direct-io: always call ->end_io if non-NULL

This way we can pass back errors to the file system, and allow for
cleanup required for all direct I/O invocations.

Also allow the ->end_io handlers to return errors on their own, so that
I/O completion errors can be passed on to the callers.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>

fs/dax.c

@@ -267,8 +267,13 @@ ssize_t dax_do_io(struct kiocb *iocb, struct inode *inode,
 	if ((flags & DIO_LOCKING) && iov_iter_rw(iter) == READ)
 		inode_unlock(inode);
 
-	if ((retval > 0) && end_io)
-		end_io(iocb, pos, retval, bh.b_private);
+	if (end_io) {
+		int err;
+
+		err = end_io(iocb, pos, retval, bh.b_private);
+		if (err)
+			retval = err;
+	}
 
 	if (!(flags & DIO_SKIP_DIO_COUNT))
 		inode_dio_end(inode);

fs/direct-io.c

@@ -253,8 +253,13 @@ static ssize_t dio_complete(struct dio *dio, loff_t offset, ssize_t ret,
 	if (ret == 0)
 		ret = transferred;
 
-	if (dio->end_io && dio->result)
-		dio->end_io(dio->iocb, offset, transferred, dio->private);
+	if (dio->end_io) {
+		int err;
+
+		err = dio->end_io(dio->iocb, offset, ret, dio->private);
+		if (err)
+			ret = err;
+	}
 
 	if (!(dio->flags & DIO_SKIP_DIO_COUNT))
 		inode_dio_end(dio->inode);

fs/ext4/inode.c

@@ -3161,14 +3161,17 @@ out:
 }
 #endif
 
-static void ext4_end_io_dio(struct kiocb *iocb, loff_t offset,
+static int ext4_end_io_dio(struct kiocb *iocb, loff_t offset,
 			    ssize_t size, void *private)
 {
         ext4_io_end_t *io_end = iocb->private;
 
+	if (size <= 0)
+		return 0;
+
 	/* if not async direct IO just return */
 	if (!io_end)
-		return;
+		return 0;
 
 	ext_debug("ext4_end_io_dio(): io_end 0x%p "
 		  "for inode %lu, iocb 0x%p, offset %llu, size %zd\n",
@@ -3179,6 +3182,8 @@ static void ext4_end_io_dio(struct kiocb *iocb, loff_t offset,
 	io_end->offset = offset;
 	io_end->size = size;
 	ext4_put_io_end(io_end);
+
+	return 0;
 }
 
 /*

fs/ocfs2/aops.c

@@ -620,7 +620,7 @@ bail:
  * particularly interested in the aio/dio case.  We use the rw_lock DLM lock
  * to protect io on one node from truncation on another.
  */
-static void ocfs2_dio_end_io(struct kiocb *iocb,
+static int ocfs2_dio_end_io(struct kiocb *iocb,
 			     loff_t offset,
 			     ssize_t bytes,
 			     void *private)
@@ -628,6 +628,9 @@ static void ocfs2_dio_end_io(struct kiocb *iocb,
 	struct inode *inode = file_inode(iocb->ki_filp);
 	int level;
 
+	if (bytes <= 0)
+		return 0;
+
 	/* this io's submitter should not have unlocked this before we could */
 	BUG_ON(!ocfs2_iocb_is_rw_locked(iocb));
 
@@ -644,6 +647,8 @@ static void ocfs2_dio_end_io(struct kiocb *iocb,
 		level = ocfs2_iocb_rw_locked_level(iocb);
 		ocfs2_rw_unlock(inode, level);
 	}
+
+	return 0;
 }
 
 static int ocfs2_releasepage(struct page *page, gfp_t wait)

fs/xfs/xfs_aops.c

@@ -1645,7 +1645,7 @@ out_end_io:
  * case the completion can be called in interrupt context, whereas if we have an
  * ioend we will always be called in task context (i.e. from a workqueue).
  */
-STATIC void
+STATIC int
 xfs_end_io_direct_write(
 	struct kiocb		*iocb,
 	loff_t			offset,
@@ -1655,15 +1655,19 @@ xfs_end_io_direct_write(
 	struct inode		*inode = file_inode(iocb->ki_filp);
 	struct xfs_ioend	*ioend = private;
 
+	if (size <= 0)
+		return 0;
+
 	trace_xfs_gbmap_direct_endio(XFS_I(inode), offset, size,
 				     ioend ? ioend->io_type : 0, NULL);
 
 	if (!ioend) {
 		ASSERT(offset + size <= i_size_read(inode));
-		return;
+		return 0;
 	}
 
 	__xfs_end_io_direct_write(inode, ioend, offset, size);
+	return 0;
 }
 
 static inline ssize_t
@@ -1672,10 +1676,7 @@ xfs_vm_do_dio(
 	struct kiocb		*iocb,
 	struct iov_iter		*iter,
 	loff_t			offset,
-	void			(*endio)(struct kiocb	*iocb,
-					 loff_t		offset,
-					 ssize_t	size,
-					 void		*private),
+	dio_iodone_t		endio,
 	int			flags)
 {
 	struct block_device	*bdev;

include/linux/fs.h

@@ -70,7 +70,7 @@ extern int sysctl_protected_hardlinks;
 struct buffer_head;
 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
 			struct buffer_head *bh_result, int create);
-typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
+typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
 			ssize_t bytes, void *private);
 typedef void (dax_iodone_t)(struct buffer_head *bh_map, int uptodate);