nfs: page group syncing in write path

Operations that modify state for a whole page must be syncronized across
all requests within a page group. In the write path, this is calling
end_page_writeback and removing the head request from an inode.
Both of these operations should not be called until all requests
in a page group have reached the point where they would call them.

This patch should have no effect yet since all page groups currently
have one request, but will come into play when pg_test functions are
modified to split pages into sub-page regions.

Signed-off-by: Weston Andros Adamson <dros@primarydata.com>
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>

3 files changed, 24 insertions, 12 deletions

diff --git a/fs/nfs/pagelist.c b/fs/nfs/pagelist.c
index 18ee4e99347e..ceb4424614aa 100644
--- a/fs/nfs/pagelist.c
+++ b/fs/nfs/pagelist.c
@@ -397,6 +397,8 @@ static void nfs_free_request(struct nfs_page *req)
 	WARN_ON_ONCE(test_bit(PG_TEARDOWN, &req->wb_flags));
 	WARN_ON_ONCE(test_bit(PG_UNLOCKPAGE, &req->wb_flags));
 	WARN_ON_ONCE(test_bit(PG_UPTODATE, &req->wb_flags));
+	WARN_ON_ONCE(test_bit(PG_WB_END, &req->wb_flags));
+	WARN_ON_ONCE(test_bit(PG_REMOVE, &req->wb_flags));
 
 	/* Release struct file and open context */
 	nfs_clear_request(req);
diff --git a/fs/nfs/write.c b/fs/nfs/write.c
index d0f30f12a8b3..5d752766139d 100644
--- a/fs/nfs/write.c
+++ b/fs/nfs/write.c
@@ -201,12 +201,15 @@ static void nfs_set_page_writeback(struct page *page)
 	}
 }
 
-static void nfs_end_page_writeback(struct page *page)
+static void nfs_end_page_writeback(struct nfs_page *req)
 {
-	struct inode *inode = page_file_mapping(page)->host;
+	struct inode *inode = page_file_mapping(req->wb_page)->host;
 	struct nfs_server *nfss = NFS_SERVER(inode);
 
-	end_page_writeback(page);
+	if (!nfs_page_group_sync_on_bit(req, PG_WB_END))
+		return;
+
+	end_page_writeback(req->wb_page);
 	if (atomic_long_dec_return(&nfss->writeback) < NFS_CONGESTION_OFF_THRESH)
 		clear_bdi_congested(&nfss->backing_dev_info, BLK_RW_ASYNC);
 }
@@ -397,15 +400,20 @@ static void nfs_inode_remove_request(struct nfs_page *req)
 {
 	struct inode *inode = req->wb_context->dentry->d_inode;
 	struct nfs_inode *nfsi = NFS_I(inode);
+	struct nfs_page *head;
 
-	spin_lock(&inode->i_lock);
-	if (likely(!PageSwapCache(req->wb_page))) {
-		set_page_private(req->wb_page, 0);
-		ClearPagePrivate(req->wb_page);
-		clear_bit(PG_MAPPED, &req->wb_flags);
+	if (nfs_page_group_sync_on_bit(req, PG_REMOVE)) {
+		head = req->wb_head;
+
+		spin_lock(&inode->i_lock);
+		if (likely(!PageSwapCache(head->wb_page))) {
+			set_page_private(head->wb_page, 0);
+			ClearPagePrivate(head->wb_page);
+			clear_bit(PG_MAPPED, &head->wb_flags);
+		}
+		nfsi->npages--;
+		spin_unlock(&inode->i_lock);
 	}
-	nfsi->npages--;
-	spin_unlock(&inode->i_lock);
 	nfs_release_request(req);
 }
 
@@ -599,7 +607,7 @@ remove_req:
 		nfs_inode_remove_request(req);
 next:
 		nfs_unlock_request(req);
-		nfs_end_page_writeback(req->wb_page);
+		nfs_end_page_writeback(req);
 		do_destroy = !test_bit(NFS_IOHDR_NEED_COMMIT, &hdr->flags);
 		nfs_release_request(req);
 	}
@@ -964,7 +972,7 @@ static void nfs_redirty_request(struct nfs_page *req)
 {
 	nfs_mark_request_dirty(req);
 	nfs_unlock_request(req);
-	nfs_end_page_writeback(req->wb_page);
+	nfs_end_page_writeback(req);
 	nfs_release_request(req);
 }
 
diff --git a/include/linux/nfs_page.h b/include/linux/nfs_page.h
index 6385175a127b..7d9096d95d4a 100644
--- a/include/linux/nfs_page.h
+++ b/include/linux/nfs_page.h
@@ -31,6 +31,8 @@ enum {
 	PG_TEARDOWN,		/* page group sync for destroy */
 	PG_UNLOCKPAGE,		/* page group sync bit in read path */
 	PG_UPTODATE,		/* page group sync bit in read path */
+	PG_WB_END,		/* page group sync bit in write path */
+	PG_REMOVE,		/* page group sync bit in write path */
 };
 
 struct nfs_inode;