[PATCH] remove i_dirty_data_buffers

Removes inode.i_dirty_data_buffers.  It's no longer used - all dirty
buffers have their pages marked dirty and filemap_fdatasync() /
filemap_fdatawait() catches it all.

Updates all callers.

This required a change in JFS - it has "metapages" which
are a container around a page which holds metadata.  They
were holding these pages locked and were relying on fsync_inode_data_buffers
for writing them out.  So fdatasync() deadlocked.

I've changed JFS to not lock those pages.  Change was acked
by Dave Kleikamp <shaggy@austin.ibm.com> as the right
thing to do, but may not be complete.  Probably igrab()
against ->host is needed to pin the address_space down.

fs/buffer.c

@@ -428,8 +428,7 @@ int inode_has_buffers(struct inode *inode)
 	int ret;
 	
 	spin_lock(&inode->i_bufferlist_lock);
-	ret = !list_empty(&inode->i_dirty_buffers) ||
-			!list_empty(&inode->i_dirty_data_buffers);
+	ret = !list_empty(&inode->i_dirty_buffers);
 	spin_unlock(&inode->i_bufferlist_lock);
 	
 	return ret;
@@ -694,9 +693,6 @@ void invalidate_inode_buffers(struct inode *inode)
 	while ((entry = inode->i_dirty_buffers.next) !=
 				&inode->i_dirty_buffers)
 		__remove_inode_queue(BH_ENTRY(entry));
-	while ((entry = inode->i_dirty_data_buffers.next) !=
-				&inode->i_dirty_data_buffers)
-		__remove_inode_queue(BH_ENTRY(entry));
 	spin_unlock(&inode->i_bufferlist_lock);
 }
 
@@ -954,10 +950,6 @@ __getblk(struct block_device *bdev, sector_t block, int size)
  * block_read_full_page() against that page will discover all the uptodate
  * buffers, will set the page uptodate and will perform no I/O.
  */
-static inline void __mark_dirty(struct buffer_head *bh)
-{
-	__set_page_dirty_nobuffers(bh->b_page);
-}
 
 /**
  * mark_buffer_dirty - mark a buffer_head as needing writeout
@@ -973,7 +965,7 @@ static inline void __mark_dirty(struct buffer_head *bh)
 void mark_buffer_dirty(struct buffer_head *bh)
 {
 	if (!atomic_set_buffer_dirty(bh))
-		__mark_dirty(bh);
+		__set_page_dirty_nobuffers(bh->b_page);
 }
 
 /*
@@ -1498,10 +1490,7 @@ static int __block_commit_write(struct inode *inode, struct page *page,
 				partial = 1;
 		} else {
 			mark_buffer_uptodate(bh, 1);
-			if (!atomic_set_buffer_dirty(bh)) {
-				__mark_dirty(bh);
-				buffer_insert_inode_data_queue(bh, inode);
-			}
+			mark_buffer_dirty(bh);
 		}
 	}
 

fs/ext2/fsync.c

@@ -38,7 +38,6 @@ int ext2_sync_file(struct file * file, struct dentry *dentry, int datasync)
 	int err;
 	
 	err  = fsync_inode_buffers(inode);
-	err |= fsync_inode_data_buffers(inode);
 	if (!(inode->i_state & I_DIRTY))
 		return err;
 	if (datasync && !(inode->i_state & I_DIRTY_DATASYNC))

fs/ext3/fsync.c

@@ -62,8 +62,6 @@ int ext3_sync_file(struct file * file, struct dentry *dentry, int datasync)
 	 * we'll end up waiting on them in commit.
 	 */
 	ret = fsync_inode_buffers(inode);
-	ret |= fsync_inode_data_buffers(inode);
-
 	ext3_force_commit(inode->i_sb);
 
 	return ret;

fs/inode.c

@@ -140,7 +140,6 @@ void inode_init_once(struct inode *inode)
 	INIT_LIST_HEAD(&inode->i_data.io_pages);
 	INIT_LIST_HEAD(&inode->i_dentry);
 	INIT_LIST_HEAD(&inode->i_dirty_buffers);
-	INIT_LIST_HEAD(&inode->i_dirty_data_buffers);
 	INIT_LIST_HEAD(&inode->i_devices);
 	sema_init(&inode->i_sem, 1);
 	INIT_RADIX_TREE(&inode->i_data.page_tree, GFP_ATOMIC);

fs/jfs/file.c

@@ -33,8 +33,6 @@ int jfs_fsync(struct file *file, struct dentry *dentry, int datasync)
 	struct inode *inode = dentry->d_inode;
 	int rc = 0;
 
-	rc = fsync_inode_data_buffers(inode);
-
 	if (!(inode->i_state & I_DIRTY))
 		return rc;
 	if (datasync || !(inode->i_state & I_DIRTY_DATASYNC))

fs/jfs/jfs_dmap.c

@@ -325,7 +325,8 @@ int dbSync(struct inode *ipbmap)
 	/*
 	 * write out dirty pages of bmap
 	 */
-	fsync_inode_data_buffers(ipbmap);
+	filemap_fdatasync(ipbmap->i_mapping);
+	filemap_fdatawait(ipbmap->i_mapping);
 
 	ipbmap->i_state |= I_DIRTY;
 	diWriteSpecial(ipbmap);

fs/jfs/jfs_imap.c

@@ -282,7 +282,8 @@ int diSync(struct inode *ipimap)
 	/*
 	 * write out dirty pages of imap
 	 */
-	fsync_inode_data_buffers(ipimap);
+	filemap_fdatasync(ipimap->i_mapping);
+	filemap_fdatawait(ipimap->i_mapping);
 
 	diWriteSpecial(ipimap);
 
@@ -607,7 +608,8 @@ void diFreeSpecial(struct inode *ip)
 		jERROR(1, ("diFreeSpecial called with NULL ip!\n"));
 		return;
 	}
-	fsync_inode_data_buffers(ip);
+	filemap_fdatasync(ip->i_mapping);
+	filemap_fdatawait(ip->i_mapping);
 	truncate_inode_pages(ip->i_mapping, 0);
 	iput(ip);
 }

fs/jfs/jfs_logmgr.c

@@ -966,9 +966,12 @@ int lmLogSync(log_t * log, int nosyncwait)
 		 * We need to make sure all of the "written" metapages
 		 * actually make it to disk
 		 */
-		fsync_inode_data_buffers(sbi->ipbmap);
-		fsync_inode_data_buffers(sbi->ipimap);
-		fsync_inode_data_buffers(sbi->direct_inode);
+		filemap_fdatasync(sbi->ipbmap->i_mapping);
+		filemap_fdatasync(sbi->ipimap->i_mapping);
+		filemap_fdatasync(sbi->direct_inode->i_mapping);
+		filemap_fdatawait(sbi->ipbmap->i_mapping);
+		filemap_fdatawait(sbi->ipimap->i_mapping);
+		filemap_fdatawait(sbi->direct_inode->i_mapping);
 
 		lrd.logtid = 0;
 		lrd.backchain = 0;

fs/jfs/jfs_metapage.c

@@ -349,7 +349,7 @@ metapage_t *__get_metapage(struct inode *inode,
 		page_index = lblock >> l2BlocksPerPage;
 		page_offset = (lblock - (page_index << l2BlocksPerPage)) <<
 		    l2bsize;
-		if ((page_offset + size) > PAGE_SIZE) {
+		if ((page_offset + size) > PAGE_CACHE_SIZE) {
 			spin_unlock(&meta_lock);
 			jERROR(1, ("MetaData crosses page boundary!!\n"));
 			return NULL;
@@ -394,8 +394,10 @@ metapage_t *__get_metapage(struct inode *inode,
 				__free_metapage(mp);
 				spin_unlock(&meta_lock);
 				return NULL;
-			} else
+			} else {
 				INCREMENT(mpStat.pagealloc);
+				unlock_page(mp->page);
+			}
 		} else {
 			jFYI(1,
 			     ("__get_metapage: Calling read_cache_page\n"));
@@ -412,7 +414,6 @@ metapage_t *__get_metapage(struct inode *inode,
 				return NULL;
 			} else
 				INCREMENT(mpStat.pagealloc);
-			lock_page(mp->page);
 		}
 		mp->data = (void *) (kmap(mp->page) + page_offset);
 	}
@@ -459,6 +460,7 @@ static void __write_metapage(metapage_t * mp)
 	page_offset =
 	    (mp->index - (page_index << l2BlocksPerPage)) << l2bsize;
 
+	lock_page(mp->page);
 	rc = mp->mapping->a_ops->prepare_write(NULL, mp->page, page_offset,
 					       page_offset +
 					       mp->logical_size);
@@ -466,6 +468,7 @@ static void __write_metapage(metapage_t * mp)
 		jERROR(1, ("prepare_write return %d!\n", rc));
 		ClearPageUptodate(mp->page);
 		kunmap(mp->page);
+		unlock_page(mp->page);
 		clear_bit(META_dirty, &mp->flag);
 		return;
 	}
@@ -476,6 +479,7 @@ static void __write_metapage(metapage_t * mp)
 		jERROR(1, ("commit_write returned %d\n", rc));
 	}
 
+	unlock_page(mp->page);
 	clear_bit(META_dirty, &mp->flag);
 
 	jFYI(1, ("__write_metapage done\n"));
@@ -525,7 +529,6 @@ void release_metapage(metapage_t * mp)
 			mp->data = 0;
 			if (test_bit(META_dirty, &mp->flag))
 				__write_metapage(mp);
-			unlock_page(mp->page);
 			if (test_bit(META_sync, &mp->flag)) {
 				sync_metapage(mp);
 				clear_bit(META_sync, &mp->flag);
@@ -585,7 +588,9 @@ void invalidate_metapages(struct inode *ip, unsigned long addr,
 			/*
 			 * If in the metapage cache, we've got the page locked
 			 */
+			lock_page(mp->page);
 			block_flushpage(mp->page, 0);
+			unlock_page(mp->page);
 		} else {
 			spin_unlock(&meta_lock);
 			page = find_lock_page(mapping, lblock>>l2BlocksPerPage);
@@ -608,7 +613,6 @@ void invalidate_inode_metapages(struct inode *inode)
 		clear_bit(META_dirty, &mp->flag);
 		set_bit(META_discard, &mp->flag);
 		kunmap(mp->page);
-		unlock_page(mp->page);
 		page_cache_release(mp->page);
 		INCREMENT(mpStat.pagefree);
 		mp->data = 0;

fs/jfs/jfs_txnmgr.c

@@ -1163,8 +1163,10 @@ int txCommit(tid_t tid,		/* transaction identifier */
 		 * committing transactions and use i_sem instead.
 		 */
 		if ((!S_ISDIR(ip->i_mode))
-		    && (tblk->flag & COMMIT_DELETE) == 0)
-			fsync_inode_data_buffers(ip);
+		    && (tblk->flag & COMMIT_DELETE) == 0) {
+			filemap_fdatasync(ip->i_mapping);
+			filemap_fdatawait(ip->i_mapping);
+		}
 
 		/*
 		 * Mark inode as not dirty.  It will still be on the dirty

fs/jfs/super.c

@@ -151,7 +151,8 @@ static void jfs_put_super(struct super_block *sb)
 	 * We need to clean out the direct_inode pages since this inode
 	 * is not in the inode hash.
 	 */
-	fsync_inode_data_buffers(sbi->direct_inode);
+	filemap_fdatasync(sbi->direct_inode->i_mapping);
+	filemap_fdatawait(sbi->direct_inode->i_mapping);
 	truncate_inode_pages(sbi->direct_mapping, 0);
 	iput(sbi->direct_inode);
 	sbi->direct_inode = NULL;
@@ -337,7 +338,8 @@ static int jfs_fill_super(struct super_block *sb, void *data, int silent)
 		jERROR(1, ("jfs_umount failed with return code %d\n", rc));
 	}
 out_mount_failed:
-	fsync_inode_data_buffers(sbi->direct_inode);
+	filemap_fdatasync(sbi->direct_inode->i_mapping);
+	filemap_fdatawait(sbi->direct_inode->i_mapping);
 	truncate_inode_pages(sbi->direct_mapping, 0);
 	make_bad_inode(sbi->direct_inode);
 	iput(sbi->direct_inode);

fs/minix/file.c

@@ -32,7 +32,6 @@ int minix_sync_file(struct file * file, struct dentry *dentry, int datasync)
 	int err;
 
 	err = fsync_inode_buffers(inode);
-	err |= fsync_inode_data_buffers(inode);
 	if (!(inode->i_state & I_DIRTY))
 		return err;
 	if (datasync && !(inode->i_state & I_DIRTY_DATASYNC))

fs/reiserfs/file.c

@@ -86,7 +86,6 @@ static int reiserfs_sync_file(
       BUG ();
 
   n_err = fsync_inode_buffers(p_s_inode) ;
-  n_err |= fsync_inode_data_buffers(p_s_inode);
   reiserfs_commit_for_inode(p_s_inode) ;
   unlock_kernel() ;
   return ( n_err < 0 ) ? -EIO : 0;

fs/sysv/file.c

@@ -38,7 +38,6 @@ int sysv_sync_file(struct file * file, struct dentry *dentry, int datasync)
 	int err;
 
 	err = fsync_inode_buffers(inode);
-	err |= fsync_inode_data_buffers(inode);
 	if (!(inode->i_state & I_DIRTY))
 		return err;
 	if (datasync && !(inode->i_state & I_DIRTY_DATASYNC))

fs/udf/fsync.c

@@ -45,7 +45,6 @@ int udf_fsync_inode(struct inode *inode, int datasync)
 	int err;
 
 	err = fsync_inode_buffers(inode);
-	err |= fsync_inode_data_buffers(inode);
 	if (!(inode->i_state & I_DIRTY))
 		return err;
 	if (datasync && !(inode->i_state & I_DIRTY_DATASYNC))

include/linux/fs.h

@@ -435,7 +435,6 @@ struct inode {
 	struct list_head	i_dentry;
 
 	struct list_head	i_dirty_buffers;   /* uses i_bufferlist_lock */
-	struct list_head	i_dirty_data_buffers;
 	spinlock_t		i_bufferlist_lock;
 
 	unsigned long		i_ino;
@@ -1266,13 +1265,6 @@ buffer_insert_inode_queue(struct buffer_head *bh, struct inode *inode)
 			bh, &inode->i_dirty_buffers);
 }
 
-static inline void
-buffer_insert_inode_data_queue(struct buffer_head *bh, struct inode *inode)
-{
-	buffer_insert_list(&inode->i_bufferlist_lock,
-			bh, &inode->i_dirty_data_buffers);
-}
-
 #define atomic_set_buffer_dirty(bh) test_and_set_bit(BH_Dirty, &(bh)->b_state)
 
 static inline void mark_buffer_async(struct buffer_head * bh, int on)
@@ -1341,11 +1333,6 @@ static inline int fsync_inode_buffers(struct inode *inode)
 	return fsync_buffers_list(&inode->i_bufferlist_lock,
 				&inode->i_dirty_buffers);
 }
-static inline int fsync_inode_data_buffers(struct inode *inode)
-{
-	return fsync_buffers_list(&inode->i_bufferlist_lock,
-				&inode->i_dirty_data_buffers);
-}
 extern int inode_has_buffers(struct inode *);
 extern int filemap_fdatasync(struct address_space *);
 extern int filemap_fdatawait(struct address_space *);

mm/filemap.c

@@ -1083,8 +1083,6 @@ static ssize_t generic_file_direct_IO(int rw, struct file * filp, char * buf, si
 	 * completly asynchronous or performance will go to /dev/null.
 	 */
 	retval = filemap_fdatasync(mapping);
-	if (retval == 0)
-		retval = fsync_inode_data_buffers(inode);
 	if (retval == 0)
 		retval = filemap_fdatawait(mapping);
 	if (retval < 0)