/*
* 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>
*/
#include <linux/pagemap.h>
#include <linux/file.h>
#include <linux/writeback.h>
+#include <linux/swap.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"
#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 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 int nfs_wb_page_priority(struct inode *inode, struct page *page, 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;
-static kmem_cache_t *nfs_wdata_cachep;
+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, SLAB_NOFS);
+ 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)
+struct nfs_write_data *nfs_writedata_alloc(unsigned int pagecount)
{
- call_rcu_bh(&wdata->task.u.tk_rcu, nfs_commit_rcu_free);
-}
-
-struct nfs_write_data *nfs_writedata_alloc(size_t len)
-{
- unsigned int pagecount = (len + PAGE_SIZE - 1) >> PAGE_SHIFT;
- struct nfs_write_data *p = mempool_alloc(nfs_wdata_mempool, SLAB_NOFS);
+ 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 {
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)
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;
}
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
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)
+static int wb_priority(struct writeback_control *wbc)
{
- struct nfs_page *req;
+ if (wbc->for_reclaim)
+ return FLUSH_HIGHPRI | FLUSH_STABLE;
+ if (wbc->for_kupdate)
+ return FLUSH_LOWPRI;
+ return 0;
+}
+
+/*
+ * NFS congestion control
+ */
+
+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);
+}
+
+/*
+ * 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 inode *inode = page->mapping->host;
+ struct nfs_page *req;
int ret;
- for (;;) {
- req = nfs_update_request(ctx, page, offset, count);
- if (!IS_ERR(req))
+ spin_lock(&inode->i_lock);
+ for(;;) {
+ req = nfs_page_find_request_locked(page);
+ if (req == NULL) {
+ spin_unlock(&inode->i_lock);
+ return 0;
+ }
+ if (nfs_set_page_tag_locked(req))
break;
- ret = PTR_ERR(req);
- if (ret != -EBUSY)
- return ret;
- ret = nfs_wb_page(page->mapping->host, page);
+ /* 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).
+ */
+ spin_unlock(&inode->i_lock);
+ ret = nfs_wait_on_request(req);
+ nfs_release_request(req);
if (ret != 0)
return ret;
+ spin_lock(&inode->i_lock);
+ }
+ if (test_bit(PG_CLEAN, &req->wb_flags)) {
+ spin_unlock(&inode->i_lock);
+ BUG();
+ }
+ if (nfs_set_page_writeback(page) != 0) {
+ spin_unlock(&inode->i_lock);
+ BUG();
+ }
+ spin_unlock(&inode->i_lock);
+ if (!nfs_pageio_add_request(pgio, req)) {
+ nfs_redirty_request(req);
+ return pgio->pg_error;
}
- /* 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)
+static int nfs_do_writepage(struct page *page, struct writeback_control *wbc, struct nfs_pageio_descriptor *pgio)
{
- if (wbc->for_reclaim)
- return FLUSH_HIGHPRI;
- if (wbc->for_kupdate)
- return FLUSH_LOWPRI;
- return 0;
+ 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);
}
/*
*/
static int nfs_writepage_locked(struct page *page, struct writeback_control *wbc)
{
- struct nfs_open_context *ctx;
- struct inode *inode = page->mapping->host;
- struct nfs_page *req;
- unsigned offset;
- int err = 0;
-
- nfs_inc_stats(inode, NFSIOS_VFSWRITEPAGE);
- nfs_add_stats(inode, NFSIOS_WRITEPAGES, 1);
-
- req = nfs_page_find_request(page);
- if (req != NULL) {
- int flushme = test_bit(PG_NEED_FLUSH, &req->wb_flags);
- nfs_release_request(req);
- if (!flushme)
- goto out;
- /* Ensure we've flushed out the invalid write */
- nfs_wb_page_priority(inode, page, wb_priority(wbc) | FLUSH_STABLE | FLUSH_NOWRITEPAGE);
- }
+ struct nfs_pageio_descriptor pgio;
+ int err;
- offset = nfs_page_length(page);
- if (!offset)
- goto out;
+ 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;
+}
- 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);
+int nfs_writepage(struct page *page, struct writeback_control *wbc)
+{
+ int ret;
-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;
}
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);
}
-/*
- * Add a request to the inode's dirty list.
- */
static void
nfs_mark_request_dirty(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);
- 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);
- inc_zone_page_state(req->wb_page, NR_FILE_DIRTY);
- mark_inode_dirty(inode);
-}
-
-/*
- * Check if a request is dirty
- */
-static inline int
-nfs_dirty_request(struct nfs_page *req)
-{
- struct nfs_inode *nfsi = NFS_I(req->wb_context->dentry->d_inode);
- return !list_empty(&req->wb_list) && req->wb_list_head == &nfsi->dirty;
+ __set_page_dirty_nobuffers(req->wb_page);
}
#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
static void
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);
- nfs_list_add_request(req, &nfsi->commit);
- nfsi->ncommit++;
- spin_unlock(&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_COMMIT);
+ spin_unlock(&inode->i_lock);
inc_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
- mark_inode_dirty(inode);
+ inc_bdi_stat(req->wb_page->mapping->backing_dev_info, BDI_RECLAIMABLE);
+ __mark_inode_dirty(inode, I_DIRTY_DATASYNC);
+}
+
+static int
+nfs_clear_request_commit(struct nfs_page *req)
+{
+ 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;
+}
+
+static inline
+int nfs_write_need_commit(struct nfs_write_data *data)
+{
+ return data->verf.committed != NFS_FILE_SYNC;
+}
+
+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)
+{
+}
+
+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;
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++;
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
* 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);
- nfs_mark_request_dirty(new);
- 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;
/*
req = nfs_page_find_request(page);
if (req == NULL)
return 0;
- do_flush = req->wb_page != page || req->wb_context != ctx
- || test_bit(PG_NEED_FLUSH, &req->wb_flags);
+ do_flush = req->wb_page != page || req->wb_context != ctx;
nfs_release_request(req);
if (!do_flush)
return 0;
}
/*
+ * 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
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",
- file->f_dentry->d_parent->d_name.name,
- file->f_dentry->d_name.name, count,
- (long long)(page_offset(page) +offset));
+ 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));
/* 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);
+ 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_mark_request_dirty(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:
/*
* 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;
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;
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);
atomic_set(&req->wb_complete, requests);
ClearPageError(page);
- set_page_writeback(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)) {
list_del(&data->pages);
nfs_writedata_release(data);
}
- nfs_mark_request_dirty(req);
- nfs_clear_page_writeback(req);
+ nfs_redirty_request(req);
return -ENOMEM;
}
* 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);
- set_page_writeback(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_mark_request_dirty(req);
- nfs_clear_page_writeback(req);
+ nfs_redirty_request(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_mark_request_dirty(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);
}
/*
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,
};
/*
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,
};
{
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);
/*
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;
}
*/
argp->stable = NFS_FILE_SYNC;
}
- rpc_restart_call(task);
+ nfs4_restart_rpc(task, server->nfs_client);
return -EAGAIN;
}
if (time_before(complain, jiffies)) {
/* 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);
- 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);
+ dprintk("NFS: %5u 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;
}
/*
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;
}
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;
}
dprintk(" mismatch\n");
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)
}
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;
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)
+{
+ int ret;
+
+ ret = nfs_writepages(mapping, wbc);
+ if (ret < 0)
+ goto out;
+ 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;
+}
+
+/* Two pass sync: first using WB_SYNC_NONE, then WB_SYNC_ALL */
+static int nfs_write_mapping(struct address_space *mapping, 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,
- .range_cyclic = 1,
+ .range_start = 0,
+ .range_end = LLONG_MAX,
+ .for_writepages = 1,
};
- int ret;
- ret = nfs_sync_mapping_wait(mapping, &wbc, 0);
- if (ret >= 0)
- return 0;
- return ret;
+ 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_sync_mapping_range(struct address_space *mapping, loff_t range_start, loff_t range_end, int how)
+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 = mapping->backing_dev_info,
+ .bdi = page->mapping->backing_dev_info,
.sync_mode = WB_SYNC_ALL,
.nr_to_write = LONG_MAX,
.range_start = range_start,
.range_end = range_end,
};
- int ret;
+ int ret = 0;
- ret = nfs_sync_mapping_wait(mapping, &wbc, how);
- if (ret >= 0)
+ 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;
+ }
+ if (!PagePrivate(page))
return 0;
+ ret = nfs_sync_mapping_wait(page->mapping, &wbc, FLUSH_INVALIDATE);
+out:
return ret;
}
-static 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);
};
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;
}
return nfs_wb_page_priority(inode, page, FLUSH_STABLE);
}
-int nfs_set_page_dirty(struct page *page)
-{
- struct nfs_page *req;
-
- 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);
-}
-
-
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;
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;
}
git://ftp.safe.ca / safe / jmp / linux-2.6 / blobdiff