writeback: get rid of wbc->for_writepages
[linux-2.6.git] / fs / nfs / write.c
index 345492e..53eb26c 100644 (file)
@@ -1,47 +1,7 @@
 /*
  * linux/fs/nfs/write.c
  *
- * Writing file data over NFS.
- *
- * We do it like this: When a (user) process wishes to write data to an
- * NFS file, a write request is allocated that contains the RPC task data
- * plus some info on the page to be written, and added to the inode's
- * write chain. If the process writes past the end of the page, an async
- * RPC call to write the page is scheduled immediately; otherwise, the call
- * is delayed for a few seconds.
- *
- * Just like readahead, no async I/O is performed if wsize < PAGE_SIZE.
- *
- * Write requests are kept on the inode's writeback list. Each entry in
- * that list references the page (portion) to be written. When the
- * cache timeout has expired, the RPC task is woken up, and tries to
- * lock the page. As soon as it manages to do so, the request is moved
- * from the writeback list to the writelock list.
- *
- * Note: we must make sure never to confuse the inode passed in the
- * write_page request with the one in page->inode. As far as I understand
- * it, these are different when doing a swap-out.
- *
- * To understand everything that goes on here and in the NFS read code,
- * one should be aware that a page is locked in exactly one of the following
- * cases:
- *
- *  -  A write request is in progress.
- *  -  A user process is in generic_file_write/nfs_update_page
- *  -  A user process is in generic_file_read
- *
- * Also note that because of the way pages are invalidated in
- * nfs_revalidate_inode, the following assertions hold:
- *
- *  -  If a page is dirty, there will be no read requests (a page will
- *     not be re-read unless invalidated by nfs_revalidate_inode).
- *  -  If the page is not uptodate, there will be no pending write
- *     requests, and no process will be in nfs_update_page.
- *
- * FIXME: Interaction with the vmscan routines is not optimal yet.
- * Either vmscan must be made nfs-savvy, or we need a different page
- * reclaim concept that supports something like FS-independent
- * buffer_heads with a b_ops-> field.
+ * Write file data over NFS.
  *
  * Copyright (C) 1996, 1997, Olaf Kirch <okir@monad.swb.de>
  */
@@ -52,6 +12,8 @@
 #include <linux/pagemap.h>
 #include <linux/file.h>
 #include <linux/writeback.h>
+#include <linux/swap.h>
+#include <linux/migrate.h>
 
 #include <linux/sunrpc/clnt.h>
 #include <linux/nfs_fs.h>
 #include <linux/backing-dev.h>
 
 #include <asm/uaccess.h>
-#include <linux/smp_lock.h>
 
 #include "delegation.h"
 #include "internal.h"
 #include "iostat.h"
+#include "nfs4_fs.h"
+#include "fscache.h"
 
 #define NFSDBG_FACILITY                NFSDBG_PAGECACHE
 
 /*
  * Local function declarations
  */
-static struct nfs_page * nfs_update_request(struct nfs_open_context*,
-                                           struct page *,
-                                           unsigned int, unsigned int);
-static void nfs_mark_request_dirty(struct nfs_page *req);
-static int nfs_wait_on_write_congestion(struct address_space *, int);
-static int nfs_wait_on_requests(struct inode *, unsigned long, unsigned int);
-static long nfs_flush_mapping(struct address_space *mapping, struct writeback_control *wbc, int how);
+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;
@@ -89,41 +48,34 @@ static struct kmem_cache *nfs_wdata_cachep;
 static mempool_t *nfs_wdata_mempool;
 static mempool_t *nfs_commit_mempool;
 
-static DECLARE_WAIT_QUEUE_HEAD(nfs_write_congestion);
-
-struct nfs_write_data *nfs_commit_alloc(void)
+struct nfs_write_data *nfs_commitdata_alloc(void)
 {
        struct nfs_write_data *p = mempool_alloc(nfs_commit_mempool, GFP_NOFS);
 
        if (p) {
                memset(p, 0, sizeof(*p));
                INIT_LIST_HEAD(&p->pages);
+               p->res.seq_res.sr_slotid = NFS4_MAX_SLOT_TABLE;
        }
        return p;
 }
 
-void nfs_commit_rcu_free(struct rcu_head *head)
+void nfs_commit_free(struct nfs_write_data *p)
 {
-       struct nfs_write_data *p = container_of(head, struct nfs_write_data, task.u.tk_rcu);
        if (p && (p->pagevec != &p->page_array[0]))
                kfree(p->pagevec);
        mempool_free(p, nfs_commit_mempool);
 }
 
-void nfs_commit_free(struct nfs_write_data *wdata)
-{
-       call_rcu_bh(&wdata->task.u.tk_rcu, nfs_commit_rcu_free);
-}
-
-struct nfs_write_data *nfs_writedata_alloc(size_t len)
+struct nfs_write_data *nfs_writedata_alloc(unsigned int pagecount)
 {
-       unsigned int pagecount = (len + PAGE_SIZE - 1) >> PAGE_SHIFT;
        struct nfs_write_data *p = mempool_alloc(nfs_wdata_mempool, GFP_NOFS);
 
        if (p) {
                memset(p, 0, sizeof(*p));
                INIT_LIST_HEAD(&p->pages);
                p->npages = pagecount;
+               p->res.seq_res.sr_slotid = NFS4_MAX_SLOT_TABLE;
                if (pagecount <= ARRAY_SIZE(p->page_array))
                        p->pagevec = p->page_array;
                else {
@@ -137,22 +89,24 @@ struct nfs_write_data *nfs_writedata_alloc(size_t len)
        return p;
 }
 
-static void nfs_writedata_rcu_free(struct rcu_head *head)
+void nfs_writedata_free(struct nfs_write_data *p)
 {
-       struct nfs_write_data *p = container_of(head, struct nfs_write_data, task.u.tk_rcu);
        if (p && (p->pagevec != &p->page_array[0]))
                kfree(p->pagevec);
        mempool_free(p, nfs_wdata_mempool);
 }
 
-static void nfs_writedata_free(struct nfs_write_data *wdata)
+static void nfs_writedata_release(struct nfs_write_data *wdata)
 {
-       call_rcu_bh(&wdata->task.u.tk_rcu, nfs_writedata_rcu_free);
+       put_nfs_open_context(wdata->args.context);
+       nfs_writedata_free(wdata);
 }
 
-void nfs_writedata_release(void *wdata)
+static void nfs_context_set_write_error(struct nfs_open_context *ctx, int error)
 {
-       nfs_writedata_free(wdata);
+       ctx->error = error;
+       smp_wmb();
+       set_bit(NFS_CONTEXT_ERROR_WRITE, &ctx->flags);
 }
 
 static struct nfs_page *nfs_page_find_request_locked(struct page *page)
@@ -162,19 +116,19 @@ static struct nfs_page *nfs_page_find_request_locked(struct page *page)
        if (PagePrivate(page)) {
                req = (struct nfs_page *)page_private(page);
                if (req != NULL)
-                       atomic_inc(&req->wb_count);
+                       kref_get(&req->wb_kref);
        }
        return req;
 }
 
 static struct nfs_page *nfs_page_find_request(struct page *page)
 {
+       struct inode *inode = page->mapping->host;
        struct nfs_page *req = NULL;
-       spinlock_t *req_lock = &NFS_I(page->mapping->host)->req_lock;
 
-       spin_lock(req_lock);
+       spin_lock(&inode->i_lock);
        req = nfs_page_find_request_locked(page);
-       spin_unlock(req_lock);
+       spin_unlock(&inode->i_lock);
        return req;
 }
 
@@ -182,16 +136,28 @@ static struct nfs_page *nfs_page_find_request(struct page *page)
 static void nfs_grow_file(struct page *page, unsigned int offset, unsigned int count)
 {
        struct inode *inode = page->mapping->host;
-       loff_t end, i_size = i_size_read(inode);
-       unsigned long end_index = (i_size - 1) >> PAGE_CACHE_SHIFT;
+       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)
-               return;
+               goto out;
        end = ((loff_t)page->index << PAGE_CACHE_SHIFT) + ((loff_t)offset+count);
        if (i_size >= end)
-               return;
-       nfs_inc_stats(inode, NFSIOS_EXTENDWRITE);
+               goto out;
        i_size_write(inode, end);
+       nfs_inc_stats(inode, NFSIOS_EXTENDWRITE);
+out:
+       spin_unlock(&inode->i_lock);
+}
+
+/* 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);
 }
 
 /* We can set the PG_uptodate flag if we see that a write request
@@ -205,173 +171,190 @@ static void nfs_mark_uptodate(struct page *page, unsigned int base, unsigned int
                return;
        if (count != nfs_page_length(page))
                return;
-       if (count != PAGE_CACHE_SIZE)
-               memclear_highpage_flush(page, count, PAGE_CACHE_SIZE - count);
        SetPageUptodate(page);
 }
 
-static int nfs_writepage_setup(struct nfs_open_context *ctx, struct page *page,
-               unsigned int offset, unsigned int count)
-{
-       struct nfs_page *req;
-       int ret;
-
-       for (;;) {
-               req = nfs_update_request(ctx, page, offset, count);
-               if (!IS_ERR(req))
-                       break;
-               ret = PTR_ERR(req);
-               if (ret != -EBUSY)
-                       return ret;
-               ret = nfs_wb_page(page->mapping->host, page);
-               if (ret != 0)
-                       return ret;
-       }
-       /* Update file length */
-       nfs_grow_file(page, offset, count);
-       /* Set the PG_uptodate flag? */
-       nfs_mark_uptodate(page, offset, count);
-       nfs_unlock_request(req);
-       return 0;
-}
-
 static int wb_priority(struct writeback_control *wbc)
 {
        if (wbc->for_reclaim)
-               return FLUSH_HIGHPRI;
+               return FLUSH_HIGHPRI | FLUSH_STABLE;
        if (wbc->for_kupdate)
                return FLUSH_LOWPRI;
        return 0;
 }
 
 /*
- * Find an associated nfs write request, and prepare to flush it out
- * Returns 1 if there was no write request, or if the request was
- * already tagged by nfs_set_page_dirty.Returns 0 if the request
- * was not tagged.
- * May also return an error if the user signalled nfs_wait_on_request().
+ * NFS congestion control
  */
-static int nfs_page_mark_flush(struct page *page)
+
+int nfs_congestion_kb;
+
+#define NFS_CONGESTION_ON_THRESH       (nfs_congestion_kb >> (PAGE_SHIFT-10))
+#define NFS_CONGESTION_OFF_THRESH      \
+       (NFS_CONGESTION_ON_THRESH - (NFS_CONGESTION_ON_THRESH >> 2))
+
+static int nfs_set_page_writeback(struct page *page)
+{
+       int ret = test_set_page_writeback(page);
+
+       if (!ret) {
+               struct inode *inode = page->mapping->host;
+               struct nfs_server *nfss = NFS_SERVER(inode);
+
+               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)
 {
+       struct inode *inode = page->mapping->host;
+       struct nfs_server *nfss = NFS_SERVER(inode);
+
+       end_page_writeback(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)
+{
+       struct inode *inode = page->mapping->host;
        struct nfs_page *req;
-       spinlock_t *req_lock = &NFS_I(page->mapping->host)->req_lock;
        int ret;
 
-       spin_lock(req_lock);
-       for(;;) {
+       spin_lock(&inode->i_lock);
+       for (;;) {
                req = nfs_page_find_request_locked(page);
-               if (req == NULL) {
-                       spin_unlock(req_lock);
-                       return 1;
-               }
-               if (nfs_lock_request_dontget(req))
+               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_lock_request_dontget() will always
+                *       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).
                 */
-               spin_unlock(req_lock);
+               spin_unlock(&inode->i_lock);
                ret = nfs_wait_on_request(req);
                nfs_release_request(req);
                if (ret != 0)
-                       return ret;
-               spin_lock(req_lock);
+                       return ERR_PTR(ret);
+               spin_lock(&inode->i_lock);
        }
-       spin_unlock(req_lock);
-       if (test_and_set_bit(PG_FLUSHING, &req->wb_flags) == 0) {
-               nfs_mark_request_dirty(req);
-               set_page_writeback(page);
+       spin_unlock(&inode->i_lock);
+       return req;
+}
+
+/*
+ * Find an associated nfs write request, and prepare to flush it out
+ * May return an error if the user signalled nfs_wait_on_request().
+ */
+static int nfs_page_async_flush(struct nfs_pageio_descriptor *pgio,
+                               struct page *page)
+{
+       struct nfs_page *req;
+       int ret = 0;
+
+       req = nfs_find_and_lock_request(page);
+       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));
+
+       if (!nfs_pageio_add_request(pgio, req)) {
+               nfs_redirty_request(req);
+               ret = pgio->pg_error;
        }
-       ret = test_bit(PG_NEED_FLUSH, &req->wb_flags);
-       nfs_unlock_request(req);
+out:
        return ret;
 }
 
+static int nfs_do_writepage(struct page *page, struct writeback_control *wbc, struct nfs_pageio_descriptor *pgio)
+{
+       struct inode *inode = page->mapping->host;
+
+       nfs_inc_stats(inode, NFSIOS_VFSWRITEPAGE);
+       nfs_add_stats(inode, NFSIOS_WRITEPAGES, 1);
+
+       nfs_pageio_cond_complete(pgio, page->index);
+       return nfs_page_async_flush(pgio, page);
+}
+
 /*
  * Write an mmapped page to the server.
  */
 static int nfs_writepage_locked(struct page *page, struct writeback_control *wbc)
 {
-       struct nfs_open_context *ctx;
-       struct inode *inode = page->mapping->host;
-       unsigned offset;
+       struct nfs_pageio_descriptor pgio;
        int err;
 
-       nfs_inc_stats(inode, NFSIOS_VFSWRITEPAGE);
-       nfs_add_stats(inode, NFSIOS_WRITEPAGES, 1);
+       nfs_pageio_init_write(&pgio, page->mapping->host, wb_priority(wbc));
+       err = nfs_do_writepage(page, wbc, &pgio);
+       nfs_pageio_complete(&pgio);
+       if (err < 0)
+               return err;
+       if (pgio.pg_error < 0)
+               return pgio.pg_error;
+       return 0;
+}
 
-       err = nfs_page_mark_flush(page);
-       if (err <= 0)
-               goto out;
-       err = 0;
-       offset = nfs_page_length(page);
-       if (!offset)
-               goto out;
+int nfs_writepage(struct page *page, struct writeback_control *wbc)
+{
+       int ret;
 
-       ctx = nfs_find_open_context(inode, NULL, FMODE_WRITE);
-       if (ctx == NULL) {
-               err = -EBADF;
-               goto out;
-       }
-       err = nfs_writepage_setup(ctx, page, 0, offset);
-       put_nfs_open_context(ctx);
-       if (err != 0)
-               goto out;
-       err = nfs_page_mark_flush(page);
-       if (err > 0)
-               err = 0;
-out:
-       if (!wbc->for_writepages)
-               nfs_flush_mapping(page->mapping, wbc, wb_priority(wbc));
-       return err;
+       ret = nfs_writepage_locked(page, wbc);
+       unlock_page(page);
+       return ret;
 }
 
-int nfs_writepage(struct page *page, struct writeback_control *wbc)
+static int nfs_writepages_callback(struct page *page, struct writeback_control *wbc, void *data)
 {
-       int err;
+       int ret;
 
-       err = nfs_writepage_locked(page, wbc);
+       ret = nfs_do_writepage(page, wbc, data);
        unlock_page(page);
-       return err; 
+       return ret;
 }
 
-/*
- * Note: causes nfs_update_request() to block on the assumption
- *      that the writeback is generated due to memory pressure.
- */
 int nfs_writepages(struct address_space *mapping, struct writeback_control *wbc)
 {
-       struct backing_dev_info *bdi = mapping->backing_dev_info;
        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(bitlock, NFS_INO_FLUSHING,
+                       nfs_wait_bit_killable, TASK_KILLABLE);
+       if (err)
+               goto out_err;
+
        nfs_inc_stats(inode, NFSIOS_VFSWRITEPAGES);
 
-       err = generic_writepages(mapping, wbc);
-       if (err)
-               return err;
-       while (test_and_set_bit(BDI_write_congested, &bdi->state) != 0) {
-               if (wbc->nonblocking)
-                       return 0;
-               nfs_wait_on_write_congestion(mapping, 0);
-       }
-       err = nfs_flush_mapping(mapping, wbc, wb_priority(wbc));
+       nfs_pageio_init_write(&pgio, inode, wb_priority(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_clear_bit();
+       wake_up_bit(bitlock, NFS_INO_FLUSHING);
+
        if (err < 0)
-               goto out;
-       nfs_add_stats(inode, NFSIOS_WRITEPAGES, err);
-       if (!wbc->nonblocking && wbc->sync_mode == WB_SYNC_ALL) {
-               err = nfs_wait_on_requests(inode, 0, 0);
-               if (err < 0)
-                       goto out;
-       }
-       err = nfs_commit_inode(inode, wb_priority(wbc));
-       if (err > 0)
-               err = 0;
-out:
-       clear_bit(BDI_write_congested, &bdi->state);
-       wake_up_all(&nfs_write_congestion);
-       congestion_end(WRITE);
+               goto out_err;
+       err = pgio.pg_error;
+       if (err < 0)
+               goto out_err;
+       return 0;
+out_err:
        return err;
 }
 
@@ -383,111 +366,151 @@ static int 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;
+
+       /* Lock the request! */
+       nfs_lock_request_dontget(req);
+
+       spin_lock(&inode->i_lock);
        error = radix_tree_insert(&nfsi->nfs_page_tree, req->wb_index, req);
-       BUG_ON(error == -EEXIST);
-       if (error)
-               return error;
+       BUG_ON(error);
        if (!nfsi->npages) {
                igrab(inode);
-               nfs_begin_data_update(inode);
                if (nfs_have_delegation(inode, FMODE_WRITE))
                        nfsi->change_attr++;
        }
        SetPagePrivate(req->wb_page);
        set_page_private(req->wb_page, (unsigned long)req);
        nfsi->npages++;
-       atomic_inc(&req->wb_count);
-       return 0;
+       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;
 }
 
 /*
- * Insert a write request into an inode
+ * Remove a write request from an inode
  */
 static void nfs_inode_remove_request(struct nfs_page *req)
 {
-       struct inode *inode = req->wb_context->dentry->d_inode;
+       struct inode *inode = req->wb_context->path.dentry->d_inode;
        struct nfs_inode *nfsi = NFS_I(inode);
 
        BUG_ON (!NFS_WBACK_BUSY(req));
 
-       spin_lock(&nfsi->req_lock);
+       spin_lock(&inode->i_lock);
        set_page_private(req->wb_page, 0);
        ClearPagePrivate(req->wb_page);
        radix_tree_delete(&nfsi->nfs_page_tree, req->wb_index);
        nfsi->npages--;
        if (!nfsi->npages) {
-               spin_unlock(&nfsi->req_lock);
-               nfs_end_data_update(inode);
+               spin_unlock(&inode->i_lock);
                iput(inode);
        } else
-               spin_unlock(&nfsi->req_lock);
+               spin_unlock(&inode->i_lock);
        nfs_clear_request(req);
        nfs_release_request(req);
 }
 
+static void
+nfs_mark_request_dirty(struct nfs_page *req)
+{
+       __set_page_dirty_nobuffers(req->wb_page);
+}
+
+#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
 /*
- * Add a request to the inode's dirty list.
+ * Add a request to the inode's commit list.
  */
 static void
-nfs_mark_request_dirty(struct nfs_page *req)
+nfs_mark_request_commit(struct nfs_page *req)
 {
-       struct inode *inode = req->wb_context->dentry->d_inode;
+       struct inode *inode = req->wb_context->path.dentry->d_inode;
        struct nfs_inode *nfsi = NFS_I(inode);
 
-       spin_lock(&nfsi->req_lock);
+       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_DIRTY);
-       nfs_list_add_request(req, &nfsi->dirty);
-       nfsi->ndirty++;
-       spin_unlock(&nfsi->req_lock);
-       __mark_inode_dirty(inode, I_DIRTY_PAGES);
+                       req->wb_index,
+                       NFS_PAGE_TAG_COMMIT);
+       spin_unlock(&inode->i_lock);
+       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_redirty_request(struct nfs_page *req)
+static int
+nfs_clear_request_commit(struct nfs_page *req)
 {
-       clear_bit(PG_FLUSHING, &req->wb_flags);
-       __set_page_dirty_nobuffers(req->wb_page);
+       struct page *page = req->wb_page;
+
+       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;
+       }
+       return 0;
 }
 
-/*
- * Check if a request is dirty
- */
-static inline int
-nfs_dirty_request(struct nfs_page *req)
+static inline
+int nfs_write_need_commit(struct nfs_write_data *data)
 {
-       return test_bit(PG_FLUSHING, &req->wb_flags) == 0;
+       return data->verf.committed != NFS_FILE_SYNC;
 }
 
-#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
-/*
- * Add a request to the inode's commit list.
- */
-static void
+static inline
+int nfs_reschedule_unstable_write(struct nfs_page *req)
+{
+       if (test_and_clear_bit(PG_NEED_COMMIT, &req->wb_flags)) {
+               nfs_mark_request_commit(req);
+               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 inode *inode = req->wb_context->dentry->d_inode;
-       struct nfs_inode *nfsi = NFS_I(inode);
+}
 
-       spin_lock(&nfsi->req_lock);
-       nfs_list_add_request(req, &nfsi->commit);
-       nfsi->ncommit++;
-       spin_unlock(&nfsi->req_lock);
-       inc_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
-       __mark_inode_dirty(inode, I_DIRTY_DATASYNC);
+static inline int
+nfs_clear_request_commit(struct nfs_page *req)
+{
+       return 0;
+}
+
+static inline
+int nfs_write_need_commit(struct nfs_write_data *data)
+{
+       return 0;
+}
+
+static inline
+int nfs_reschedule_unstable_write(struct nfs_page *req)
+{
+       return 0;
 }
 #endif
 
 /*
  * Wait for a request to complete.
  *
- * Interruptible by signals only if mounted with intr flag.
+ * Interruptible by fatal signals only.
  */
-static int nfs_wait_on_requests_locked(struct inode *inode, unsigned long idx_start, unsigned int npages)
+static int nfs_wait_on_requests_locked(struct inode *inode, pgoff_t idx_start, unsigned int npages)
 {
        struct nfs_inode *nfsi = NFS_I(inode);
        struct nfs_page *req;
-       unsigned long           idx_end, next;
+       pgoff_t idx_end, next;
        unsigned int            res = 0;
        int                     error;
 
@@ -497,18 +520,18 @@ static int nfs_wait_on_requests_locked(struct inode *inode, unsigned long idx_st
                idx_end = idx_start + npages - 1;
 
        next = idx_start;
-       while (radix_tree_gang_lookup_tag(&nfsi->nfs_page_tree, (void **)&req, next, 1, NFS_PAGE_TAG_WRITEBACK)) {
+       while (radix_tree_gang_lookup_tag(&nfsi->nfs_page_tree, (void **)&req, next, 1, NFS_PAGE_TAG_LOCKED)) {
                if (req->wb_index > idx_end)
                        break;
 
                next = req->wb_index + 1;
                BUG_ON(!NFS_WBACK_BUSY(req));
 
-               atomic_inc(&req->wb_count);
-               spin_unlock(&nfsi->req_lock);
+               kref_get(&req->wb_kref);
+               spin_unlock(&inode->i_lock);
                error = nfs_wait_on_request(req);
                nfs_release_request(req);
-               spin_lock(&nfsi->req_lock);
+               spin_lock(&inode->i_lock);
                if (error < 0)
                        return error;
                res++;
@@ -516,42 +539,26 @@ static int nfs_wait_on_requests_locked(struct inode *inode, unsigned long idx_st
        return res;
 }
 
-static int nfs_wait_on_requests(struct inode *inode, unsigned long idx_start, unsigned int npages)
-{
-       struct nfs_inode *nfsi = NFS_I(inode);
-       int ret;
-
-       spin_lock(&nfsi->req_lock);
-       ret = nfs_wait_on_requests_locked(inode, idx_start, npages);
-       spin_unlock(&nfsi->req_lock);
-       return ret;
-}
-
-static void nfs_cancel_dirty_list(struct list_head *head)
-{
-       struct nfs_page *req;
-       while(!list_empty(head)) {
-               req = nfs_list_entry(head->next);
-               nfs_list_remove_request(req);
-               nfs_inode_remove_request(req);
-               nfs_clear_page_writeback(req);
-       }
-}
-
 static void nfs_cancel_commit_list(struct list_head *head)
 {
        struct nfs_page *req;
 
        while(!list_empty(head)) {
                req = nfs_list_entry(head->next);
-               dec_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
                nfs_list_remove_request(req);
+               nfs_clear_request_commit(req);
                nfs_inode_remove_request(req);
                nfs_unlock_request(req);
        }
 }
 
 #if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
+static int
+nfs_need_commit(struct nfs_inode *nfsi)
+{
+       return radix_tree_tagged(&nfsi->nfs_page_tree, NFS_PAGE_TAG_COMMIT);
+}
+
 /*
  * nfs_scan_commit - Scan an inode for commit requests
  * @inode: NFS inode to scan
@@ -563,157 +570,149 @@ static void nfs_cancel_commit_list(struct list_head *head)
  * 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, unsigned long idx_start, unsigned int npages)
+nfs_scan_commit(struct inode *inode, struct list_head *dst, pgoff_t idx_start, unsigned int npages)
 {
        struct nfs_inode *nfsi = NFS_I(inode);
-       int res = 0;
 
-       if (nfsi->ncommit != 0) {
-               res = nfs_scan_list(nfsi, &nfsi->commit, dst, idx_start, npages);
-               nfsi->ncommit -= res;
-               if ((nfsi->ncommit == 0) != list_empty(&nfsi->commit))
-                       printk(KERN_ERR "NFS: desynchronized value of nfs_i.ncommit.\n");
-       }
-       return res;
+       if (!nfs_need_commit(nfsi))
+               return 0;
+
+       return nfs_scan_list(nfsi, dst, idx_start, npages, NFS_PAGE_TAG_COMMIT);
 }
 #else
-static inline int nfs_scan_commit(struct inode *inode, struct list_head *dst, unsigned long idx_start, unsigned int npages)
+static inline int nfs_need_commit(struct nfs_inode *nfsi)
 {
        return 0;
 }
-#endif
 
-static int nfs_wait_on_write_congestion(struct address_space *mapping, int intr)
+static inline int nfs_scan_commit(struct inode *inode, struct list_head *dst, pgoff_t idx_start, unsigned int npages)
 {
-       struct backing_dev_info *bdi = mapping->backing_dev_info;
-       DEFINE_WAIT(wait);
-       int ret = 0;
-
-       might_sleep();
-
-       if (!bdi_write_congested(bdi))
-               return 0;
-
-       nfs_inc_stats(mapping->host, NFSIOS_CONGESTIONWAIT);
-
-       if (intr) {
-               struct rpc_clnt *clnt = NFS_CLIENT(mapping->host);
-               sigset_t oldset;
-
-               rpc_clnt_sigmask(clnt, &oldset);
-               prepare_to_wait(&nfs_write_congestion, &wait, TASK_INTERRUPTIBLE);
-               if (bdi_write_congested(bdi)) {
-                       if (signalled())
-                               ret = -ERESTARTSYS;
-                       else
-                               schedule();
-               }
-               rpc_clnt_sigunmask(clnt, &oldset);
-       } else {
-               prepare_to_wait(&nfs_write_congestion, &wait, TASK_UNINTERRUPTIBLE);
-               if (bdi_write_congested(bdi))
-                       schedule();
-       }
-       finish_wait(&nfs_write_congestion, &wait);
-       return ret;
+       return 0;
 }
-
+#endif
 
 /*
- * Try to update any existing write request, or create one if there is none.
- * In order to match, the request's credentials must match those of
- * the calling process.
+ * Search for an existing write request, and attempt to update
+ * it to reflect a new dirty region on a given page.
  *
- * Note: Should always be called with the Page Lock held!
+ * If the attempt fails, then the existing request is flushed out
+ * to disk.
  */
-static struct nfs_page * nfs_update_request(struct nfs_open_context* ctx,
-               struct page *page, unsigned int offset, unsigned int bytes)
+static struct nfs_page *nfs_try_to_update_request(struct inode *inode,
+               struct page *page,
+               unsigned int offset,
+               unsigned int bytes)
 {
-       struct inode *inode = page->mapping->host;
-       struct nfs_inode *nfsi = NFS_I(inode);
-       struct nfs_page         *req, *new = NULL;
-       unsigned long           rqend, end;
+       struct nfs_page *req;
+       unsigned int rqend;
+       unsigned int end;
+       int error;
+
+       if (!PagePrivate(page))
+               return NULL;
 
        end = offset + bytes;
+       spin_lock(&inode->i_lock);
 
-       if (nfs_wait_on_write_congestion(page->mapping, NFS_SERVER(inode)->flags & NFS_MOUNT_INTR))
-               return ERR_PTR(-ERESTARTSYS);
        for (;;) {
-               /* Loop over all inode entries and see if we find
-                * A request for the page we wish to update
-                */
-               spin_lock(&nfsi->req_lock);
                req = nfs_page_find_request_locked(page);
-               if (req) {
-                       if (!nfs_lock_request_dontget(req)) {
-                               int error;
-
-                               spin_unlock(&nfsi->req_lock);
-                               error = nfs_wait_on_request(req);
-                               nfs_release_request(req);
-                               if (error < 0) {
-                                       if (new)
-                                               nfs_release_request(new);
-                                       return ERR_PTR(error);
-                               }
-                               continue;
-                       }
-                       spin_unlock(&nfsi->req_lock);
-                       if (new)
-                               nfs_release_request(new);
-                       break;
-               }
+               if (req == NULL)
+                       goto out_unlock;
+
+               rqend = req->wb_offset + req->wb_bytes;
+               /*
+                * Tell the caller to flush out the request if
+                * the offsets are non-contiguous.
+                * Note: nfs_flush_incompatible() will already
+                * have flushed out requests having wrong owners.
+                */
+               if (offset > rqend
+                   || end < req->wb_offset)
+                       goto out_flushme;
 
-               if (new) {
-                       int error;
-                       nfs_lock_request_dontget(new);
-                       error = nfs_inode_add_request(inode, new);
-                       if (error) {
-                               spin_unlock(&nfsi->req_lock);
-                               nfs_unlock_request(new);
-                               return ERR_PTR(error);
-                       }
-                       spin_unlock(&nfsi->req_lock);
-                       return new;
-               }
-               spin_unlock(&nfsi->req_lock);
+               if (nfs_set_page_tag_locked(req))
+                       break;
 
-               new = nfs_create_request(ctx, inode, page, offset, bytes);
-               if (IS_ERR(new))
-                       return new;
+               /* The request is locked, so wait and then retry */
+               spin_unlock(&inode->i_lock);
+               error = nfs_wait_on_request(req);
+               nfs_release_request(req);
+               if (error != 0)
+                       goto out_err;
+               spin_lock(&inode->i_lock);
        }
 
-       /* We have a request for our page.
-        * If the creds don't match, or the
-        * page addresses don't match,
-        * tell the caller to wait on the conflicting
-        * request.
-        */
-       rqend = req->wb_offset + req->wb_bytes;
-       if (req->wb_context != ctx
-           || req->wb_page != page
-           || !nfs_dirty_request(req)
-           || offset > rqend || end < req->wb_offset) {
-               nfs_unlock_request(req);
-               return ERR_PTR(-EBUSY);
-       }
+       if (nfs_clear_request_commit(req))
+               radix_tree_tag_clear(&NFS_I(inode)->nfs_page_tree,
+                               req->wb_index, NFS_PAGE_TAG_COMMIT);
 
        /* Okay, the request matches. Update the region */
        if (offset < req->wb_offset) {
                req->wb_offset = offset;
                req->wb_pgbase = offset;
-               req->wb_bytes = rqend - req->wb_offset;
        }
-
        if (end > rqend)
                req->wb_bytes = end - req->wb_offset;
+       else
+               req->wb_bytes = rqend - req->wb_offset;
+out_unlock:
+       spin_unlock(&inode->i_lock);
+       return req;
+out_flushme:
+       spin_unlock(&inode->i_lock);
+       nfs_release_request(req);
+       error = nfs_wb_page(inode, page);
+out_err:
+       return ERR_PTR(error);
+}
 
+/*
+ * Try to update an existing write request, or create one if there is none.
+ *
+ * Note: Should always be called with the Page Lock held to prevent races
+ * if we have to add a new request. Also assumes that the caller has
+ * already called nfs_flush_incompatible() if necessary.
+ */
+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 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);
+       if (IS_ERR(req))
+               goto out;
+       error = nfs_inode_add_request(inode, req);
+       if (error != 0) {
+               nfs_release_request(req);
+               req = ERR_PTR(error);
+       }
+out:
        return req;
 }
 
+static int nfs_writepage_setup(struct nfs_open_context *ctx, struct page *page,
+               unsigned int offset, unsigned int count)
+{
+       struct nfs_page *req;
+
+       req = nfs_setup_write_request(ctx, page, offset, count);
+       if (IS_ERR(req))
+               return PTR_ERR(req);
+       /* Update file length */
+       nfs_grow_file(page, offset, count);
+       nfs_mark_uptodate(page, req->wb_pgbase, req->wb_bytes);
+       nfs_clear_page_tag_locked(req);
+       return 0;
+}
+
 int nfs_flush_incompatible(struct file *file, struct page *page)
 {
-       struct nfs_open_context *ctx = (struct nfs_open_context *)file->private_data;
+       struct nfs_open_context *ctx = nfs_file_open_context(file);
        struct nfs_page *req;
        int do_flush, status;
        /*
@@ -728,8 +727,7 @@ int nfs_flush_incompatible(struct file *file, struct page *page)
                req = nfs_page_find_request(page);
                if (req == NULL)
                        return 0;
-               do_flush = req->wb_page != page || req->wb_context != ctx
-                       || !nfs_dirty_request(req);
+               do_flush = req->wb_page != page || req->wb_context != ctx;
                nfs_release_request(req);
                if (!do_flush)
                        return 0;
@@ -739,6 +737,17 @@ int nfs_flush_incompatible(struct file *file, struct page *page)
 }
 
 /*
+ * 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)
+{
+       return PageUptodate(page) &&
+               !(NFS_I(inode)->cache_validity & (NFS_INO_REVAL_PAGECACHE|NFS_INO_INVALID_DATA));
+}
+
+/*
  * Update and possibly write a cached page of an NFS file.
  *
  * XXX: Keep an eye on generic_file_read to make sure it doesn't do bad
@@ -747,62 +756,52 @@ int nfs_flush_incompatible(struct file *file, struct page *page)
 int nfs_updatepage(struct file *file, struct page *page,
                unsigned int offset, unsigned int count)
 {
-       struct nfs_open_context *ctx = (struct nfs_open_context *)file->private_data;
+       struct nfs_open_context *ctx = nfs_file_open_context(file);
        struct inode    *inode = page->mapping->host;
        int             status = 0;
 
        nfs_inc_stats(inode, NFSIOS_VFSUPDATEPAGE);
 
-       dprintk("NFS:      nfs_updatepage(%s/%s %d@%Ld)\n",
+       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));
+               (long long)(page_offset(page) + offset));
 
        /* If we're not using byte range locks, and we know the page
-        * is entirely in cache, it may be more efficient to avoid
-        * fragmenting write requests.
+        * 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 (PageUptodate(page) && inode->i_flock == NULL && !(file->f_mode & O_SYNC)) {
+       if (nfs_write_pageuptodate(page, inode) &&
+                       inode->i_flock == NULL &&
+                       !(file->f_flags & O_SYNC)) {
                count = max(count + offset, nfs_page_length(page));
                offset = 0;
        }
 
        status = nfs_writepage_setup(ctx, page, offset, count);
-       __set_page_dirty_nobuffers(page);
+       if (status < 0)
+               nfs_set_pageerror(page);
+       else
+               __set_page_dirty_nobuffers(page);
 
-        dprintk("NFS:      nfs_updatepage returns %d (isize %Ld)\n",
+       dprintk("NFS:       nfs_updatepage returns %d (isize %lld)\n",
                        status, (long long)i_size_read(inode));
-       if (status < 0)
-               ClearPageUptodate(page);
        return status;
 }
 
 static void nfs_writepage_release(struct nfs_page *req)
 {
-       end_page_writeback(req->wb_page);
-
-#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
-       if (!PageError(req->wb_page)) {
-               if (NFS_NEED_RESCHED(req)) {
-                       nfs_redirty_request(req);
-                       goto out;
-               } else if (NFS_NEED_COMMIT(req)) {
-                       nfs_mark_request_commit(req);
-                       goto out;
-               }
-       }
-       nfs_inode_remove_request(req);
 
-out:
-       nfs_clear_commit(req);
-       nfs_clear_reschedule(req);
-#else
-       nfs_inode_remove_request(req);
-#endif
-       nfs_clear_page_writeback(req);
+       if (PageError(req->wb_page) || !nfs_reschedule_unstable_write(req)) {
+               nfs_end_page_writeback(req->wb_page);
+               nfs_inode_remove_request(req);
+       } else
+               nfs_end_page_writeback(req->wb_page);
+       nfs_clear_page_tag_locked(req);
 }
 
-static inline int flush_task_priority(int how)
+static int flush_task_priority(int how)
 {
        switch (how & (FLUSH_HIGHPRI|FLUSH_LOWPRI)) {
                case FLUSH_HIGHPRI:
@@ -816,28 +815,51 @@ static inline int flush_task_priority(int how)
 /*
  * Set up the argument/result storage required for the RPC call.
  */
-static void nfs_write_rpcsetup(struct nfs_page *req,
+static int nfs_write_rpcsetup(struct nfs_page *req,
                struct nfs_write_data *data,
                const struct rpc_call_ops *call_ops,
                unsigned int count, unsigned int offset,
                int how)
 {
-       struct inode            *inode;
-       int flags;
+       struct inode *inode = req->wb_context->path.dentry->d_inode;
+       int flags = (how & FLUSH_SYNC) ? 0 : RPC_TASK_ASYNC;
+       int priority = flush_task_priority(how);
+       struct rpc_task *task;
+       struct rpc_message msg = {
+               .rpc_argp = &data->args,
+               .rpc_resp = &data->res,
+               .rpc_cred = req->wb_context->cred,
+       };
+       struct rpc_task_setup task_setup_data = {
+               .rpc_client = NFS_CLIENT(inode),
+               .task = &data->task,
+               .rpc_message = &msg,
+               .callback_ops = call_ops,
+               .callback_data = data,
+               .workqueue = nfsiod_workqueue,
+               .flags = flags,
+               .priority = priority,
+       };
 
        /* 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->inode = inode = req->wb_context->path.dentry->d_inode;
+       data->cred = msg.rpc_cred;
 
        data->args.fh     = NFS_FH(inode);
        data->args.offset = req_offset(req) + offset;
        data->args.pgbase = req->wb_pgbase + offset;
        data->args.pages  = data->pagevec;
        data->args.count  = count;
-       data->args.context = req->wb_context;
+       data->args.context = get_nfs_open_context(req->wb_context);
+       data->args.stable  = NFS_UNSTABLE;
+       if (how & FLUSH_STABLE) {
+               data->args.stable = NFS_DATA_SYNC;
+               if (!nfs_need_commit(NFS_I(inode)))
+                       data->args.stable = NFS_FILE_SYNC;
+       }
 
        data->res.fattr   = &data->fattr;
        data->res.count   = count;
@@ -845,36 +867,39 @@ static void nfs_write_rpcsetup(struct nfs_page *req,
        nfs_fattr_init(&data->fattr);
 
        /* Set up the initial task struct.  */
-       flags = (how & FLUSH_SYNC) ? 0 : RPC_TASK_ASYNC;
-       rpc_init_task(&data->task, NFS_CLIENT(inode), flags, call_ops, data);
-       NFS_PROTO(inode)->write_setup(data, how);
+       NFS_PROTO(inode)->write_setup(data, &msg);
 
-       data->task.tk_priority = flush_task_priority(how);
-       data->task.tk_cookie = (unsigned long)inode;
-
-       dprintk("NFS: %4d initiated write call (req %s/%Ld, %u bytes @ offset %Lu)\n",
+       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),
                count,
                (unsigned long long)data->args.offset);
+
+       task = rpc_run_task(&task_setup_data);
+       if (IS_ERR(task))
+               return PTR_ERR(task);
+       rpc_put_task(task);
+       return 0;
 }
 
-static void nfs_execute_write(struct nfs_write_data *data)
+/* If a nfs_flush_* function fails, it should remove reqs from @head and
+ * call this on each, which will prepare them to be retried on next
+ * writeback using standard nfs.
+ */
+static void nfs_redirty_request(struct nfs_page *req)
 {
-       struct rpc_clnt *clnt = NFS_CLIENT(data->inode);
-       sigset_t oldset;
-
-       rpc_clnt_sigmask(clnt, &oldset);
-       rpc_execute(&data->task);
-       rpc_clnt_sigunmask(clnt, &oldset);
+       nfs_mark_request_dirty(req);
+       nfs_end_page_writeback(req->wb_page);
+       nfs_clear_page_tag_locked(req);
 }
 
 /*
  * Generate multiple small requests to write out a single
  * contiguous dirty area on one page.
  */
-static int nfs_flush_multi(struct inode *inode, struct list_head *head, int how)
+static int nfs_flush_multi(struct inode *inode, struct list_head *head, unsigned int npages, size_t count, int how)
 {
        struct nfs_page *req = nfs_list_entry(head->next);
        struct page *page = req->wb_page;
@@ -882,15 +907,16 @@ static int nfs_flush_multi(struct inode *inode, struct list_head *head, int how)
        size_t wsize = NFS_SERVER(inode)->wsize, nbytes;
        unsigned int offset;
        int requests = 0;
+       int ret = 0;
        LIST_HEAD(list);
 
        nfs_list_remove_request(req);
 
-       nbytes = req->wb_bytes;
+       nbytes = count;
        do {
                size_t len = min(nbytes, wsize);
 
-               data = nfs_writedata_alloc(len);
+               data = nfs_writedata_alloc(1);
                if (!data)
                        goto out_bad;
                list_add(&data->pages, &list);
@@ -901,27 +927,26 @@ static int nfs_flush_multi(struct inode *inode, struct list_head *head, int how)
 
        ClearPageError(page);
        offset = 0;
-       nbytes = req->wb_bytes;
+       nbytes = count;
        do {
+               int ret2;
+
                data = list_entry(list.next, struct nfs_write_data, pages);
                list_del_init(&data->pages);
 
                data->pagevec[0] = page;
 
-               if (nbytes > wsize) {
-                       nfs_write_rpcsetup(req, data, &nfs_write_partial_ops,
-                                       wsize, offset, how);
-                       offset += wsize;
-                       nbytes -= wsize;
-               } else {
-                       nfs_write_rpcsetup(req, data, &nfs_write_partial_ops,
-                                       nbytes, offset, how);
-                       nbytes = 0;
-               }
-               nfs_execute_write(data);
+               if (nbytes < wsize)
+                       wsize = nbytes;
+               ret2 = nfs_write_rpcsetup(req, data, &nfs_write_partial_ops,
+                                  wsize, offset, how);
+               if (ret == 0)
+                       ret = ret2;
+               offset += wsize;
+               nbytes -= wsize;
        } while (nbytes != 0);
 
-       return 0;
+       return ret;
 
 out_bad:
        while (!list_empty(&list)) {
@@ -930,7 +955,6 @@ out_bad:
                nfs_writedata_release(data);
        }
        nfs_redirty_request(req);
-       nfs_clear_page_writeback(req);
        return -ENOMEM;
 }
 
@@ -942,77 +966,46 @@ out_bad:
  * This is the case if nfs_updatepage detects a conflicting request
  * that has been written but not committed.
  */
-static int nfs_flush_one(struct inode *inode, struct list_head *head, int how)
+static int nfs_flush_one(struct inode *inode, struct list_head *head, unsigned int npages, size_t count, int how)
 {
        struct nfs_page         *req;
        struct page             **pages;
        struct nfs_write_data   *data;
-       unsigned int            count;
 
-       data = nfs_writedata_alloc(NFS_SERVER(inode)->wsize);
+       data = nfs_writedata_alloc(npages);
        if (!data)
                goto out_bad;
 
        pages = data->pagevec;
-       count = 0;
        while (!list_empty(head)) {
                req = nfs_list_entry(head->next);
                nfs_list_remove_request(req);
                nfs_list_add_request(req, &data->pages);
                ClearPageError(req->wb_page);
                *pages++ = req->wb_page;
-               count += req->wb_bytes;
        }
        req = nfs_list_entry(data->pages.next);
 
        /* Set up the argument struct */
-       nfs_write_rpcsetup(req, data, &nfs_write_full_ops, count, 0, how);
-
-       nfs_execute_write(data);
-       return 0;
+       return nfs_write_rpcsetup(req, data, &nfs_write_full_ops, count, 0, how);
  out_bad:
        while (!list_empty(head)) {
-               struct nfs_page *req = nfs_list_entry(head->next);
+               req = nfs_list_entry(head->next);
                nfs_list_remove_request(req);
                nfs_redirty_request(req);
-               nfs_clear_page_writeback(req);
        }
        return -ENOMEM;
 }
 
-static int nfs_flush_list(struct inode *inode, struct list_head *head, int npages, int how)
+static void nfs_pageio_init_write(struct nfs_pageio_descriptor *pgio,
+                                 struct inode *inode, int ioflags)
 {
-       LIST_HEAD(one_request);
-       int (*flush_one)(struct inode *, struct list_head *, int);
-       struct nfs_page *req;
-       int wpages = NFS_SERVER(inode)->wpages;
-       int wsize = NFS_SERVER(inode)->wsize;
-       int error;
+       size_t wsize = NFS_SERVER(inode)->wsize;
 
-       flush_one = nfs_flush_one;
        if (wsize < PAGE_CACHE_SIZE)
-               flush_one = nfs_flush_multi;
-       /* For single writes, FLUSH_STABLE is more efficient */
-       if (npages <= wpages && npages == NFS_I(inode)->npages
-                       && nfs_list_entry(head->next)->wb_bytes <= wsize)
-               how |= FLUSH_STABLE;
-
-       do {
-               nfs_coalesce_requests(head, &one_request, wpages);
-               req = nfs_list_entry(one_request.next);
-               error = flush_one(inode, &one_request, how);
-               if (error < 0)
-                       goto out_err;
-       } while (!list_empty(head));
-       return 0;
-out_err:
-       while (!list_empty(head)) {
-               req = nfs_list_entry(head->next);
-               nfs_list_remove_request(req);
-               nfs_redirty_request(req);
-               nfs_clear_page_writeback(req);
-       }
-       return error;
+               nfs_pageio_init(pgio, inode, nfs_flush_multi, wsize, ioflags);
+       else
+               nfs_pageio_init(pgio, inode, nfs_flush_one, wsize, ioflags);
 }
 
 /*
@@ -1021,46 +1014,74 @@ out_err:
 static void nfs_writeback_done_partial(struct rpc_task *task, void *calldata)
 {
        struct nfs_write_data   *data = calldata;
-       struct nfs_page         *req = data->req;
-       struct page             *page = req->wb_page;
 
-       dprintk("NFS: write (%s/%Ld %d@%Ld)",
-               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));
+       dprintk("NFS: %5u write(%s/%lld %d@%lld)",
+               task->tk_pid,
+               data->req->wb_context->path.dentry->d_inode->i_sb->s_id,
+               (long long)
+                 NFS_FILEID(data->req->wb_context->path.dentry->d_inode),
+               data->req->wb_bytes, (long long)req_offset(data->req));
 
-       if (nfs_writeback_done(task, data) != 0)
-               return;
+       nfs_writeback_done(task, data);
+}
 
-       if (task->tk_status < 0) {
-               ClearPageUptodate(page);
-               SetPageError(page);
-               req->wb_context->error = task->tk_status;
-               dprintk(", error = %d\n", task->tk_status);
-       } else {
-#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
-               if (data->verf.committed < NFS_FILE_SYNC) {
-                       if (!NFS_NEED_COMMIT(req)) {
-                               nfs_defer_commit(req);
-                               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))) {
-                               nfs_defer_reschedule(req);
-                               dprintk(" server reboot detected\n");
-                       }
-               } else
-#endif
-                       dprintk(" OK\n");
+static void nfs_writeback_release_partial(void *calldata)
+{
+       struct nfs_write_data   *data = calldata;
+       struct nfs_page         *req = data->req;
+       struct page             *page = req->wb_page;
+       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 (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);
+       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;
+       struct nfs_client *clp = (NFS_SERVER(data->inode))->nfs_client;
+
+       if (nfs4_setup_sequence(clp, &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_writedata_release,
+       .rpc_release = nfs_writeback_release_partial,
 };
 
 /*
@@ -1073,55 +1094,59 @@ static const struct rpc_call_ops nfs_write_partial_ops = {
 static void nfs_writeback_done_full(struct rpc_task *task, void *calldata)
 {
        struct nfs_write_data   *data = calldata;
-       struct nfs_page         *req;
-       struct page             *page;
 
-       if (nfs_writeback_done(task, data) != 0)
-               return;
+       nfs_writeback_done(task, data);
+}
+
+static void nfs_writeback_release_full(void *calldata)
+{
+       struct nfs_write_data   *data = calldata;
+       int status = data->task.tk_status;
 
        /* Update attributes as result of writeback. */
        while (!list_empty(&data->pages)) {
-               req = nfs_list_entry(data->pages.next);
+               struct nfs_page *req = nfs_list_entry(data->pages.next);
+               struct page *page = req->wb_page;
+
                nfs_list_remove_request(req);
-               page = req->wb_page;
 
-               dprintk("NFS: write (%s/%Ld %d@%Ld)",
-                       req->wb_context->dentry->d_inode->i_sb->s_id,
-                       (long long)NFS_FILEID(req->wb_context->dentry->d_inode),
+               dprintk("NFS: %5u write (%s/%lld %d@%lld)",
+                       data->task.tk_pid,
+                       req->wb_context->path.dentry->d_inode->i_sb->s_id,
+                       (long long)NFS_FILEID(req->wb_context->path.dentry->d_inode),
                        req->wb_bytes,
                        (long long)req_offset(req));
 
-               if (task->tk_status < 0) {
-                       ClearPageUptodate(page);
-                       SetPageError(page);
-                       req->wb_context->error = task->tk_status;
-                       end_page_writeback(page);
-                       nfs_inode_remove_request(req);
-                       dprintk(", error = %d\n", task->tk_status);
-                       goto next;
+               if (status < 0) {
+                       nfs_set_pageerror(page);
+                       nfs_context_set_write_error(req->wb_context, status);
+                       dprintk(", error = %d\n", status);
+                       goto remove_request;
                }
-               end_page_writeback(page);
 
-#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
-               if (data->args.stable != NFS_UNSTABLE || data->verf.committed == NFS_FILE_SYNC) {
-                       nfs_inode_remove_request(req);
-                       dprintk(" OK\n");
+               if (nfs_write_need_commit(data)) {
+                       memcpy(&req->wb_verf, &data->verf, sizeof(req->wb_verf));
+                       nfs_mark_request_commit(req);
+                       nfs_end_page_writeback(page);
+                       dprintk(" marked for commit\n");
                        goto next;
                }
-               memcpy(&req->wb_verf, &data->verf, sizeof(req->wb_verf));
-               nfs_mark_request_commit(req);
-               dprintk(" marked for commit\n");
-#else
+               dprintk(" OK\n");
+remove_request:
+               nfs_end_page_writeback(page);
                nfs_inode_remove_request(req);
-#endif
        next:
-               nfs_clear_page_writeback(req);
+               nfs_clear_page_tag_locked(req);
        }
+       nfs_writedata_release(calldata);
 }
 
 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_writedata_release,
+       .rpc_release = nfs_writeback_release_full,
 };
 
 
@@ -1132,9 +1157,10 @@ int nfs_writeback_done(struct rpc_task *task, struct nfs_write_data *data)
 {
        struct nfs_writeargs    *argp = &data->args;
        struct nfs_writeres     *resp = &data->res;
+       struct nfs_server       *server = NFS_SERVER(data->inode);
        int status;
 
-       dprintk("NFS: %4d nfs_writeback_done (status %d)\n",
+       dprintk("NFS: %5u nfs_writeback_done (status %d)\n",
                task->tk_pid, task->tk_status);
 
        /*
@@ -1162,9 +1188,9 @@ int nfs_writeback_done(struct rpc_task *task, struct nfs_write_data *data)
                static unsigned long    complain;
 
                if (time_before(complain, jiffies)) {
-                       dprintk("NFS: faulty NFS server %s:"
+                       dprintk("NFS:       faulty NFS server %s:"
                                " (committed = %d) != (stable = %d)\n",
-                               NFS_SERVER(data->inode)->nfs_client->cl_hostname,
+                               server->nfs_client->cl_hostname,
                                resp->verf->committed, argp->stable);
                        complain = jiffies + 300 * HZ;
                }
@@ -1190,7 +1216,7 @@ int nfs_writeback_done(struct rpc_task *task, struct nfs_write_data *data)
                                 */
                                argp->stable = NFS_FILE_SYNC;
                        }
-                       rpc_restart_call(task);
+                       nfs4_restart_rpc(task, server->nfs_client);
                        return -EAGAIN;
                }
                if (time_before(complain, jiffies)) {
@@ -1202,55 +1228,76 @@ int nfs_writeback_done(struct rpc_task *task, struct nfs_write_data *data)
                /* Can't do anything about it except throw an error. */
                task->tk_status = -EIO;
        }
+       nfs4_sequence_free_slot(server->nfs_client, &data->res.seq_res);
        return 0;
 }
 
 
 #if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
-void nfs_commit_release(void *wdata)
+void nfs_commitdata_release(void *data)
 {
+       struct nfs_write_data *wdata = data;
+
+       put_nfs_open_context(wdata->args.context);
        nfs_commit_free(wdata);
 }
 
 /*
  * Set up the argument/result storage required for the RPC call.
  */
-static void nfs_commit_rpcsetup(struct list_head *head,
+static int nfs_commit_rpcsetup(struct list_head *head,
                struct nfs_write_data *data,
                int how)
 {
-       struct nfs_page         *first;
-       struct inode            *inode;
-       int flags;
+       struct nfs_page *first = nfs_list_entry(head->next);
+       struct inode *inode = first->wb_context->path.dentry->d_inode;
+       int flags = (how & FLUSH_SYNC) ? 0 : RPC_TASK_ASYNC;
+       int priority = flush_task_priority(how);
+       struct rpc_task *task;
+       struct rpc_message msg = {
+               .rpc_argp = &data->args,
+               .rpc_resp = &data->res,
+               .rpc_cred = first->wb_context->cred,
+       };
+       struct rpc_task_setup task_setup_data = {
+               .task = &data->task,
+               .rpc_client = NFS_CLIENT(inode),
+               .rpc_message = &msg,
+               .callback_ops = &nfs_commit_ops,
+               .callback_data = data,
+               .workqueue = nfsiod_workqueue,
+               .flags = flags,
+               .priority = priority,
+       };
 
        /* Set up the RPC argument and reply structs
         * NB: take care not to mess about with data->commit et al. */
 
        list_splice_init(head, &data->pages);
-       first = nfs_list_entry(data->pages.next);
-       inode = first->wb_context->dentry->d_inode;
 
        data->inode       = inode;
-       data->cred        = first->wb_context->cred;
+       data->cred        = msg.rpc_cred;
 
        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->res.fattr   = &data->fattr;
        data->res.verf    = &data->verf;
        nfs_fattr_init(&data->fattr);
 
        /* Set up the initial task struct.  */
-       flags = (how & FLUSH_SYNC) ? 0 : RPC_TASK_ASYNC;
-       rpc_init_task(&data->task, NFS_CLIENT(inode), flags, &nfs_commit_ops, data);
-       NFS_PROTO(inode)->commit_setup(data, how);
+       NFS_PROTO(inode)->commit_setup(data, &msg);
+
+       dprintk("NFS: %5u initiated commit call\n", data->task.tk_pid);
 
-       data->task.tk_priority = flush_task_priority(how);
-       data->task.tk_cookie = (unsigned long)inode;
-       
-       dprintk("NFS: %4d initiated commit call\n", data->task.tk_pid);
+       task = rpc_run_task(&task_setup_data);
+       if (IS_ERR(task))
+               return PTR_ERR(task);
+       rpc_put_task(task);
+       return 0;
 }
 
 /*
@@ -1262,23 +1309,22 @@ nfs_commit_list(struct inode *inode, struct list_head *head, int how)
        struct nfs_write_data   *data;
        struct nfs_page         *req;
 
-       data = nfs_commit_alloc();
+       data = nfs_commitdata_alloc();
 
        if (!data)
                goto out_bad;
 
        /* Set up the argument struct */
-       nfs_commit_rpcsetup(head, data, how);
-
-       nfs_execute_write(data);
-       return 0;
+       return nfs_commit_rpcsetup(head, data, how);
  out_bad:
        while (!list_empty(head)) {
                req = nfs_list_entry(head->next);
                nfs_list_remove_request(req);
                nfs_mark_request_commit(req);
                dec_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
-               nfs_clear_page_writeback(req);
+               dec_bdi_stat(req->wb_page->mapping->backing_dev_info,
+                               BDI_RECLAIMABLE);
+               nfs_clear_page_tag_locked(req);
        }
        return -ENOMEM;
 }
@@ -1289,29 +1335,35 @@ 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_page         *req;
 
-        dprintk("NFS: %4d nfs_commit_done (status %d)\n",
+        dprintk("NFS: %5u nfs_commit_done (status %d)\n",
                                 task->tk_pid, task->tk_status);
 
        /* Call the NFS version-specific code */
        if (NFS_PROTO(data->inode)->commit_done(task, data) != 0)
                return;
+}
+
+static void nfs_commit_release(void *calldata)
+{
+       struct nfs_write_data   *data = calldata;
+       struct nfs_page         *req;
+       int status = data->task.tk_status;
 
        while (!list_empty(&data->pages)) {
                req = nfs_list_entry(data->pages.next);
                nfs_list_remove_request(req);
-               dec_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
+               nfs_clear_request_commit(req);
 
-               dprintk("NFS: commit (%s/%Ld %d@%Ld)",
-                       req->wb_context->dentry->d_inode->i_sb->s_id,
-                       (long long)NFS_FILEID(req->wb_context->dentry->d_inode),
+               dprintk("NFS:       commit (%s/%lld %d@%lld)",
+                       req->wb_context->path.dentry->d_inode->i_sb->s_id,
+                       (long long)NFS_FILEID(req->wb_context->path.dentry->d_inode),
                        req->wb_bytes,
                        (long long)req_offset(req));
-               if (task->tk_status < 0) {
-                       req->wb_context->error = task->tk_status;
+               if (status < 0) {
+                       nfs_context_set_write_error(req->wb_context, status);
                        nfs_inode_remove_request(req);
-                       dprintk(", error = %d\n", task->tk_status);
+                       dprintk(", error = %d\n", status);
                        goto next;
                }
 
@@ -1325,50 +1377,29 @@ static void nfs_commit_done(struct rpc_task *task, void *calldata)
                }
                /* We have a mismatch. Write the page again */
                dprintk(" mismatch\n");
-               nfs_redirty_request(req);
+               nfs_mark_request_dirty(req);
        next:
-               nfs_clear_page_writeback(req);
+               nfs_clear_page_tag_locked(req);
        }
+       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_done = nfs_commit_done,
        .rpc_release = nfs_commit_release,
 };
-#else
-static inline int nfs_commit_list(struct inode *inode, struct list_head *head, int how)
-{
-       return 0;
-}
-#endif
 
-static long nfs_flush_mapping(struct address_space *mapping, struct writeback_control *wbc, int how)
-{
-       struct nfs_inode *nfsi = NFS_I(mapping->host);
-       LIST_HEAD(head);
-       long res;
-
-       spin_lock(&nfsi->req_lock);
-       res = nfs_scan_dirty(mapping, wbc, &head);
-       spin_unlock(&nfsi->req_lock);
-       if (res) {
-               int error = nfs_flush_list(mapping->host, &head, res, how);
-               if (error < 0)
-                       return error;
-       }
-       return res;
-}
-
-#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
 int nfs_commit_inode(struct inode *inode, int how)
 {
-       struct nfs_inode *nfsi = NFS_I(inode);
        LIST_HEAD(head);
        int res;
 
-       spin_lock(&nfsi->req_lock);
+       spin_lock(&inode->i_lock);
        res = nfs_scan_commit(inode, &head, 0, 0);
-       spin_unlock(&nfsi->req_lock);
+       spin_unlock(&inode->i_lock);
        if (res) {
                int error = nfs_commit_list(inode, &head, how);
                if (error < 0)
@@ -1376,13 +1407,17 @@ int nfs_commit_inode(struct inode *inode, int how)
        }
        return res;
 }
+#else
+static inline int nfs_commit_list(struct inode *inode, struct list_head *head, int how)
+{
+       return 0;
+}
 #endif
 
 long nfs_sync_mapping_wait(struct address_space *mapping, struct writeback_control *wbc, int how)
 {
        struct inode *inode = mapping->host;
-       struct nfs_inode *nfsi = NFS_I(inode);
-       unsigned long idx_start, idx_end;
+       pgoff_t idx_start, idx_end;
        unsigned int npages = 0;
        LIST_HEAD(head);
        int nocommit = how & FLUSH_NOCOMMIT;
@@ -1395,109 +1430,130 @@ long nfs_sync_mapping_wait(struct address_space *mapping, struct writeback_contr
                idx_start = wbc->range_start >> PAGE_CACHE_SHIFT;
                idx_end = wbc->range_end >> PAGE_CACHE_SHIFT;
                if (idx_end > idx_start) {
-                       unsigned long l_npages = 1 + idx_end - idx_start;
+                       pgoff_t l_npages = 1 + idx_end - idx_start;
                        npages = l_npages;
                        if (sizeof(npages) != sizeof(l_npages) &&
-                                       (unsigned long)npages != l_npages)
+                                       (pgoff_t)npages != l_npages)
                                npages = 0;
                }
        }
        how &= ~FLUSH_NOCOMMIT;
-       spin_lock(&nfsi->req_lock);
+       spin_lock(&inode->i_lock);
        do {
-               wbc->pages_skipped = 0;
                ret = nfs_wait_on_requests_locked(inode, idx_start, npages);
                if (ret != 0)
                        continue;
-               pages = nfs_scan_dirty(mapping, wbc, &head);
-               if (pages != 0) {
-                       spin_unlock(&nfsi->req_lock);
-                       if (how & FLUSH_INVALIDATE) {
-                               nfs_cancel_dirty_list(&head);
-                               ret = pages;
-                       } else
-                               ret = nfs_flush_list(inode, &head, pages, how);
-                       spin_lock(&nfsi->req_lock);
-                       continue;
-               }
-               if (wbc->pages_skipped != 0)
-                       continue;
                if (nocommit)
                        break;
                pages = nfs_scan_commit(inode, &head, idx_start, npages);
-               if (pages == 0) {
-                       if (wbc->pages_skipped != 0)
-                               continue;
+               if (pages == 0)
                        break;
-               }
                if (how & FLUSH_INVALIDATE) {
-                       spin_unlock(&nfsi->req_lock);
+                       spin_unlock(&inode->i_lock);
                        nfs_cancel_commit_list(&head);
                        ret = pages;
-                       spin_lock(&nfsi->req_lock);
+                       spin_lock(&inode->i_lock);
                        continue;
                }
                pages += nfs_scan_commit(inode, &head, 0, 0);
-               spin_unlock(&nfsi->req_lock);
+               spin_unlock(&inode->i_lock);
                ret = nfs_commit_list(inode, &head, how);
-               spin_lock(&nfsi->req_lock);
+               spin_lock(&inode->i_lock);
+
        } while (ret >= 0);
-       spin_unlock(&nfsi->req_lock);
+       spin_unlock(&inode->i_lock);
        return ret;
 }
 
-/*
- * flush the inode to disk.
- */
-int nfs_wb_all(struct inode *inode)
+static int __nfs_write_mapping(struct address_space *mapping, struct writeback_control *wbc, int how)
 {
-       struct address_space *mapping = inode->i_mapping;
-       struct writeback_control wbc = {
-               .bdi = mapping->backing_dev_info,
-               .sync_mode = WB_SYNC_ALL,
-               .nr_to_write = LONG_MAX,
-               .for_writepages = 1,
-               .range_cyclic = 1,
-       };
        int ret;
 
-       ret = generic_writepages(mapping, &wbc);
+       ret = nfs_writepages(mapping, wbc);
        if (ret < 0)
                goto out;
-       ret = nfs_sync_mapping_wait(mapping, &wbc, 0);
-       if (ret >= 0)
-               return 0;
+       ret = nfs_sync_mapping_wait(mapping, wbc, how);
+       if (ret < 0)
+               goto out;
+       return 0;
 out:
        __mark_inode_dirty(mapping->host, I_DIRTY_PAGES);
        return ret;
 }
 
-int nfs_sync_mapping_range(struct address_space *mapping, loff_t range_start, loff_t range_end, int how)
+/* Two pass sync: first using WB_SYNC_NONE, then WB_SYNC_ALL */
+static int nfs_write_mapping(struct address_space *mapping, int how)
 {
        struct writeback_control wbc = {
                .bdi = mapping->backing_dev_info,
                .sync_mode = WB_SYNC_ALL,
                .nr_to_write = LONG_MAX,
+               .range_start = 0,
+               .range_end = LLONG_MAX,
+       };
+
+       return __nfs_write_mapping(mapping, &wbc, how);
+}
+
+/*
+ * flush the inode to disk.
+ */
+int nfs_wb_all(struct inode *inode)
+{
+       return nfs_write_mapping(inode->i_mapping, 0);
+}
+
+int nfs_wb_nocommit(struct inode *inode)
+{
+       return nfs_write_mapping(inode->i_mapping, FLUSH_NOCOMMIT);
+}
+
+int nfs_wb_page_cancel(struct inode *inode, struct page *page)
+{
+       struct nfs_page *req;
+       loff_t range_start = page_offset(page);
+       loff_t range_end = range_start + (loff_t)(PAGE_CACHE_SIZE - 1);
+       struct writeback_control wbc = {
+               .bdi = page->mapping->backing_dev_info,
+               .sync_mode = WB_SYNC_ALL,
+               .nr_to_write = LONG_MAX,
                .range_start = range_start,
                .range_end = range_end,
-               .for_writepages = 1,
        };
-       int ret;
+       int ret = 0;
 
-       if (!(how & FLUSH_NOWRITEPAGE)) {
-               ret = generic_writepages(mapping, &wbc);
+       BUG_ON(!PageLocked(page));
+       for (;;) {
+               req = nfs_page_find_request(page);
+               if (req == NULL)
+                       goto out;
+               if (test_bit(PG_CLEAN, &req->wb_flags)) {
+                       nfs_release_request(req);
+                       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);
                if (ret < 0)
                        goto out;
        }
-       ret = nfs_sync_mapping_wait(mapping, &wbc, how);
-       if (ret >= 0)
+       if (!PagePrivate(page))
                return 0;
+       ret = nfs_sync_mapping_wait(page->mapping, &wbc, FLUSH_INVALIDATE);
 out:
-       __mark_inode_dirty(mapping->host, I_DIRTY_PAGES);
        return ret;
 }
 
-int nfs_wb_page_priority(struct inode *inode, struct page *page, int how)
+static int nfs_wb_page_priority(struct inode *inode, struct page *page,
+                               int how)
 {
        loff_t range_start = page_offset(page);
        loff_t range_end = range_start + (loff_t)(PAGE_CACHE_SIZE - 1);
@@ -1510,16 +1566,19 @@ int nfs_wb_page_priority(struct inode *inode, struct page *page, int how)
        };
        int ret;
 
-       BUG_ON(!PageLocked(page));
-       if (!(how & FLUSH_NOWRITEPAGE) && clear_page_dirty_for_io(page)) {
-               ret = nfs_writepage_locked(page, &wbc);
+       do {
+               if (clear_page_dirty_for_io(page)) {
+                       ret = nfs_writepage_locked(page, &wbc);
+                       if (ret < 0)
+                               goto out_error;
+               } else if (!PagePrivate(page))
+                       break;
+               ret = nfs_sync_mapping_wait(page->mapping, &wbc, how);
                if (ret < 0)
-                       goto out;
-       }
-       ret = nfs_sync_mapping_wait(page->mapping, &wbc, how);
-       if (ret >= 0)
-               return 0;
-out:
+                       goto out_error;
+       } while (PagePrivate(page));
+       return 0;
+out_error:
        __mark_inode_dirty(inode, I_DIRTY_PAGES);
        return ret;
 }
@@ -1532,26 +1591,47 @@ int nfs_wb_page(struct inode *inode, struct page* page)
        return nfs_wb_page_priority(inode, page, FLUSH_STABLE);
 }
 
-int nfs_set_page_dirty(struct page *page)
+#ifdef CONFIG_MIGRATION
+int nfs_migrate_page(struct address_space *mapping, struct page *newpage,
+               struct page *page)
 {
        struct nfs_page *req;
+       int ret;
 
-       req = nfs_page_find_request(page);
-       if (req != NULL) {
-               /* Mark any existing write requests for flushing */
-               set_bit(PG_NEED_FLUSH, &req->wb_flags);
-               nfs_release_request(req);
-       }
-       return __set_page_dirty_nobuffers(page);
-}
+       if (PageFsCache(page))
+               nfs_fscache_release_page(page, GFP_KERNEL);
 
+       req = nfs_find_and_lock_request(page);
+       ret = PTR_ERR(req);
+       if (IS_ERR(req))
+               goto out;
+
+       ret = migrate_page(mapping, newpage, page);
+       if (!req)
+               goto out;
+       if (ret)
+               goto out_unlock;
+       page_cache_get(newpage);
+       req->wb_page = newpage;
+       SetPagePrivate(newpage);
+       set_page_private(newpage, page_private(page));
+       ClearPagePrivate(page);
+       set_page_private(page, 0);
+       page_cache_release(page);
+out_unlock:
+       nfs_clear_page_tag_locked(req);
+       nfs_release_request(req);
+out:
+       return ret;
+}
+#endif
 
 int __init nfs_init_writepagecache(void)
 {
        nfs_wdata_cachep = kmem_cache_create("nfs_write_data",
                                             sizeof(struct nfs_write_data),
                                             0, SLAB_HWCACHE_ALIGN,
-                                            NULL, NULL);
+                                            NULL);
        if (nfs_wdata_cachep == NULL)
                return -ENOMEM;
 
@@ -1565,6 +1645,26 @@ int __init nfs_init_writepagecache(void)
        if (nfs_commit_mempool == NULL)
                return -ENOMEM;
 
+       /*
+        * NFS congestion size, scale with available memory.
+        *
+        *  64MB:    8192k
+        * 128MB:   11585k
+        * 256MB:   16384k
+        * 512MB:   23170k
+        *   1GB:   32768k
+        *   2GB:   46340k
+        *   4GB:   65536k
+        *   8GB:   92681k
+        *  16GB:  131072k
+        *
+        * This allows larger machines to have larger/more transfers.
+        * Limit the default to 256M
+        */
+       nfs_congestion_kb = (16*int_sqrt(totalram_pages)) << (PAGE_SHIFT-10);
+       if (nfs_congestion_kb > 256*1024)
+               nfs_congestion_kb = 256*1024;
+
        return 0;
 }