#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 "delegation.h"
#include "internal.h"
#include "iostat.h"
+#include "nfs4_fs.h"
+#include "fscache.h"
+#include "pnfs.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_pageio_init_write(struct nfs_pageio_descriptor *desc,
struct inode *inode, int ioflags);
static void nfs_redirty_request(struct nfs_page *req);
static mempool_t *nfs_wdata_mempool;
static mempool_t *nfs_commit_mempool;
-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);
}
return p;
}
+EXPORT_SYMBOL_GPL(nfs_commitdata_alloc);
void nfs_commit_free(struct nfs_write_data *p)
{
kfree(p->pagevec);
mempool_free(p, nfs_commit_mempool);
}
+EXPORT_SYMBOL_GPL(nfs_commit_free);
struct nfs_write_data *nfs_writedata_alloc(unsigned int pagecount)
{
return p;
}
-static void nfs_writedata_free(struct nfs_write_data *p)
+void nfs_writedata_free(struct nfs_write_data *p)
{
if (p && (p->pagevec != &p->page_array[0]))
kfree(p->pagevec);
mempool_free(p, nfs_wdata_mempool);
}
-void nfs_writedata_release(void *data)
+static void nfs_writedata_release(struct nfs_write_data *wdata)
{
- struct nfs_write_data *wdata = data;
-
+ put_lseg(wdata->lseg);
put_nfs_open_context(wdata->args.context);
nfs_writedata_free(wdata);
}
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);
- pgoff_t 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 */
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);
- nfs_clear_page_tag_locked(req);
- return 0;
-}
-
static int wb_priority(struct writeback_control *wbc)
{
if (wbc->for_reclaim)
return FLUSH_HIGHPRI | FLUSH_STABLE;
- if (wbc->for_kupdate)
- return FLUSH_LOWPRI;
- return 0;
+ if (wbc->for_kupdate || wbc->for_background)
+ return FLUSH_LOWPRI | FLUSH_COND_STABLE;
+ return FLUSH_COND_STABLE;
}
/*
struct inode *inode = page->mapping->host;
struct nfs_server *nfss = NFS_SERVER(inode);
+ page_cache_get(page);
if (atomic_long_inc_return(&nfss->writeback) >
- NFS_CONGESTION_ON_THRESH)
- set_bdi_congested(&nfss->backing_dev_info, WRITE);
+ NFS_CONGESTION_ON_THRESH) {
+ set_bdi_congested(&nfss->backing_dev_info,
+ BLK_RW_ASYNC);
+ }
}
return ret;
}
struct nfs_server *nfss = NFS_SERVER(inode);
end_page_writeback(page);
+ page_cache_release(page);
if (atomic_long_dec_return(&nfss->writeback) < NFS_CONGESTION_OFF_THRESH)
- clear_bdi_congested(&nfss->backing_dev_info, WRITE);
+ 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)
+static struct nfs_page *nfs_find_and_lock_request(struct page *page, bool nonblock)
{
struct inode *inode = page->mapping->host;
struct nfs_page *req;
int ret;
spin_lock(&inode->i_lock);
- for(;;) {
+ for (;;) {
req = nfs_page_find_request_locked(page);
- if (req == NULL) {
- spin_unlock(&inode->i_lock);
- return 0;
- }
+ 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,
* request as dirty (in which case we don't care).
*/
spin_unlock(&inode->i_lock);
- ret = nfs_wait_on_request(req);
+ if (!nonblock)
+ ret = nfs_wait_on_request(req);
+ else
+ ret = -EAGAIN;
nfs_release_request(req);
if (ret != 0)
- return ret;
+ return ERR_PTR(ret);
spin_lock(&inode->i_lock);
}
- if (test_bit(PG_NEED_COMMIT, &req->wb_flags)) {
- /* This request is marked for commit */
- spin_unlock(&inode->i_lock);
- nfs_clear_page_tag_locked(req);
- nfs_pageio_complete(pgio);
- return 0;
- }
- if (nfs_set_page_writeback(page) != 0) {
- spin_unlock(&inode->i_lock);
- BUG();
- }
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, bool nonblock)
+{
+ struct nfs_page *req;
+ int ret = 0;
+
+ req = nfs_find_and_lock_request(page, nonblock);
+ if (!req)
+ goto out;
+ ret = PTR_ERR(req);
+ if (IS_ERR(req))
+ goto out;
+
+ ret = nfs_set_page_writeback(page);
+ BUG_ON(ret != 0);
+ BUG_ON(test_bit(PG_CLEAN, &req->wb_flags));
+
if (!nfs_pageio_add_request(pgio, req)) {
nfs_redirty_request(req);
- nfs_end_page_writeback(page);
- nfs_clear_page_tag_locked(req);
- return pgio->pg_error;
+ ret = pgio->pg_error;
}
- return 0;
+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;
+ int ret;
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);
+ ret = nfs_page_async_flush(pgio, page, wbc->sync_mode == WB_SYNC_NONE);
+ if (ret == -EAGAIN) {
+ redirty_page_for_writepage(wbc, page);
+ ret = 0;
+ }
+ return ret;
}
/*
int nfs_writepages(struct address_space *mapping, struct writeback_control *wbc)
{
struct inode *inode = mapping->host;
+ unsigned long *bitlock = &NFS_I(inode)->flags;
struct nfs_pageio_descriptor pgio;
int err;
+ /* Stop dirtying of new pages while we sync */
+ err = wait_on_bit_lock(bitlock, NFS_INO_FLUSHING,
+ nfs_wait_bit_killable, TASK_KILLABLE);
+ if (err)
+ goto out_err;
+
nfs_inc_stats(inode, NFSIOS_VFSWRITEPAGES);
nfs_pageio_init_write(&pgio, inode, wb_priority(wbc));
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)
- return err;
- if (pgio.pg_error < 0)
- return pgio.pg_error;
+ goto out_err;
+ err = pgio.pg_error;
+ if (err < 0)
+ goto out_err;
return 0;
+out_err:
+ return err;
}
/*
* Insert a write request into an inode
*/
-static void nfs_inode_add_request(struct inode *inode, struct nfs_page *req)
+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);
if (!nfsi->npages) {
if (nfs_have_delegation(inode, FMODE_WRITE))
nfsi->change_attr++;
}
+ set_bit(PG_MAPPED, &req->wb_flags);
SetPagePrivate(req->wb_page);
set_page_private(req->wb_page, (unsigned long)req);
nfsi->npages++;
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;
}
/*
spin_lock(&inode->i_lock);
set_page_private(req->wb_page, 0);
ClearPagePrivate(req->wb_page);
+ clear_bit(PG_MAPPED, &req->wb_flags);
radix_tree_delete(&nfsi->nfs_page_tree, req->wb_index);
nfsi->npages--;
if (!nfsi->npages) {
iput(inode);
} else
spin_unlock(&inode->i_lock);
- nfs_clear_request(req);
nfs_release_request(req);
}
static void
-nfs_redirty_request(struct nfs_page *req)
+nfs_mark_request_dirty(struct nfs_page *req)
{
__set_page_dirty_nobuffers(req->wb_page);
-}
-
-/*
- * Check if a request is dirty
- */
-static inline int
-nfs_dirty_request(struct nfs_page *req)
-{
- struct page *page = req->wb_page;
-
- if (page == NULL || test_bit(PG_NEED_COMMIT, &req->wb_flags))
- return 0;
- return !PageWriteback(req->wb_page);
+ __mark_inode_dirty(req->wb_page->mapping->host, I_DIRTY_DATASYNC);
}
#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
* Add a request to the inode's commit list.
*/
static void
-nfs_mark_request_commit(struct nfs_page *req)
+nfs_mark_request_commit(struct nfs_page *req, struct pnfs_layout_segment *lseg)
{
struct inode *inode = req->wb_context->path.dentry->d_inode;
struct nfs_inode *nfsi = NFS_I(inode);
spin_lock(&inode->i_lock);
- nfsi->ncommit++;
- set_bit(PG_NEED_COMMIT, &(req)->wb_flags);
+ set_bit(PG_CLEAN, &(req)->wb_flags);
radix_tree_tag_set(&nfsi->nfs_page_tree,
req->wb_index,
NFS_PAGE_TAG_COMMIT);
+ nfsi->ncommit++;
spin_unlock(&inode->i_lock);
+ pnfs_mark_request_commit(req, lseg);
inc_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
inc_bdi_stat(req->wb_page->mapping->backing_dev_info, BDI_RECLAIMABLE);
__mark_inode_dirty(inode, I_DIRTY_DATASYNC);
}
+static 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;
+ if (data->verf.committed == NFS_DATA_SYNC)
+ return data->lseg == NULL;
+ else
+ return data->verf.committed != NFS_FILE_SYNC;
}
static inline
-int nfs_reschedule_unstable_write(struct nfs_page *req)
+int nfs_reschedule_unstable_write(struct nfs_page *req,
+ struct nfs_write_data *data)
{
- if (test_bit(PG_NEED_COMMIT, &req->wb_flags)) {
- nfs_mark_request_commit(req);
+ if (test_and_clear_bit(PG_NEED_COMMIT, &req->wb_flags)) {
+ nfs_mark_request_commit(req, data->lseg);
return 1;
}
if (test_and_clear_bit(PG_NEED_RESCHED, &req->wb_flags)) {
- nfs_redirty_request(req);
+ nfs_mark_request_dirty(req);
return 1;
}
return 0;
}
#else
static inline void
-nfs_mark_request_commit(struct nfs_page *req)
+nfs_mark_request_commit(struct nfs_page *req, struct pnfs_layout_segment *lseg)
{
}
-static inline
-int nfs_write_need_commit(struct nfs_write_data *data)
+static inline int
+nfs_clear_request_commit(struct nfs_page *req)
{
return 0;
}
static inline
-int nfs_reschedule_unstable_write(struct nfs_page *req)
+int nfs_write_need_commit(struct nfs_write_data *data)
{
return 0;
}
-#endif
-/*
- * Wait for a request to complete.
- *
- * Interruptible by fatal signals only.
- */
-static int nfs_wait_on_requests_locked(struct inode *inode, pgoff_t idx_start, unsigned int npages)
+static inline
+int nfs_reschedule_unstable_write(struct nfs_page *req,
+ struct nfs_write_data *data)
{
- struct nfs_inode *nfsi = NFS_I(inode);
- struct nfs_page *req;
- pgoff_t idx_end, next;
- unsigned int res = 0;
- int error;
-
- if (npages == 0)
- idx_end = ~0;
- else
- 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_LOCKED)) {
- if (req->wb_index > idx_end)
- break;
-
- next = req->wb_index + 1;
- BUG_ON(!NFS_WBACK_BUSY(req));
-
- kref_get(&req->wb_kref);
- spin_unlock(&inode->i_lock);
- error = nfs_wait_on_request(req);
- nfs_release_request(req);
- spin_lock(&inode->i_lock);
- if (error < 0)
- return error;
- res++;
- }
- return res;
+ return 0;
}
+#endif
-static void nfs_cancel_commit_list(struct list_head *head)
+#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
+static int
+nfs_need_commit(struct nfs_inode *nfsi)
{
- struct nfs_page *req;
-
- while(!list_empty(head)) {
- req = nfs_list_entry(head->next);
- dec_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
- dec_bdi_stat(req->wb_page->mapping->backing_dev_info,
- BDI_RECLAIMABLE);
- nfs_list_remove_request(req);
- clear_bit(PG_NEED_COMMIT, &(req)->wb_flags);
- nfs_inode_remove_request(req);
- nfs_unlock_request(req);
- }
+ return radix_tree_tagged(&nfsi->nfs_page_tree, NFS_PAGE_TAG_COMMIT);
}
-#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
/*
* nfs_scan_commit - Scan an inode for commit requests
* @inode: NFS inode to scan
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;
+ int ret;
- if (nfsi->ncommit != 0) {
- res = nfs_scan_list(nfsi, dst, idx_start, npages,
- NFS_PAGE_TAG_COMMIT);
- nfsi->ncommit -= res;
- }
- return res;
+ if (!nfs_need_commit(nfsi))
+ return 0;
+
+ ret = nfs_scan_list(nfsi, dst, idx_start, npages, NFS_PAGE_TAG_COMMIT);
+ if (ret > 0)
+ nfsi->ncommit -= ret;
+ if (nfs_need_commit(NFS_I(inode)))
+ __mark_inode_dirty(inode, I_DIRTY_DATASYNC);
+ return ret;
}
#else
+static inline int nfs_need_commit(struct nfs_inode *nfsi)
+{
+ return 0;
+}
+
static inline int nfs_scan_commit(struct inode *inode, struct list_head *dst, pgoff_t idx_start, unsigned int npages)
{
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 address_space *mapping = page->mapping;
- struct inode *inode = mapping->host;
- struct nfs_page *req, *new = NULL;
- pgoff_t 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);
for (;;) {
- /* Loop over all inode entries and see if we find
- * A request for the page we wish to update
+ req = nfs_page_find_request_locked(page);
+ 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 (new) {
- if (radix_tree_preload(GFP_NOFS)) {
- nfs_release_request(new);
- return ERR_PTR(-ENOMEM);
- }
- }
+ if (offset > rqend
+ || end < req->wb_offset)
+ goto out_flushme;
- spin_lock(&inode->i_lock);
- req = nfs_page_find_request_locked(page);
- if (req) {
- if (!nfs_set_page_tag_locked(req)) {
- int error;
-
- spin_unlock(&inode->i_lock);
- error = nfs_wait_on_request(req);
- nfs_release_request(req);
- if (error < 0) {
- if (new) {
- radix_tree_preload_end();
- nfs_release_request(new);
- }
- return ERR_PTR(error);
- }
- continue;
- }
- spin_unlock(&inode->i_lock);
- if (new) {
- radix_tree_preload_end();
- nfs_release_request(new);
- }
+ if (nfs_set_page_tag_locked(req))
break;
- }
- if (new) {
- nfs_lock_request_dontget(new);
- nfs_inode_add_request(inode, new);
- spin_unlock(&inode->i_lock);
- radix_tree_preload_end();
- req = new;
- goto zero_page;
- }
+ /* The request is locked, so wait and then retry */
spin_unlock(&inode->i_lock);
-
- new = nfs_create_request(ctx, inode, page, offset, bytes);
- if (IS_ERR(new))
- return new;
+ 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_clear_page_tag_locked(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) != NULL) {
+ NFS_I(inode)->ncommit--;
+ pnfs_clear_request_commit(req);
}
/* Okay, the request matches. Update the region */
if (offset < req->wb_offset) {
req->wb_offset = offset;
req->wb_pgbase = offset;
- req->wb_bytes = max(end, rqend) - req->wb_offset;
- goto zero_page;
}
-
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;
-zero_page:
- /* If this page might potentially be marked as up to date,
- * then we need to zero any uninitalised data. */
- if (req->wb_pgbase == 0 && req->wb_bytes != PAGE_CACHE_SIZE
- && !PageUptodate(req->wb_page))
- zero_user_segment(req->wb_page, req->wb_bytes, PAGE_CACHE_SIZE);
+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);
+ nfs_mark_request_dirty(req);
+ /* Update file length */
+ nfs_grow_file(page, offset, count);
+ nfs_mark_uptodate(page, req->wb_pgbase, req->wb_bytes);
+ nfs_mark_request_dirty(req);
+ nfs_clear_page_tag_locked(req);
+ return 0;
+}
+
int nfs_flush_incompatible(struct file *file, struct page *page)
{
struct nfs_open_context *ctx = nfs_file_open_context(file);
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 ||
+ req->wb_lock_context->lockowner != current->files ||
+ req->wb_lock_context->pid != current->tgid;
nfs_release_request(req);
if (!do_flush)
return 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 up to date, it may be more efficient to extend the write
*/
if (nfs_write_pageuptodate(page, inode) &&
inode->i_flock == NULL &&
- !(file->f_flags & O_SYNC)) {
+ !(file->f_flags & O_DSYNC)) {
count = max(count + offset, nfs_page_length(page));
offset = 0;
}
status = nfs_writepage_setup(ctx, page, offset, count);
- __set_page_dirty_nobuffers(page);
-
- dprintk("NFS: nfs_updatepage returns %d (isize %Ld)\n",
- status, (long long)i_size_read(inode));
if (status < 0)
nfs_set_pageerror(page);
+
+ dprintk("NFS: nfs_updatepage returns %d (isize %lld)\n",
+ status, (long long)i_size_read(inode));
return status;
}
-static void nfs_writepage_release(struct nfs_page *req)
+static void nfs_writepage_release(struct nfs_page *req,
+ struct nfs_write_data *data)
{
+ struct page *page = req->wb_page;
- if (PageError(req->wb_page)) {
- nfs_end_page_writeback(req->wb_page);
+ if (PageError(req->wb_page) || !nfs_reschedule_unstable_write(req, data))
nfs_inode_remove_request(req);
- } else if (!nfs_reschedule_unstable_write(req)) {
- /* Set the PG_uptodate flag */
- nfs_mark_uptodate(req->wb_page, req->wb_pgbase, req->wb_bytes);
- 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);
+ nfs_end_page_writeback(page);
}
static int flush_task_priority(int how)
return RPC_PRIORITY_NORMAL;
}
-/*
- * Set up the argument/result storage required for the RPC call.
- */
-static void nfs_write_rpcsetup(struct nfs_page *req,
- struct nfs_write_data *data,
- const struct rpc_call_ops *call_ops,
- unsigned int count, unsigned int offset,
- int how)
+int nfs_initiate_write(struct nfs_write_data *data,
+ struct rpc_clnt *clnt,
+ const struct rpc_call_ops *call_ops,
+ int how)
{
- struct inode *inode = req->wb_context->path.dentry->d_inode;
- int flags = (how & FLUSH_SYNC) ? 0 : RPC_TASK_ASYNC;
+ struct inode *inode = data->inode;
int priority = flush_task_priority(how);
struct rpc_task *task;
struct rpc_message msg = {
.rpc_argp = &data->args,
.rpc_resp = &data->res,
- .rpc_cred = req->wb_context->cred,
+ .rpc_cred = data->cred,
};
struct rpc_task_setup task_setup_data = {
- .rpc_client = NFS_CLIENT(inode),
+ .rpc_client = clnt,
.task = &data->task,
.rpc_message = &msg,
.callback_ops = call_ops,
.callback_data = data,
.workqueue = nfsiod_workqueue,
- .flags = flags,
+ .flags = RPC_TASK_ASYNC,
.priority = priority,
};
+ int ret = 0;
+
+ /* Set up the initial task struct. */
+ NFS_PROTO(inode)->write_setup(data, &msg);
+
+ dprintk("NFS: %5u initiated write call "
+ "(req %s/%lld, %u bytes @ offset %llu)\n",
+ data->task.tk_pid,
+ inode->i_sb->s_id,
+ (long long)NFS_FILEID(inode),
+ data->args.count,
+ (unsigned long long)data->args.offset);
+
+ task = rpc_run_task(&task_setup_data);
+ if (IS_ERR(task)) {
+ ret = PTR_ERR(task);
+ goto out;
+ }
+ if (how & FLUSH_SYNC) {
+ ret = rpc_wait_for_completion_task(task);
+ if (ret == 0)
+ ret = task->tk_status;
+ }
+ rpc_put_task(task);
+out:
+ return ret;
+}
+EXPORT_SYMBOL_GPL(nfs_initiate_write);
+
+/*
+ * Set up the argument/result storage required for the RPC call.
+ */
+static 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,
+ struct pnfs_layout_segment *lseg,
+ int how)
+{
+ struct inode *inode = req->wb_context->path.dentry->d_inode;
/* Set up the RPC argument and reply structs
* NB: take care not to mess about with data->commit et al. */
data->req = req;
data->inode = inode = req->wb_context->path.dentry->d_inode;
- data->cred = msg.rpc_cred;
+ data->cred = req->wb_context->cred;
+ data->lseg = get_lseg(lseg);
data->args.fh = NFS_FH(inode);
data->args.offset = req_offset(req) + offset;
data->args.pages = data->pagevec;
data->args.count = count;
data->args.context = get_nfs_open_context(req->wb_context);
+ data->args.lock_context = req->wb_lock_context;
data->args.stable = NFS_UNSTABLE;
- if (how & FLUSH_STABLE) {
+ if (how & (FLUSH_STABLE | FLUSH_COND_STABLE)) {
data->args.stable = NFS_DATA_SYNC;
- if (!NFS_I(inode)->ncommit)
+ if (!nfs_need_commit(NFS_I(inode)))
data->args.stable = NFS_FILE_SYNC;
}
data->res.verf = &data->verf;
nfs_fattr_init(&data->fattr);
- /* Set up the initial task struct. */
- NFS_PROTO(inode)->write_setup(data, &msg);
+ if (data->lseg &&
+ (pnfs_try_to_write_data(data, call_ops, how) == PNFS_ATTEMPTED))
+ return 0;
- dprintk("NFS: %5u initiated write call "
- "(req %s/%Ld, %u bytes @ offset %Lu)\n",
- data->task.tk_pid,
- inode->i_sb->s_id,
- (long long)NFS_FILEID(inode),
- count,
- (unsigned long long)data->args.offset);
+ return nfs_initiate_write(data, NFS_CLIENT(inode), call_ops, how);
+}
- task = rpc_run_task(&task_setup_data);
- if (!IS_ERR(task))
- rpc_put_task(task);
+/* 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 page *page = req->wb_page;
+
+ nfs_mark_request_dirty(req);
+ nfs_clear_page_tag_locked(req);
+ nfs_end_page_writeback(page);
}
/*
* Generate multiple small requests to write out a single
* contiguous dirty area on one page.
*/
-static int nfs_flush_multi(struct inode *inode, struct list_head *head, unsigned int npages, size_t count, int how)
+static int nfs_flush_multi(struct nfs_pageio_descriptor *desc)
{
- struct nfs_page *req = nfs_list_entry(head->next);
+ struct nfs_page *req = nfs_list_entry(desc->pg_list.next);
struct page *page = req->wb_page;
struct nfs_write_data *data;
- size_t wsize = NFS_SERVER(inode)->wsize, nbytes;
+ size_t wsize = NFS_SERVER(desc->pg_inode)->wsize, nbytes;
unsigned int offset;
int requests = 0;
+ int ret = 0;
+ struct pnfs_layout_segment *lseg;
LIST_HEAD(list);
nfs_list_remove_request(req);
- nbytes = count;
+ if ((desc->pg_ioflags & FLUSH_COND_STABLE) &&
+ (desc->pg_moreio || NFS_I(desc->pg_inode)->ncommit ||
+ desc->pg_count > wsize))
+ desc->pg_ioflags &= ~FLUSH_COND_STABLE;
+
+
+ nbytes = desc->pg_count;
do {
size_t len = min(nbytes, wsize);
} while (nbytes != 0);
atomic_set(&req->wb_complete, requests);
+ BUG_ON(desc->pg_lseg);
+ lseg = pnfs_update_layout(desc->pg_inode, req->wb_context, IOMODE_RW);
ClearPageError(page);
offset = 0;
- nbytes = count;
+ nbytes = desc->pg_count;
do {
+ int ret2;
+
data = list_entry(list.next, struct nfs_write_data, pages);
list_del_init(&data->pages);
if (nbytes < wsize)
wsize = nbytes;
- nfs_write_rpcsetup(req, data, &nfs_write_partial_ops,
- wsize, offset, how);
+ ret2 = nfs_write_rpcsetup(req, data, &nfs_write_partial_ops,
+ wsize, offset, lseg, desc->pg_ioflags);
+ if (ret == 0)
+ ret = ret2;
offset += wsize;
nbytes -= wsize;
} while (nbytes != 0);
- return 0;
+ put_lseg(lseg);
+ desc->pg_lseg = NULL;
+ return ret;
out_bad:
while (!list_empty(&list)) {
data = list_entry(list.next, struct nfs_write_data, pages);
list_del(&data->pages);
- nfs_writedata_release(data);
+ nfs_writedata_free(data);
}
nfs_redirty_request(req);
- nfs_end_page_writeback(req->wb_page);
- nfs_clear_page_tag_locked(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, unsigned int npages, size_t count, int how)
+static int nfs_flush_one(struct nfs_pageio_descriptor *desc)
{
struct nfs_page *req;
struct page **pages;
struct nfs_write_data *data;
+ struct list_head *head = &desc->pg_list;
+ struct pnfs_layout_segment *lseg = desc->pg_lseg;
+ int ret;
- data = nfs_writedata_alloc(npages);
- if (!data)
- goto out_bad;
-
+ data = nfs_writedata_alloc(nfs_page_array_len(desc->pg_base,
+ desc->pg_count));
+ if (!data) {
+ while (!list_empty(head)) {
+ req = nfs_list_entry(head->next);
+ nfs_list_remove_request(req);
+ nfs_redirty_request(req);
+ }
+ ret = -ENOMEM;
+ goto out;
+ }
pages = data->pagevec;
while (!list_empty(head)) {
req = nfs_list_entry(head->next);
*pages++ = req->wb_page;
}
req = nfs_list_entry(data->pages.next);
+ if ((!lseg) && list_is_singular(&data->pages))
+ lseg = pnfs_update_layout(desc->pg_inode, req->wb_context, IOMODE_RW);
- /* Set up the argument struct */
- nfs_write_rpcsetup(req, data, &nfs_write_full_ops, count, 0, how);
+ if ((desc->pg_ioflags & FLUSH_COND_STABLE) &&
+ (desc->pg_moreio || NFS_I(desc->pg_inode)->ncommit))
+ desc->pg_ioflags &= ~FLUSH_COND_STABLE;
- return 0;
- out_bad:
- while (!list_empty(head)) {
- req = nfs_list_entry(head->next);
- nfs_list_remove_request(req);
- nfs_redirty_request(req);
- nfs_end_page_writeback(req->wb_page);
- nfs_clear_page_tag_locked(req);
- }
- return -ENOMEM;
+ /* Set up the argument struct */
+ ret = nfs_write_rpcsetup(req, data, &nfs_write_full_ops, desc->pg_count, 0, lseg, desc->pg_ioflags);
+out:
+ put_lseg(lseg); /* Cleans any gotten in ->pg_test */
+ desc->pg_lseg = NULL;
+ return ret;
}
static void nfs_pageio_init_write(struct nfs_pageio_descriptor *pgio,
{
size_t wsize = NFS_SERVER(inode)->wsize;
+ pnfs_pageio_init_write(pgio, inode);
+
if (wsize < PAGE_CACHE_SIZE)
nfs_pageio_init(pgio, inode, nfs_flush_multi, wsize, ioflags);
else
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->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));
+ 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) {
+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, task->tk_status);
- dprintk(", error = %d\n", task->tk_status);
+ nfs_context_set_write_error(req->wb_context, status);
+ dprintk(", error = %d\n", status);
goto out;
}
out:
if (atomic_dec_and_test(&req->wb_complete))
- nfs_writepage_release(req);
+ nfs_writepage_release(req, data);
+ nfs_writedata_release(calldata);
+}
+
+#if defined(CONFIG_NFS_V4_1)
+void nfs_write_prepare(struct rpc_task *task, void *calldata)
+{
+ struct nfs_write_data *data = calldata;
+
+ if (nfs4_setup_sequence(NFS_SERVER(data->inode),
+ &data->args.seq_args,
+ &data->res.seq_res, 1, task))
+ return;
+ rpc_call_start(task);
}
+#endif /* CONFIG_NFS_V4_1 */
static const struct rpc_call_ops nfs_write_partial_ops = {
+#if defined(CONFIG_NFS_V4_1)
+ .rpc_call_prepare = nfs_write_prepare,
+#endif /* CONFIG_NFS_V4_1 */
.rpc_call_done = nfs_writeback_done_partial,
- .rpc_release = nfs_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)",
+ 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) {
+ if (status < 0) {
nfs_set_pageerror(page);
- nfs_context_set_write_error(req->wb_context, task->tk_status);
- dprintk(", error = %d\n", task->tk_status);
+ nfs_context_set_write_error(req->wb_context, status);
+ dprintk(", error = %d\n", status);
goto remove_request;
}
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);
+ nfs_mark_request_commit(req, data->lseg);
dprintk(" marked for commit\n");
goto next;
}
- /* Set the PG_uptodate flag? */
- nfs_mark_uptodate(page, req->wb_pgbase, req->wb_bytes);
dprintk(" OK\n");
remove_request:
- nfs_end_page_writeback(page);
nfs_inode_remove_request(req);
next:
nfs_clear_page_tag_locked(req);
+ nfs_end_page_writeback(page);
}
+ 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,
};
/*
* This function is called when the WRITE call is complete.
*/
-int nfs_writeback_done(struct rpc_task *task, struct nfs_write_data *data)
+void 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: %5u nfs_writeback_done (status %d)\n",
*/
status = NFS_PROTO(data->inode)->write_done(task, data);
if (status != 0)
- return status;
+ return;
nfs_add_stats(data->inode, NFSIOS_SERVERWRITTENBYTES, resp->count);
#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
*/
static unsigned long complain;
+ /* Note this will print the MDS for a DS write */
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;
}
/* Was this an NFSv2 write or an NFSv3 stable write? */
if (resp->verf->committed != NFS_UNSTABLE) {
/* Resend from where the server left off */
+ data->mds_offset += resp->count;
argp->offset += resp->count;
argp->pgbase += resp->count;
argp->count -= resp->count;
*/
argp->stable = NFS_FILE_SYNC;
}
- rpc_restart_call(task);
- return -EAGAIN;
+ nfs_restart_rpc(task, server->nfs_client);
+ return;
}
if (time_before(complain, jiffies)) {
printk(KERN_WARNING
/* Can't do anything about it except throw an error. */
task->tk_status = -EIO;
}
- return 0;
+ return;
}
#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
-void nfs_commit_release(void *data)
+static int nfs_commit_set_lock(struct nfs_inode *nfsi, int may_wait)
+{
+ int ret;
+
+ if (!test_and_set_bit(NFS_INO_COMMIT, &nfsi->flags))
+ return 1;
+ if (!may_wait)
+ return 0;
+ ret = out_of_line_wait_on_bit_lock(&nfsi->flags,
+ NFS_INO_COMMIT,
+ nfs_wait_bit_killable,
+ TASK_KILLABLE);
+ return (ret < 0) ? ret : 1;
+}
+
+void nfs_commit_clear_lock(struct nfs_inode *nfsi)
+{
+ clear_bit(NFS_INO_COMMIT, &nfsi->flags);
+ smp_mb__after_clear_bit();
+ wake_up_bit(&nfsi->flags, NFS_INO_COMMIT);
+}
+EXPORT_SYMBOL_GPL(nfs_commit_clear_lock);
+
+void nfs_commitdata_release(void *data)
{
struct nfs_write_data *wdata = data;
+ put_lseg(wdata->lseg);
put_nfs_open_context(wdata->args.context);
nfs_commit_free(wdata);
}
+EXPORT_SYMBOL_GPL(nfs_commitdata_release);
-/*
- * Set up the argument/result storage required for the RPC call.
- */
-static void nfs_commit_rpcsetup(struct list_head *head,
- struct nfs_write_data *data,
- int how)
+int nfs_initiate_commit(struct nfs_write_data *data, struct rpc_clnt *clnt,
+ const struct rpc_call_ops *call_ops,
+ int how)
{
- 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;
+ int priority = flush_task_priority(how);
struct rpc_message msg = {
.rpc_argp = &data->args,
.rpc_resp = &data->res,
- .rpc_cred = first->wb_context->cred,
+ .rpc_cred = data->cred,
};
struct rpc_task_setup task_setup_data = {
.task = &data->task,
- .rpc_client = NFS_CLIENT(inode),
+ .rpc_client = clnt,
.rpc_message = &msg,
- .callback_ops = &nfs_commit_ops,
+ .callback_ops = call_ops,
.callback_data = data,
.workqueue = nfsiod_workqueue,
- .flags = flags,
+ .flags = RPC_TASK_ASYNC,
.priority = priority,
};
+ /* Set up the initial task struct. */
+ NFS_PROTO(data->inode)->commit_setup(data, &msg);
+
+ 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);
+ if (how & FLUSH_SYNC)
+ rpc_wait_for_completion_task(task);
+ rpc_put_task(task);
+ return 0;
+}
+EXPORT_SYMBOL_GPL(nfs_initiate_commit);
+
+/*
+ * Set up the argument/result storage required for the RPC call.
+ */
+void nfs_init_commit(struct nfs_write_data *data,
+ struct list_head *head,
+ struct pnfs_layout_segment *lseg)
+{
+ struct nfs_page *first = nfs_list_entry(head->next);
+ struct inode *inode = first->wb_context->path.dentry->d_inode;
/* 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);
data->inode = inode;
- data->cred = msg.rpc_cred;
+ data->cred = first->wb_context->cred;
+ data->lseg = lseg; /* reference transferred */
+ data->mds_ops = &nfs_commit_ops;
data->args.fh = NFS_FH(data->inode);
/* Note: we always request a commit of the entire inode */
data->res.fattr = &data->fattr;
data->res.verf = &data->verf;
nfs_fattr_init(&data->fattr);
+}
+EXPORT_SYMBOL_GPL(nfs_init_commit);
- /* Set up the initial task struct. */
- NFS_PROTO(inode)->commit_setup(data, &msg);
-
- dprintk("NFS: %5u initiated commit call\n", data->task.tk_pid);
+void nfs_retry_commit(struct list_head *page_list,
+ struct pnfs_layout_segment *lseg)
+{
+ struct nfs_page *req;
- task = rpc_run_task(&task_setup_data);
- if (!IS_ERR(task))
- rpc_put_task(task);
+ while (!list_empty(page_list)) {
+ req = nfs_list_entry(page_list->next);
+ nfs_list_remove_request(req);
+ nfs_mark_request_commit(req, lseg);
+ dec_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
+ dec_bdi_stat(req->wb_page->mapping->backing_dev_info,
+ BDI_RECLAIMABLE);
+ nfs_clear_page_tag_locked(req);
+ }
}
+EXPORT_SYMBOL_GPL(nfs_retry_commit);
/*
* Commit dirty pages
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);
-
- return 0;
+ nfs_init_commit(data, head, NULL);
+ return nfs_initiate_commit(data, NFS_CLIENT(inode), data->mds_ops, 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);
- dec_bdi_stat(req->wb_page->mapping->backing_dev_info,
- BDI_RECLAIMABLE);
- nfs_clear_page_tag_locked(req);
- }
+ nfs_retry_commit(head, NULL);
+ nfs_commit_clear_lock(NFS_I(inode));
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: %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;
+}
+
+void nfs_commit_release_pages(struct nfs_write_data *data)
+{
+ 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);
- clear_bit(PG_NEED_COMMIT, &(req)->wb_flags);
- dec_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
- dec_bdi_stat(req->wb_page->mapping->backing_dev_info,
- BDI_RECLAIMABLE);
+ nfs_clear_request_commit(req);
- dprintk("NFS: commit (%s/%Ld %d@%Ld)",
+ 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) {
- nfs_context_set_write_error(req->wb_context, 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;
}
* returned by the server against all stored verfs. */
if (!memcmp(req->wb_verf.verifier, data->verf.verifier, sizeof(data->verf.verifier))) {
/* We have a match */
- /* Set the PG_uptodate flag */
- nfs_mark_uptodate(req->wb_page, req->wb_pgbase,
- req->wb_bytes);
nfs_inode_remove_request(req);
dprintk(" OK\n");
goto next;
}
/* We have a mismatch. Write the page again */
dprintk(" mismatch\n");
- nfs_redirty_request(req);
+ nfs_mark_request_dirty(req);
next:
nfs_clear_page_tag_locked(req);
}
}
+EXPORT_SYMBOL_GPL(nfs_commit_release_pages);
+
+static void nfs_commit_release(void *calldata)
+{
+ struct nfs_write_data *data = calldata;
+
+ nfs_commit_release_pages(data);
+ nfs_commit_clear_lock(NFS_I(data->inode));
+ 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,
};
int nfs_commit_inode(struct inode *inode, int how)
{
LIST_HEAD(head);
+ int may_wait = how & FLUSH_SYNC;
int res;
+ res = nfs_commit_set_lock(NFS_I(inode), may_wait);
+ if (res <= 0)
+ goto out_mark_dirty;
spin_lock(&inode->i_lock);
res = nfs_scan_commit(inode, &head, 0, 0);
spin_unlock(&inode->i_lock);
if (res) {
- int error = nfs_commit_list(inode, &head, how);
+ int error;
+
+ error = pnfs_commit_list(inode, &head, how);
+ if (error == PNFS_NOT_ATTEMPTED)
+ error = nfs_commit_list(inode, &head, how);
if (error < 0)
return error;
- }
+ if (!may_wait)
+ goto out_mark_dirty;
+ error = wait_on_bit(&NFS_I(inode)->flags,
+ NFS_INO_COMMIT,
+ nfs_wait_bit_killable,
+ TASK_KILLABLE);
+ if (error < 0)
+ return error;
+ } else
+ nfs_commit_clear_lock(NFS_I(inode));
+ return res;
+ /* Note: If we exit without ensuring that the commit is complete,
+ * we must mark the inode as dirty. Otherwise, future calls to
+ * sync_inode() with the WB_SYNC_ALL flag set will fail to ensure
+ * that the data is on the disk.
+ */
+out_mark_dirty:
+ __mark_inode_dirty(inode, I_DIRTY_DATASYNC);
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)
+static int nfs_commit_unstable_pages(struct inode *inode, struct writeback_control *wbc)
{
- struct inode *inode = mapping->host;
- pgoff_t idx_start, idx_end;
- unsigned int npages = 0;
- LIST_HEAD(head);
- int nocommit = how & FLUSH_NOCOMMIT;
- long pages, ret;
-
- /* FIXME */
- if (wbc->range_cyclic)
- idx_start = 0;
- else {
- idx_start = wbc->range_start >> PAGE_CACHE_SHIFT;
- idx_end = wbc->range_end >> PAGE_CACHE_SHIFT;
- if (idx_end > idx_start) {
- pgoff_t l_npages = 1 + idx_end - idx_start;
- npages = l_npages;
- if (sizeof(npages) != sizeof(l_npages) &&
- (pgoff_t)npages != l_npages)
- npages = 0;
- }
+ struct nfs_inode *nfsi = NFS_I(inode);
+ int flags = FLUSH_SYNC;
+ int ret = 0;
+
+ if (wbc->sync_mode == WB_SYNC_NONE) {
+ /* Don't commit yet if this is a non-blocking flush and there
+ * are a lot of outstanding writes for this mapping.
+ */
+ if (nfsi->ncommit <= (nfsi->npages >> 1))
+ goto out_mark_dirty;
+
+ /* don't wait for the COMMIT response */
+ flags = 0;
}
- how &= ~FLUSH_NOCOMMIT;
- spin_lock(&inode->i_lock);
- do {
- ret = nfs_wait_on_requests_locked(inode, idx_start, npages);
- if (ret != 0)
- continue;
- if (nocommit)
- break;
- pages = nfs_scan_commit(inode, &head, idx_start, npages);
- if (pages == 0)
- break;
- if (how & FLUSH_INVALIDATE) {
- spin_unlock(&inode->i_lock);
- nfs_cancel_commit_list(&head);
- ret = pages;
- spin_lock(&inode->i_lock);
- continue;
- }
- pages += nfs_scan_commit(inode, &head, 0, 0);
- spin_unlock(&inode->i_lock);
- ret = nfs_commit_list(inode, &head, how);
- spin_lock(&inode->i_lock);
- } while (ret >= 0);
- spin_unlock(&inode->i_lock);
+ ret = nfs_commit_inode(inode, flags);
+ if (ret >= 0) {
+ if (wbc->sync_mode == WB_SYNC_NONE) {
+ if (ret < wbc->nr_to_write)
+ wbc->nr_to_write -= ret;
+ else
+ wbc->nr_to_write = 0;
+ }
+ return 0;
+ }
+out_mark_dirty:
+ __mark_inode_dirty(inode, I_DIRTY_DATASYNC);
return ret;
}
-
-static int __nfs_write_mapping(struct address_space *mapping, struct writeback_control *wbc, int how)
+#else
+static int nfs_commit_unstable_pages(struct inode *inode, struct writeback_control *wbc)
{
- 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;
}
+#endif
-/* Two pass sync: first using WB_SYNC_NONE, then WB_SYNC_ALL */
-static int nfs_write_mapping(struct address_space *mapping, int how)
+int nfs_write_inode(struct inode *inode, struct writeback_control *wbc)
{
- struct writeback_control wbc = {
- .bdi = mapping->backing_dev_info,
- .sync_mode = WB_SYNC_NONE,
- .nr_to_write = LONG_MAX,
- .for_writepages = 1,
- .range_cyclic = 1,
- };
int ret;
- ret = __nfs_write_mapping(mapping, &wbc, how);
- if (ret < 0)
- return ret;
- wbc.sync_mode = WB_SYNC_ALL;
- return __nfs_write_mapping(mapping, &wbc, how);
+ ret = nfs_commit_unstable_pages(inode, wbc);
+ if (ret >= 0 && test_bit(NFS_INO_LAYOUTCOMMIT, &NFS_I(inode)->flags)) {
+ int status, sync = wbc->sync_mode;
+
+ if (wbc->nonblocking || wbc->for_background)
+ sync = 0;
+
+ status = pnfs_layoutcommit_inode(inode, sync);
+ if (status < 0)
+ return status;
+ }
+ return ret;
}
/*
*/
int nfs_wb_all(struct inode *inode)
{
- return nfs_write_mapping(inode->i_mapping, 0);
-}
+ struct writeback_control wbc = {
+ .sync_mode = WB_SYNC_ALL,
+ .nr_to_write = LONG_MAX,
+ .range_start = 0,
+ .range_end = LLONG_MAX,
+ };
-int nfs_wb_nocommit(struct inode *inode)
-{
- return nfs_write_mapping(inode->i_mapping, FLUSH_NOCOMMIT);
+ return sync_inode(inode, &wbc);
}
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,
- };
int ret = 0;
BUG_ON(!PageLocked(page));
for (;;) {
+ wait_on_page_writeback(page);
req = nfs_page_find_request(page);
if (req == NULL)
- goto out;
- if (test_bit(PG_NEED_COMMIT, &req->wb_flags)) {
- nfs_release_request(req);
break;
- }
if (nfs_lock_request_dontget(req)) {
nfs_inode_remove_request(req);
/*
break;
}
ret = nfs_wait_on_request(req);
+ nfs_release_request(req);
if (ret < 0)
- goto out;
+ break;
}
- 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)
+/*
+ * Write back all requests on one page - we do this before reading it.
+ */
+int nfs_wb_page(struct inode *inode, struct page *page)
{
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,
+ .nr_to_write = 0,
.range_start = range_start,
.range_end = range_end,
};
int ret;
- BUG_ON(!PageLocked(page));
- if (clear_page_dirty_for_io(page)) {
- ret = nfs_writepage_locked(page, &wbc);
+ for (;;) {
+ wait_on_page_writeback(page);
+ if (clear_page_dirty_for_io(page)) {
+ ret = nfs_writepage_locked(page, &wbc);
+ if (ret < 0)
+ goto out_error;
+ continue;
+ }
+ if (!PagePrivate(page))
+ break;
+ ret = nfs_commit_inode(inode, FLUSH_SYNC);
if (ret < 0)
- goto out;
+ goto out_error;
}
- if (!PagePrivate(page))
- return 0;
- ret = nfs_sync_mapping_wait(page->mapping, &wbc, how);
- if (ret >= 0)
- return 0;
-out:
- __mark_inode_dirty(inode, I_DIRTY_PAGES);
+ return 0;
+out_error:
return ret;
}
-/*
- * Write back all requests on one page - we do this before reading it.
- */
-int nfs_wb_page(struct inode *inode, struct page* page)
+#ifdef CONFIG_MIGRATION
+int nfs_migrate_page(struct address_space *mapping, struct page *newpage,
+ struct page *page)
{
- return nfs_wb_page_priority(inode, page, FLUSH_STABLE);
+ struct nfs_page *req;
+ int ret;
+
+ nfs_fscache_release_page(page, GFP_KERNEL);
+
+ req = nfs_find_and_lock_request(page, false);
+ 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);
+ spin_lock(&mapping->host->i_lock);
+ req->wb_page = newpage;
+ SetPagePrivate(newpage);
+ set_page_private(newpage, (unsigned long)req);
+ ClearPagePrivate(page);
+ set_page_private(page, 0);
+ spin_unlock(&mapping->host->i_lock);
+ page_cache_release(page);
+out_unlock:
+ nfs_clear_page_tag_locked(req);
+out:
+ return ret;
}
+#endif
int __init nfs_init_writepagecache(void)
{
Code Review / linux-2.6.git / blobdiff