struct inode *inode,
 		spin_lock(&inode->i_lock);
 	}
 
-	if (nfs_clear_request_commit(req) &&
-	    radix_tree_tag_clear(&NFS_I(inode)->nfs_page_tree,
-				 req->wb_index, NFS_PAGE_TAG_COMMIT) != NULL) {
-		NFS_I(inode)->ncommit--;
-		pnfs_clear_request_commit(req);
-	}
-
 	/* Okay, the request matches. Update the region */
 	if (offset < req->wb_offset) {
 		req->wb_offset = offset;
@@ -634,6 +1039,8 @@ static struct nfs_page *nfs_try_to_update_request(struct inode *inode,
 		req->wb_bytes = rqend - req->wb_offset;
 out_unlock:
 	spin_unlock(&inode->i_lock);
+	if (req)
+		nfs_clear_request_commit(req);
 	return req;
 out_flushme:
 	spin_unlock(&inode->i_lock);
@@ -653,21 +1060,16 @@ out_err:
 static struct nfs_page * nfs_setup_write_request(struct nfs_open_context* ctx,
 		struct page *page, unsigned int offset, unsigned int bytes)
 {
-	struct inode *inode = page->mapping->host;
+	struct inode *inode = page_file_mapping(page)->host;
 	struct nfs_page	*req;
-	int error;
 
 	req = nfs_try_to_update_request(inode, page, offset, bytes);
 	if (req != NULL)
 		goto out;
-	req = nfs_create_request(ctx, inode, page, offset, bytes);
+	req = nfs_create_request(ctx, page, NULL, offset, bytes);
 	if (IS_ERR(req))
 		goto out;
-	error = nfs_inode_add_request(inode, req);
-	if (error != 0) {
-		nfs_release_request(req);
-		req = ERR_PTR(error);
-	}
+	nfs_inode_add_request(inode, req);
 out:
 	return req;
 }
@@ -682,15 +1084,16 @@ static int nfs_writepage_setup(struct nfs_open_context *ctx, struct page *page,
 		return PTR_ERR(req);
 	/* Update file length */
 	nfs_grow_file(page, offset, count);
-	nfs_mark_uptodate(page, req->wb_pgbase, req->wb_bytes);
+	nfs_mark_uptodate(req);
 	nfs_mark_request_dirty(req);
-	nfs_clear_page_tag_locked(req);
+	nfs_unlock_and_release_request(req);
 	return 0;
 }
 
 int nfs_flush_incompatible(struct file *file, struct page *page)
 {
 	struct nfs_open_context *ctx = nfs_file_open_context(file);
+	struct nfs_lock_context *l_ctx;
 	struct nfs_page	*req;
 	int do_flush, status;
 	/*
@@ -702,29 +1105,95 @@ int nfs_flush_incompatible(struct file *file, struct page *page)
 	 * dropped page.
 	 */
 	do {
-		req = nfs_page_find_request(page);
+		req = nfs_page_find_head_request(page);
 		if (req == NULL)
 			return 0;
-		do_flush = req->wb_page != page || req->wb_context != ctx ||
-			req->wb_lock_context->lockowner != current->files ||
-			req->wb_lock_context->pid != current->tgid;
+		l_ctx = req->wb_lock_context;
+		do_flush = req->wb_page != page || req->wb_context != ctx;
+		/* for now, flush if more than 1 request in page_group */
+		do_flush |= req->wb_this_page != req;
+		if (l_ctx && ctx->dentry->d_inode->i_flock != NULL) {
+			do_flush |= l_ctx->lockowner.l_owner != current->files
+				|| l_ctx->lockowner.l_pid != current->tgid;
+		}
 		nfs_release_request(req);
 		if (!do_flush)
 			return 0;
-		status = nfs_wb_page(page->mapping->host, page);
+		status = nfs_wb_page(page_file_mapping(page)->host, page);
 	} while (status == 0);
 	return status;
 }
 
 /*
+ * Avoid buffered writes when a open context credential's key would
+ * expire soon.
+ *
+ * Returns -EACCES if the key will expire within RPC_KEY_EXPIRE_FAIL.
+ *
+ * Return 0 and set a credential flag which triggers the inode to flush
+ * and performs  NFS_FILE_SYNC writes if the key will expired within
+ * RPC_KEY_EXPIRE_TIMEO.
+ */
+int
+nfs_key_timeout_notify(struct file *filp, struct inode *inode)
+{
+	struct nfs_open_context *ctx = nfs_file_open_context(filp);
+	struct rpc_auth *auth = NFS_SERVER(inode)->client->cl_auth;
+
+	return rpcauth_key_timeout_notify(auth, ctx->cred);
+}
+
+/*
+ * Test if the open context credential key is marked to expire soon.
+ */
+bool nfs_ctx_key_to_expire(struct nfs_open_context *ctx)
+{
+	return rpcauth_cred_key_to_expire(ctx->cred);
+}
+
+/*
  * If the page cache is marked as unsafe or invalid, then we can't rely on
  * the PageUptodate() flag. In this case, we will need to turn off
  * write optimisations that depend on the page contents being correct.
  */
-static int nfs_write_pageuptodate(struct page *page, struct inode *inode)
+static bool nfs_write_pageuptodate(struct page *page, struct inode *inode)
+{
+	struct nfs_inode *nfsi = NFS_I(inode);
+
+	if (nfs_have_delegated_attributes(inode))
+		goto out;
+	if (nfsi->cache_validity & NFS_INO_REVAL_PAGECACHE)
+		return false;
+	smp_rmb();
+	if (test_bit(NFS_INO_INVALIDATING, &nfsi->flags))
+		return false;
+out:
+	if (nfsi->cache_validity & NFS_INO_INVALID_DATA)
+		return false;
+	return PageUptodate(page) != 0;
+}
+
+/* If we know the page is up to date, and we're not using byte range locks (or
+ * if we have the whole file locked for writing), it may be more efficient to
+ * extend the write to cover the entire page in order to avoid fragmentation
+ * inefficiencies.
+ *
+ * If the file is opened for synchronous writes then we can just skip the rest
+ * of the checks.
+ */
+static int nfs_can_extend_write(struct file *file, struct page *page, struct inode *inode)
 {
-	return PageUptodate(page) &&
-		!(NFS_I(inode)->cache_validity & (NFS_INO_REVAL_PAGECACHE|NFS_INO_INVALID_DATA));
+	if (file->f_flags & O_DSYNC)
+		return 0;
+	if (!nfs_write_pageuptodate(page, inode))
+		return 0;
+	if (NFS_PROTO(inode)->have_delegation(inode, FMODE_WRITE))
+		return 1;
+	if (inode->i_flock == NULL || (inode->i_flock->fl_start == 0 &&
+			inode->i_flock->fl_end == OFFSET_MAX &&
+			inode->i_flock->fl_type != F_RDLCK))
+		return 1;
+	return 0;
 }
 
 /*
@@ -737,24 +1206,15 @@ int nfs_updatepage(struct file *file, struct page *page,
 		unsigned int offset, unsigned int count)
 {
 	struct nfs_open_context *ctx = nfs_file_open_context(file);
-	struct inode	*inode = page->mapping->host;
+	struct inode	*inode = page_file_mapping(page)->host;
 	int		status = 0;
 
 	nfs_inc_stats(inode, NFSIOS_VFSUPDATEPAGE);
 
-	dprintk("NFS:       nfs_updatepage(%s/%s %d@%lld)\n",
-		file->f_path.dentry->d_parent->d_name.name,
-		file->f_path.dentry->d_name.name, count,
-		(long long)(page_offset(page) + offset));
+	dprintk("NFS:       nfs_updatepage(%pD2 %d@%lld)\n",
+		file, count, (long long)(page_file_offset(page) + offset));
 
-	/* If we're not using byte range locks, and we know the page
-	 * is up to date, it may be more efficient to extend the write
-	 * to cover the entire page in order to avoid fragmentation
-	 * inefficiencies.
-	 */
-	if (nfs_write_pageuptodate(page, inode) &&
-			inode->i_flock == NULL &&
-			!(file->f_flags & O_DSYNC)) {
+	if (nfs_can_extend_write(file, page, inode)) {
 		count = max(count + offset, nfs_page_length(page));
 		offset = 0;
 	}
@@ -770,17 +1230,6 @@ int nfs_updatepage(struct file *file, struct page *page,
 	return status;
 }
 
-static void nfs_writepage_release(struct nfs_page *req,
-				  struct nfs_write_data *data)
-{
-	struct page *page = req->wb_page;
-
-	if (PageError(req->wb_page) || !nfs_reschedule_unstable_write(req, data))
-		nfs_inode_remove_request(req);
-	nfs_clear_page_tag_locked(req);
-	nfs_end_page_writeback(page);
-}
-
 static int flush_task_priority(int how)
 {
 	switch (how & (FLUSH_HIGHPRI|FLUSH_LOWPRI)) {
@@ -792,128 +1241,17 @@ static int flush_task_priority(int how)
 	return RPC_PRIORITY_NORMAL;
 }
 
-int nfs_initiate_write(struct nfs_write_data *data,
-		       struct rpc_clnt *clnt,
-		       const struct rpc_call_ops *call_ops,
-		       int how)
+static void nfs_initiate_write(struct nfs_pgio_data *data, struct rpc_message *msg,
+			       struct rpc_task_setup *task_setup_data, int how)
 {
-	struct inode *inode = data->inode;
+	struct inode *inode = data->header->inode;
 	int priority = flush_task_priority(how);
-	struct rpc_task *task;
-	struct rpc_message msg = {
-		.rpc_argp = &data->args,
-		.rpc_resp = &data->res,
-		.rpc_cred = data->cred,
-	};
-	struct rpc_task_setup task_setup_data = {
-		.rpc_client = clnt,
-		.task = &data->task,
-		.rpc_message = &msg,
-		.callback_ops = call_ops,
-		.callback_data = data,
-		.workqueue = nfsiod_workqueue,
-		.flags = RPC_TASK_ASYNC,
-		.priority = priority,
-	};
-	int ret = 0;
-
-	/* Set up the initial task struct.  */
-	NFS_PROTO(inode)->write_setup(data, &msg);
-
-	dprintk("NFS: %5u initiated write call "
-		"(req %s/%lld, %u bytes @ offset %llu)\n",
-		data->task.tk_pid,
-		inode->i_sb->s_id,
-		(long long)NFS_FILEID(inode),
-		data->args.count,
-		(unsigned long long)data->args.offset);
-
-	task = rpc_run_task(&task_setup_data);
-	if (IS_ERR(task)) {
-		ret = PTR_ERR(task);
-		goto out;
-	}
-	if (how & FLUSH_SYNC) {
-		ret = rpc_wait_for_completion_task(task);
-		if (ret == 0)
-			ret = task->tk_status;
-	}
-	rpc_put_task(task);
-out:
-	return ret;
-}
-EXPORT_SYMBOL_GPL(nfs_initiate_write);
-
-/*
- * Set up the argument/result storage required for the RPC call.
- */
-static void nfs_write_rpcsetup(struct nfs_page *req,
-		struct nfs_write_data *data,
-		unsigned int count, unsigned int offset,
-		int how)
-{
-	struct inode *inode = req->wb_context->dentry->d_inode;
-
-	/* Set up the RPC argument and reply structs
-	 * NB: take care not to mess about with data->commit et al. */
-
-	data->req = req;
-	data->inode = inode = req->wb_context->dentry->d_inode;
-	data->cred = req->wb_context->cred;
-
-	data->args.fh     = NFS_FH(inode);
-	data->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->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;
-	switch (how & (FLUSH_STABLE | FLUSH_COND_STABLE)) {
-	case 0:
-		break;
-	case FLUSH_COND_STABLE:
-		if (nfs_need_commit(NFS_I(inode)))
-			break;
-	default:
-		data->args.stable = NFS_FILE_SYNC;
-	}
-
-	data->res.fattr   = &data->fattr;
-	data->res.count   = count;
-	data->res.verf    = &data->verf;
-	nfs_fattr_init(&data->fattr);
-}
-
-static int nfs_do_write(struct nfs_write_data *data,
-		const struct rpc_call_ops *call_ops,
-		int how)
-{
-	struct inode *inode = data->args.context->dentry->d_inode;
-
-	return nfs_initiate_write(data, NFS_CLIENT(inode), call_ops, how);
-}
 
-static int nfs_do_multiple_writes(struct list_head *head,
-		const struct rpc_call_ops *call_ops,
-		int how)
-{
-	struct nfs_write_data *data;
-	int ret = 0;
+	task_setup_data->priority = priority;
+	NFS_PROTO(inode)->write_setup(data, msg);
 
-	while (!list_empty(head)) {
-		int ret2;
-
-		data = list_entry(head->next, struct nfs_write_data, list);
-		list_del_init(&data->list);
-		
-		ret2 = nfs_do_write(data, call_ops, how);
-		 if (ret == 0)
-			 ret = ret2;
-	}
-	return ret;
+	nfs4_state_protect_write(NFS_SERVER(inode)->nfs_client,
+				 &task_setup_data->rpc_client, msg, data);
 }
 
 /* If a nfs_flush_* function fails, it should remove reqs from @head and
@@ -922,310 +1260,109 @@ static int nfs_do_multiple_writes(struct list_head *head,
  */
 static void nfs_redirty_request(struct nfs_page *req)
 {
-	struct page *page = req->wb_page;
-
 	nfs_mark_request_dirty(req);
-	nfs_clear_page_tag_locked(req);
-	nfs_end_page_writeback(page);
-}
-
-/*
- * Generate multiple small requests to write out a single
- * contiguous dirty area on one page.
- */
-static int nfs_flush_multi(struct nfs_pageio_descriptor *desc, struct list_head *res)
-{
-	struct nfs_page *req = nfs_list_entry(desc->pg_list.next);
-	struct page *page = req->wb_page;
-	struct nfs_write_data *data;
-	size_t wsize = desc->pg_bsize, nbytes;
-	unsigned int offset;
-	int requests = 0;
-	int ret = 0;
-
-	nfs_list_remove_request(req);
-
-	if ((desc->pg_ioflags & FLUSH_COND_STABLE) &&
-	    (desc->pg_moreio || NFS_I(desc->pg_inode)->ncommit ||
-	     desc->pg_count > wsize))
-		desc->pg_ioflags &= ~FLUSH_COND_STABLE;
-
-
-	offset = 0;
-	nbytes = desc->pg_count;
-	do {
-		size_t len = min(nbytes, wsize);
-
-		data = nfs_writedata_alloc(1);
-		if (!data)
-			goto out_bad;
-		data->pagevec[0] = page;
-		nfs_write_rpcsetup(req, data, len, offset, desc->pg_ioflags);
-		list_add(&data->list, res);
-		requests++;
-		nbytes -= len;
-		offset += len;
-	} while (nbytes != 0);
-	atomic_set(&req->wb_complete, requests);
-	desc->pg_rpc_callops = &nfs_write_partial_ops;
-	return ret;
-
-out_bad:
-	while (!list_empty(res)) {
-		data = list_entry(res->next, struct nfs_write_data, list);
-		list_del(&data->list);
-		nfs_writedata_free(data);
-	}
-	nfs_redirty_request(req);
-	return -ENOMEM;
+	nfs_unlock_request(req);
+	nfs_end_page_writeback(req);
+	nfs_release_request(req);
 }
 
-/*
- * Create an RPC task for the given write request and kick it.
- * The page must have been locked by the caller.
- *
- * It may happen that the page we're passed is not marked dirty.
- * This is the case if nfs_updatepage detects a conflicting request
- * that has been written but not committed.
- */
-static int nfs_flush_one(struct nfs_pageio_descriptor *desc, struct list_head *res)
+static void nfs_async_write_error(struct list_head *head)
 {
-	struct nfs_page		*req;
-	struct page		**pages;
-	struct nfs_write_data	*data;
-	struct list_head *head = &desc->pg_list;
-	int ret = 0;
+	struct nfs_page	*req;
 
-	data = nfs_writedata_alloc(nfs_page_array_len(desc->pg_base,
-						      desc->pg_count));
-	if (!data) {
-		while (!list_empty(head)) {
-			req = nfs_list_entry(head->next);
-			nfs_list_remove_request(req);
-			nfs_redirty_request(req);
-		}
-		ret = -ENOMEM;
-		goto out;
-	}
-	pages = data->pagevec;
 	while (!list_empty(head)) {
 		req = nfs_list_entry(head->next);
 		nfs_list_remove_request(req);
-		nfs_list_add_request(req, &data->pages);
-		*pages++ = req->wb_page;
+		nfs_redirty_request(req);
 	}
-	req = nfs_list_entry(data->pages.next);
-
-	if ((desc->pg_ioflags & FLUSH_COND_STABLE) &&
-	    (desc->pg_moreio || NFS_I(desc->pg_inode)->ncommit))
-		desc->pg_ioflags &= ~FLUSH_COND_STABLE;
-
-	/* Set up the argument struct */
-	nfs_write_rpcsetup(req, data, desc->pg_count, 0, desc->pg_ioflags);
-	list_add(&data->list, res);
-	desc->pg_rpc_callops = &nfs_write_full_ops;
-out:
-	return ret;
-}
-
-int nfs_generic_flush(struct nfs_pageio_descriptor *desc, struct list_head *head)
-{
-	if (desc->pg_bsize < PAGE_CACHE_SIZE)
-		return nfs_flush_multi(desc, head);
-	return nfs_flush_one(desc, head);
-}
-
-static int nfs_generic_pg_writepages(struct nfs_pageio_descriptor *desc)
-{
-	LIST_HEAD(head);
-	int ret;
-
-	ret = nfs_generic_flush(desc, &head);
-	if (ret == 0)
-		ret = nfs_do_multiple_writes(&head, desc->pg_rpc_callops,
-				desc->pg_ioflags);
-	return ret;
 }
 
-static const struct nfs_pageio_ops nfs_pageio_write_ops = {
-	.pg_test = nfs_generic_pg_test,
-	.pg_doio = nfs_generic_pg_writepages,
+static const struct nfs_pgio_completion_ops nfs_async_write_completion_ops = {
+	.error_cleanup = nfs_async_write_error,
+	.completion = nfs_write_completion,
 };
 
-void nfs_pageio_init_write_mds(struct nfs_pageio_descriptor *pgio,
-				  struct inode *inode, int ioflags)
+void nfs_pageio_init_write(struct nfs_pageio_descriptor *pgio,
+			       struct inode *inode, int ioflags, bool force_mds,
+			       const struct nfs_pgio_completion_ops *compl_ops)
 {
-	nfs_pageio_init(pgio, inode, &nfs_pageio_write_ops,
-				NFS_SERVER(inode)->wsize, ioflags);
+	struct nfs_server *server = NFS_SERVER(inode);
+	const struct nfs_pageio_ops *pg_ops = &nfs_pgio_rw_ops;
+
+#ifdef CONFIG_NFS_V4_1
+	if (server->pnfs_curr_ld && !force_mds)
+		pg_ops = server->pnfs_curr_ld->pg_write_ops;
+#endif
+	nfs_pageio_init(pgio, inode, pg_ops, compl_ops, &nfs_rw_write_ops,
+			server->wsize, ioflags);
 }
+EXPORT_SYMBOL_GPL(nfs_pageio_init_write);
 
 void nfs_pageio_reset_write_mds(struct nfs_pageio_descriptor *pgio)
 {
-	pgio->pg_ops = &nfs_pageio_write_ops;
+	pgio->pg_ops = &nfs_pgio_rw_ops;
 	pgio->pg_bsize = NFS_SERVER(pgio->pg_inode)->wsize;
 }
 EXPORT_SYMBOL_GPL(nfs_pageio_reset_write_mds);
 
-static void nfs_pageio_init_write(struct nfs_pageio_descriptor *pgio,
-				  struct inode *inode, int ioflags)
-{
-	if (!pnfs_pageio_init_write(pgio, inode, ioflags))
-		nfs_pageio_init_write_mds(pgio, inode, ioflags);
-}
 
-/*
- * Handle a write reply that flushed part of a page.
- */
-static void nfs_writeback_done_partial(struct rpc_task *task, void *calldata)
+void nfs_commit_prepare(struct rpc_task *task, void *calldata)
 {
-	struct nfs_write_data	*data = calldata;
-
-	dprintk("NFS: %5u write(%s/%lld %d@%lld)",
-		task->tk_pid,
-		data->req->wb_context->dentry->d_inode->i_sb->s_id,
-		(long long)
-		  NFS_FILEID(data->req->wb_context->dentry->d_inode),
-		data->req->wb_bytes, (long long)req_offset(data->req));
+	struct nfs_commit_data *data = calldata;
 
-	nfs_writeback_done(task, data);
+	NFS_PROTO(data->inode)->commit_rpc_prepare(task, data);
 }
 
-static void nfs_writeback_release_partial(void *calldata)
+static void nfs_writeback_release_common(struct nfs_pgio_data *data)
 {
-	struct nfs_write_data	*data = calldata;
-	struct nfs_page		*req = data->req;
-	struct page		*page = req->wb_page;
+	struct nfs_pgio_header *hdr = data->header;
 	int status = data->task.tk_status;
 
-	if (status < 0) {
-		nfs_set_pageerror(page);
-		nfs_context_set_write_error(req->wb_context, status);
-		dprintk(", error = %d\n", status);
-		goto out;
+	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);
 	}
-
-	if (nfs_write_need_commit(data)) {
-		struct inode *inode = page->mapping->host;
-
-		spin_lock(&inode->i_lock);
-		if (test_bit(PG_NEED_RESCHED, &req->wb_flags)) {
-			/* Do nothing we need to resend the writes */
-		} else if (!test_and_set_bit(PG_NEED_COMMIT, &req->wb_flags)) {
-			memcpy(&req->wb_verf, &data->verf, sizeof(req->wb_verf));
-			dprintk(" defer commit\n");
-		} else if (memcmp(&req->wb_verf, &data->verf, sizeof(req->wb_verf))) {
-			set_bit(PG_NEED_RESCHED, &req->wb_flags);
-			clear_bit(PG_NEED_COMMIT, &req->wb_flags);
-			dprintk(" server reboot detected\n");
-		}
-		spin_unlock(&inode->i_lock);
-	} else
-		dprintk(" OK\n");
-
-out:
-	if (atomic_dec_and_test(&req->wb_complete))
-		nfs_writepage_release(req, data);
-	nfs_writedata_release(calldata);
-}
-
-#if defined(CONFIG_NFS_V4_1)
-void nfs_write_prepare(struct rpc_task *task, void *calldata)
-{
-	struct nfs_write_data *data = calldata;
-
-	if (nfs4_setup_sequence(NFS_SERVER(data->inode),
-				&data->args.seq_args,
-				&data->res.seq_res, 1, task))
-		return;
-	rpc_call_start(task);
 }
-#endif /* CONFIG_NFS_V4_1 */
-
-static const struct rpc_call_ops nfs_write_partial_ops = {
-#if defined(CONFIG_NFS_V4_1)
-	.rpc_call_prepare = nfs_write_prepare,
-#endif /* CONFIG_NFS_V4_1 */
-	.rpc_call_done = nfs_writeback_done_partial,
-	.rpc_release = nfs_writeback_release_partial,
-};
 
 /*
- * Handle a write reply that flushes a whole page.
- *
- * FIXME: There is an inherent race with invalidate_inode_pages and
- *	  writebacks since the page->count is kept > 1 for as long
- *	  as the page has a write request pending.
+ * Special version of should_remove_suid() that ignores capabilities.
  */
-static void nfs_writeback_done_full(struct rpc_task *task, void *calldata)
-{
-	struct nfs_write_data	*data = calldata;
-
-	nfs_writeback_done(task, data);
-}
-
-static void nfs_writeback_release_full(void *calldata)
+static int nfs_should_remove_suid(const struct inode *inode)
 {
-	struct nfs_write_data	*data = calldata;
-	int status = data->task.tk_status;
-
-	/* Update attributes as result of writeback. */
-	while (!list_empty(&data->pages)) {
-		struct nfs_page *req = nfs_list_entry(data->pages.next);
-		struct page *page = req->wb_page;
+	umode_t mode = inode->i_mode;
+	int kill = 0;
 
-		nfs_list_remove_request(req);
+	/* suid always must be killed */
+	if (unlikely(mode & S_ISUID))
+		kill = ATTR_KILL_SUID;
 
-		dprintk("NFS: %5u write (%s/%lld %d@%lld)",
-			data->task.tk_pid,
-			req->wb_context->dentry->d_inode->i_sb->s_id,
-			(long long)NFS_FILEID(req->wb_context->dentry->d_inode),
-			req->wb_bytes,
-			(long long)req_offset(req));
+	/*
+	 * sgid without any exec bits is just a mandatory locking mark; leave
+	 * it alone.  If some exec bits are set, it's a real sgid; kill it.
+	 */
+	if (unlikely((mode & S_ISGID) && (mode & S_IXGRP)))
+		kill |= ATTR_KILL_SGID;
 
-		if (status < 0) {
-			nfs_set_pageerror(page);
-			nfs_context_set_write_error(req->wb_context, status);
-			dprintk(", error = %d\n", status);
-			goto remove_request;
-		}
+	if (unlikely(kill && S_ISREG(mode)))
+		return kill;
 
-		if (nfs_write_need_commit(data)) {
-			memcpy(&req->wb_verf, &data->verf, sizeof(req->wb_verf));
-			nfs_mark_request_commit(req, data->lseg);
-			dprintk(" marked for commit\n");
-			goto next;
-		}
-		dprintk(" OK\n");
-remove_request:
-		nfs_inode_remove_request(req);
-	next:
-		nfs_clear_page_tag_locked(req);
-		nfs_end_page_writeback(page);
-	}
-	nfs_writedata_release(calldata);
+	return 0;
 }
 
-static const struct rpc_call_ops nfs_write_full_ops = {
-#if defined(CONFIG_NFS_V4_1)
-	.rpc_call_prepare = nfs_write_prepare,
-#endif /* CONFIG_NFS_V4_1 */
-	.rpc_call_done = nfs_writeback_done_full,
-	.rpc_release = nfs_writeback_release_full,
-};
-
-
 /*
  * This function is called when the WRITE call is complete.
  */
-void nfs_writeback_done(struct rpc_task *task, struct nfs_write_data *data)
+static int nfs_writeback_done(struct rpc_task *task, struct nfs_pgio_data *data,
+			      struct inode *inode)
 {
-	struct nfs_writeargs	*argp = &data->args;
-	struct nfs_writeres	*resp = &data->res;
 	int status;
 
-	dprintk("NFS: %5u nfs_writeback_done (status %d)\n",
-		task->tk_pid, task->tk_status);
-
 	/*
 	 * ->write_done will attempt to use post-op attributes to detect
 	 * conflicting writes by other clients.  A strict interpretation
@@ -1233,13 +1370,13 @@ void nfs_writeback_done(struct rpc_task *task, struct nfs_write_data *data)
 	 * another writer had changed the file, but some applications
 	 * depend on tighter cache coherency when writing.
 	 */
-	status = NFS_PROTO(data->inode)->write_done(task, data);
+	status = NFS_PROTO(inode)->write_done(task, data);
 	if (status != 0)
-		return;
-	nfs_add_stats(data->inode, NFSIOS_SERVERWRITTENBYTES, resp->count);
+		return status;
+	nfs_add_stats(inode, NFSIOS_SERVERWRITTENBYTES, data->res.count);
 
-#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
-	if (resp->verf->committed < argp->stable && task->tk_status >= 0) {
+#if IS_ENABLED(CONFIG_NFS_V3) || IS_ENABLED(CONFIG_NFS_V4)
+	if (data->res.verf->committed < data->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.
@@ -1254,50 +1391,64 @@ void nfs_writeback_done(struct rpc_task *task, struct nfs_write_data *data)
 		if (time_before(complain, jiffies)) {
 			dprintk("NFS:       faulty NFS server %s:"
 				" (committed = %d) != (stable = %d)\n",
-				NFS_SERVER(data->inode)->nfs_client->cl_hostname,
-				resp->verf->committed, argp->stable);
+				NFS_SERVER(inode)->nfs_client->cl_hostname,
+				data->res.verf->committed, data->args.stable);
 			complain = jiffies + 300 * HZ;
 		}
 	}
 #endif
-	/* Is this a short write? */
-	if (task->tk_status >= 0 && resp->count < argp->count) {
+
+	/* Deal with the suid/sgid bit corner case */
+	if (nfs_should_remove_suid(inode))
+		nfs_mark_for_revalidate(inode);
+	return 0;
+}
+
+/*
+ * This function is called when the WRITE call is complete.
+ */
+static void nfs_writeback_result(struct rpc_task *task, struct nfs_pgio_data *data)
+{
+	struct nfs_pgio_args	*argp = &data->args;
+	struct nfs_pgio_res	*resp = &data->res;
+
+	if (resp->count < argp->count) {
 		static unsigned long    complain;
 
-		nfs_inc_stats(data->inode, NFSIOS_SHORTWRITE);
+		/* This a short write! */
+		nfs_inc_stats(data->header->inode, NFSIOS_SHORTWRITE);
 
 		/* Has the server at least made some progress? */
-		if (resp->count != 0) {
-			/* 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;
-				argp->offset += resp->count;
-				argp->pgbase += resp->count;
-				argp->count -= resp->count;
-			} else {
-				/* Resend as a stable write in order to avoid
-				 * headaches in the case of a server crash.
-				 */
-				argp->stable = NFS_FILE_SYNC;
+		if (resp->count == 0) {
+			if (time_before(complain, jiffies)) {
+				printk(KERN_WARNING
+				       "NFS: Server wrote zero bytes, expected %u.\n",
+				       argp->count);
+				complain = jiffies + 300 * HZ;
 			}
-			rpc_restart_call_prepare(task);
+			nfs_set_pgio_error(data->header, -EIO, argp->offset);
+			task->tk_status = -EIO;
 			return;
 		}
-		if (time_before(complain, jiffies)) {
-			printk(KERN_WARNING
-			       "NFS: Server wrote zero bytes, expected %u.\n",
-					argp->count);
-			complain = jiffies + 300 * HZ;
+		/* 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;
+			argp->offset += resp->count;
+			argp->pgbase += resp->count;
+			argp->count -= resp->count;
+		} else {
+			/* Resend as a stable write in order to avoid
+			 * headaches in the case of a server crash.
+			 */
+			argp->stable = NFS_FILE_SYNC;
 		}
-		/* Can't do anything about it except throw an error. */
-		task->tk_status = -EIO;
+		rpc_restart_call_prepare(task);
 	}
-	return;
 }
 
 
-#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
+#if IS_ENABLED(CONFIG_NFS_V3) || IS_ENABLED(CONFIG_NFS_V4)
 static int nfs_commit_set_lock(struct nfs_inode *nfsi, int may_wait)
 {
 	int ret;
@@ -1313,27 +1464,23 @@ static int nfs_commit_set_lock(struct nfs_inode *nfsi, int may_wait)
 	return (ret < 0) ? ret : 1;
 }
 
-void nfs_commit_clear_lock(struct nfs_inode *nfsi)
+static void nfs_commit_clear_lock(struct nfs_inode *nfsi)
 {
 	clear_bit(NFS_INO_COMMIT, &nfsi->flags);
-	smp_mb__after_clear_bit();
+	smp_mb__after_atomic();
 	wake_up_bit(&nfsi->flags, NFS_INO_COMMIT);
 }
-EXPORT_SYMBOL_GPL(nfs_commit_clear_lock);
 
-void nfs_commitdata_release(void *data)
+void nfs_commitdata_release(struct nfs_commit_data *data)
 {
-	struct nfs_write_data *wdata = data;
-
-	put_lseg(wdata->lseg);
-	put_nfs_open_context(wdata->args.context);
-	nfs_commit_free(wdata);
+	put_nfs_open_context(data->context);
+	nfs_commit_free(data);
 }
 EXPORT_SYMBOL_GPL(nfs_commitdata_release);
 
-int nfs_initiate_commit(struct nfs_write_data *data, struct rpc_clnt *clnt,
+int nfs_initiate_commit(struct rpc_clnt *clnt, struct nfs_commit_data *data,
 			const struct rpc_call_ops *call_ops,
-			int how)
+			int how, int flags)
 {
 	struct rpc_task *task;
 	int priority = flush_task_priority(how);
@@ -1349,7 +1496,7 @@ int nfs_initiate_commit(struct nfs_write_data *data, struct rpc_clnt *clnt,
 		.callback_ops = call_ops,
 		.callback_data = data,
 		.workqueue = nfsiod_workqueue,
-		.flags = RPC_TASK_ASYNC,
+		.flags = RPC_TASK_ASYNC | flags,
 		.priority = priority,
 	};
 	/* Set up the initial task struct.  */
@@ -1357,6 +1504,9 @@ int nfs_initiate_commit(struct nfs_write_data *data, struct rpc_clnt *clnt,
 
 	dprintk("NFS: %5u initiated commit call\n", data->task.tk_pid);
 
+	nfs4_state_protect(NFS_SERVER(data->inode)->nfs_client,
+		NFS_SP4_MACH_CRED_COMMIT, &task_setup_data.rpc_client, &msg);
+
 	task = rpc_run_task(&task_setup_data);
 	if (IS_ERR(task))
 		return PTR_ERR(task);
@@ -1370,9 +1520,10 @@ EXPORT_SYMBOL_GPL(nfs_initiate_commit);
 /*
  * Set up the argument/result storage required for the RPC call.
  */
-void nfs_init_commit(struct nfs_write_data *data,
-			    struct list_head *head,
-			    struct pnfs_layout_segment *lseg)
+void nfs_init_commit(struct nfs_commit_data *data,
+		     struct list_head *head,
+		     struct pnfs_layout_segment *lseg,
+		     struct nfs_commit_info *cinfo)
 {
 	struct nfs_page *first = nfs_list_entry(head->next);
 	struct inode *inode = first->wb_context->dentry->d_inode;
@@ -1386,13 +1537,14 @@ void nfs_init_commit(struct nfs_write_data *data,
 	data->cred	  = first->wb_context->cred;
 	data->lseg	  = lseg; /* reference transferred */
 	data->mds_ops     = &nfs_commit_ops;
+	data->completion_ops = cinfo->completion_ops;
+	data->dreq	  = cinfo->dreq;
 
 	data->args.fh     = NFS_FH(data->inode);
 	/* Note: we always request a commit of the entire inode */
 	data->args.offset = 0;
 	data->args.count  = 0;
-	data->args.context = get_nfs_open_context(first->wb_context);
-	data->res.count   = 0;
+	data->context     = get_nfs_open_context(first->wb_context);
 	data->res.fattr   = &data->fattr;
 	data->res.verf    = &data->verf;
 	nfs_fattr_init(&data->fattr);
@@ -1400,18 +1552,21 @@ void nfs_init_commit(struct nfs_write_data *data,
 EXPORT_SYMBOL_GPL(nfs_init_commit);
 
 void nfs_retry_commit(struct list_head *page_list,
-		      struct pnfs_layout_segment *lseg)
+		      struct pnfs_layout_segment *lseg,
+		      struct nfs_commit_info *cinfo)
 {
 	struct nfs_page *req;
 
 	while (!list_empty(page_list)) {
 		req = nfs_list_entry(page_list->next);
 		nfs_list_remove_request(req);
-		nfs_mark_request_commit(req, lseg);
-		dec_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
-		dec_bdi_stat(req->wb_page->mapping->backing_dev_info,
-			     BDI_RECLAIMABLE);
-		nfs_clear_page_tag_locked(req);
+		nfs_mark_request_commit(req, lseg, cinfo);
+		if (!cinfo->dreq) {
+			dec_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
+			dec_bdi_stat(page_file_mapping(req->wb_page)->backing_dev_info,
+				     BDI_RECLAIMABLE);
+		}
+		nfs_unlock_and_release_request(req);
 	}
 }
 EXPORT_SYMBOL_GPL(nfs_retry_commit);
@@ -1420,9 +1575,10 @@ EXPORT_SYMBOL_GPL(nfs_retry_commit);
  * Commit dirty pages
  */
 static int
-nfs_commit_list(struct inode *inode, struct list_head *head, int how)
+nfs_commit_list(struct inode *inode, struct list_head *head, int how,
+		struct nfs_commit_info *cinfo)
 {
-	struct nfs_write_data	*data;
+	struct nfs_commit_data	*data;
 
 	data = nfs_commitdata_alloc();
 
@@ -1430,11 +1586,13 @@ nfs_commit_list(struct inode *inode, struct list_head *head, int how)
 		goto out_bad;
 
 	/* Set up the argument struct */
-	nfs_init_commit(data, head, NULL);
-	return nfs_initiate_commit(data, NFS_CLIENT(inode), data->mds_ops, how);
+	nfs_init_commit(data, head, NULL, cinfo);
+	atomic_inc(&cinfo->mds->rpcs_out);
+	return nfs_initiate_commit(NFS_CLIENT(inode), data, data->mds_ops,
+				   how, 0);
  out_bad:
-	nfs_retry_commit(head, NULL);
-	nfs_commit_clear_lock(NFS_I(inode));
+	nfs_retry_commit(head, NULL, cinfo);
+	cinfo->completion_ops->error_cleanup(NFS_I(inode));
 	return -ENOMEM;
 }
 
@@ -1443,7 +1601,7 @@ nfs_commit_list(struct inode *inode, struct list_head *head, int how)
  */
 static void nfs_commit_done(struct rpc_task *task, void *calldata)
 {
-	struct nfs_write_data	*data = calldata;
+	struct nfs_commit_data	*data = calldata;
 
         dprintk("NFS: %5u nfs_commit_done (status %d)\n",
                                 task->tk_pid, task->tk_status);
@@ -1452,19 +1610,20 @@ static void nfs_commit_done(struct rpc_task *task, void *calldata)
 	NFS_PROTO(data->inode)->commit_done(task, data);
 }
 
-void nfs_commit_release_pages(struct nfs_write_data *data)
+static void nfs_commit_release_pages(struct nfs_commit_data *data)
 {
 	struct nfs_page	*req;
 	int status = data->task.tk_status;
+	struct nfs_commit_info cinfo;
 
 	while (!list_empty(&data->pages)) {
 		req = nfs_list_entry(data->pages.next);
 		nfs_list_remove_request(req);
-		nfs_clear_request_commit(req);
+		nfs_clear_page_commit(req->wb_page);
 
-		dprintk("NFS:       commit (%s/%lld %d@%lld)",
+		dprintk("NFS:       commit (%s/%llu %d@%lld)",
 			req->wb_context->dentry->d_sb->s_id,
-			(long long)NFS_FILEID(req->wb_context->dentry->d_inode),
+			(unsigned long long)NFS_FILEID(req->wb_context->dentry->d_inode),
 			req->wb_bytes,
 			(long long)req_offset(req));
 		if (status < 0) {
@@ -1476,7 +1635,7 @@ void nfs_commit_release_pages(struct nfs_write_data *data)
 
 		/* Okay, COMMIT succeeded, apparently. Check the verifier
 		 * returned by the server against all stored verfs. */
-		if (!memcmp(req->wb_verf.verifier, data->verf.verifier, sizeof(data->verf.verifier))) {
+		if (!memcmp(&req->wb_verf, &data->verf.verifier, sizeof(req->wb_verf))) {
 			/* We have a match */
 			nfs_inode_remove_request(req);
 			dprintk(" OK\n");
@@ -1485,45 +1644,61 @@ void nfs_commit_release_pages(struct nfs_write_data *data)
 		/* We have a mismatch. Write the page again */
 		dprintk(" mismatch\n");
 		nfs_mark_request_dirty(req);
+		set_bit(NFS_CONTEXT_RESEND_WRITES, &req->wb_context->flags);
 	next:
-		nfs_clear_page_tag_locked(req);
+		nfs_unlock_and_release_request(req);
 	}
+	nfs_init_cinfo(&cinfo, data->inode, data->dreq);
+	if (atomic_dec_and_test(&cinfo.mds->rpcs_out))
+		nfs_commit_clear_lock(NFS_I(data->inode));
 }
-EXPORT_SYMBOL_GPL(nfs_commit_release_pages);
 
 static void nfs_commit_release(void *calldata)
 {
-	struct nfs_write_data *data = calldata;
+	struct nfs_commit_data *data = calldata;
 
-	nfs_commit_release_pages(data);
-	nfs_commit_clear_lock(NFS_I(data->inode));
+	data->completion_ops->completion(data);
 	nfs_commitdata_release(calldata);
 }
 
 static const struct rpc_call_ops nfs_commit_ops = {
-#if defined(CONFIG_NFS_V4_1)
-	.rpc_call_prepare = nfs_write_prepare,
-#endif /* CONFIG_NFS_V4_1 */
+	.rpc_call_prepare = nfs_commit_prepare,
 	.rpc_call_done = nfs_commit_done,
 	.rpc_release = nfs_commit_release,
 };
 
+static const struct nfs_commit_completion_ops nfs_commit_completion_ops = {
+	.completion = nfs_commit_release_pages,
+	.error_cleanup = nfs_commit_clear_lock,
+};
+
+int nfs_generic_commit_list(struct inode *inode, struct list_head *head,
+			    int how, struct nfs_commit_info *cinfo)
+{
+	int status;
+
+	status = pnfs_commit_list(inode, head, how, cinfo);
+	if (status == PNFS_NOT_ATTEMPTED)
+		status = nfs_commit_list(inode, head, how, cinfo);
+	return status;
+}
+
 int nfs_commit_inode(struct inode *inode, int how)
 {
 	LIST_HEAD(head);
+	struct nfs_commit_info cinfo;
 	int may_wait = how & FLUSH_SYNC;
 	int res;
 
 	res = nfs_commit_set_lock(NFS_I(inode), may_wait);
 	if (res <= 0)
 		goto out_mark_dirty;
-	res = nfs_scan_commit(inode, &head, 0, 0);
+	nfs_init_cinfo_from_inode(&cinfo, inode);
+	res = nfs_scan_commit(inode, &head, &cinfo);
 	if (res) {
 		int error;
 
-		error = pnfs_commit_list(inode, &head, how);
-		if (error == PNFS_NOT_ATTEMPTED)
-			error = nfs_commit_list(inode, &head, how);
+		error = nfs_generic_commit_list(inode, &head, how, &cinfo);
 		if (error < 0)
 			return error;
 		if (!may_wait)
@@ -1554,14 +1729,14 @@ static int nfs_commit_unstable_pages(struct inode *inode, struct writeback_contr
 	int ret = 0;
 
 	/* no commits means nothing needs to be done */
-	if (!nfsi->ncommit)
+	if (!nfsi->commit_info.ncommit)
 		return ret;
 
 	if (wbc->sync_mode == WB_SYNC_NONE) {
 		/* Don't commit yet if this is a non-blocking flush and there
 		 * are a lot of outstanding writes for this mapping.
 		 */
-		if (nfsi->ncommit <= (nfsi->npages >> 1))
+		if (nfsi->commit_info.ncommit <= (nfsi->npages >> 1))
 			goto out_mark_dirty;
 
 		/* don't wait for the COMMIT response */
@@ -1591,22 +1766,9 @@ static int nfs_commit_unstable_pages(struct inode *inode, struct writeback_contr
 
 int nfs_write_inode(struct inode *inode, struct writeback_control *wbc)
 {
-	int ret;
-
-	ret = nfs_commit_unstable_pages(inode, wbc);
-	if (ret >= 0 && test_bit(NFS_INO_LAYOUTCOMMIT, &NFS_I(inode)->flags)) {
-		int status;
-		bool sync = true;
-
-		if (wbc->sync_mode == WB_SYNC_NONE)
-			sync = false;
-
-		status = pnfs_layoutcommit_inode(inode, sync);
-		if (status < 0)
-			return status;
-	}
-	return ret;
+	return nfs_commit_unstable_pages(inode, wbc);
 }
+EXPORT_SYMBOL_GPL(nfs_write_inode);
 
 /*
  * flush the inode to disk.
@@ -1619,36 +1781,44 @@ int nfs_wb_all(struct inode *inode)
 		.range_start = 0,
 		.range_end = LLONG_MAX,
 	};
+	int ret;
+
+	trace_nfs_writeback_inode_enter(inode);
 
-	return sync_inode(inode, &wbc);
+	ret = sync_inode(inode, &wbc);
+
+	trace_nfs_writeback_inode_exit(inode, ret);
+	return ret;
 }
+EXPORT_SYMBOL_GPL(nfs_wb_all);
 
 int nfs_wb_page_cancel(struct inode *inode, struct page *page)
 {
 	struct nfs_page *req;
 	int ret = 0;
 
-	BUG_ON(!PageLocked(page));
-	for (;;) {
-		wait_on_page_writeback(page);
-		req = nfs_page_find_request(page);
-		if (req == NULL)
-			break;
-		if (nfs_lock_request_dontget(req)) {
-			nfs_inode_remove_request(req);
-			/*
-			 * In case nfs_inode_remove_request has marked the
-			 * page as being dirty
-			 */
-			cancel_dirty_page(page, PAGE_CACHE_SIZE);
-			nfs_unlock_request(req);
-			break;
-		}
-		ret = nfs_wait_on_request(req);
-		nfs_release_request(req);
-		if (ret < 0)
-			break;
+	wait_on_page_writeback(page);
+
+	/* blocking call to cancel all requests and join to a single (head)
+	 * request */
+	req = nfs_lock_and_join_requests(page, false);
+
+	if (IS_ERR(req)) {
+		ret = PTR_ERR(req);
+	} else if (req) {
+		/* all requests from this page have been cancelled by
+		 * nfs_lock_and_join_requests, so just remove the head
+		 * request from the inode / page_private pointer and
+		 * release it */
+		nfs_inode_remove_request(req);
+		/*
+		 * In case nfs_inode_remove_request has marked the
+		 * page as being dirty
+		 */
+		cancel_dirty_page(page, PAGE_CACHE_SIZE);
+		nfs_unlock_and_release_request(req);
 	}
+
 	return ret;
 }
 
@@ -1657,7 +1827,7 @@ int nfs_wb_page_cancel(struct inode *inode, struct page *page)
  */
 int nfs_wb_page(struct inode *inode, struct page *page)
 {
-	loff_t range_start = page_offset(page);
+	loff_t range_start = page_file_offset(page);
 	loff_t range_end = range_start + (loff_t)(PAGE_CACHE_SIZE - 1);
 	struct writeback_control wbc = {
 		.sync_mode = WB_SYNC_ALL,
@@ -1667,6 +1837,8 @@ int nfs_wb_page(struct inode *inode, struct page *page)
 	};
 	int ret;
 
+	trace_nfs_writeback_page_enter(inode);
+
 	for (;;) {
 		wait_on_page_writeback(page);
 		if (clear_page_dirty_for_io(page)) {
@@ -1675,14 +1847,15 @@ int nfs_wb_page(struct inode *inode, struct page *page)
 				goto out_error;
 			continue;
 		}
+		ret = 0;
 		if (!PagePrivate(page))
 			break;
 		ret = nfs_commit_inode(inode, FLUSH_SYNC);
 		if (ret < 0)
 			goto out_error;
 	}
-	return 0;
 out_error:
+	trace_nfs_writeback_page_exit(inode, ret);
 	return ret;
 }
 
@@ -1701,7 +1874,8 @@ int nfs_migrate_page(struct address_space *mapping, struct page *newpage,
 	if (PagePrivate(page))
 		return -EBUSY;
 
-	nfs_fscache_release_page(page, GFP_KERNEL);
+	if (!nfs_fscache_release_page(page, GFP_KERNEL))
+		return -EBUSY;
 
 	return migrate_page(mapping, newpage, page, mode);
 }
@@ -1710,7 +1884,7 @@ 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_write_data),
+					     sizeof(struct nfs_rw_header),
 					     0, SLAB_HWCACHE_ALIGN,
 					     NULL);
 	if (nfs_wdata_cachep == NULL)
@@ -1719,12 +1893,19 @@ int __init nfs_init_writepagecache(void)
 	nfs_wdata_mempool = mempool_create_slab_pool(MIN_POOL_WRITE,
 						     nfs_wdata_cachep);
 	if (nfs_wdata_mempool == NULL)
-		return -ENOMEM;
+		goto out_destroy_write_cache;
+
+	nfs_cdata_cachep = kmem_cache_create("nfs_commit_data",
+					     sizeof(struct nfs_commit_data),
+					     0, SLAB_HWCACHE_ALIGN,
+					     NULL);
+	if (nfs_cdata_cachep == NULL)
+		goto out_destroy_write_mempool;
 
 	nfs_commit_mempool = mempool_create_slab_pool(MIN_POOL_COMMIT,
-						      nfs_wdata_cachep);
+						      nfs_cdata_cachep);
 	if (nfs_commit_mempool == NULL)
-		return -ENOMEM;
+		goto out_destroy_commit_cache;
 
 	/*
 	 * NFS congestion size, scale with available memory.
@@ -1747,12 +1928,30 @@ int __init nfs_init_writepagecache(void)
 		nfs_congestion_kb = 256*1024;
 
 	return 0;
+
+out_destroy_commit_cache:
+	kmem_cache_destroy(nfs_cdata_cachep);
+out_destroy_write_mempool:
+	mempool_destroy(nfs_wdata_mempool);
+out_destroy_write_cache:
+	kmem_cache_destroy(nfs_wdata_cachep);
+	return -ENOMEM;
 }
 
 void nfs_destroy_writepagecache(void)
 {
 	mempool_destroy(nfs_commit_mempool);
+	kmem_cache_destroy(nfs_cdata_cachep);
 	mempool_destroy(nfs_wdata_mempool);
 	kmem_cache_destroy(nfs_wdata_cachep);
 }
 
+static const struct nfs_rw_ops nfs_rw_write_ops = {
+	.rw_mode		= FMODE_WRITE,
+	.rw_alloc_header	= nfs_writehdr_alloc,
+	.rw_free_header		= nfs_writehdr_free,
+	.rw_release		= nfs_writeback_release_common,
+	.rw_done		= nfs_writeback_done,
+	.rw_result		= nfs_writeback_result,
+	.rw_initiate		= nfs_initiate_write,
+};