* be asked to write less pages once, this purely depends on
* implementation. Anyway, we should be careful to avoid deadlocking.
*/
- mutex_lock(&inode->i_mutex);
+ inode_lock(inode);
bytes = ll_direct_rw_pages(env, io, rw, inode, pvec);
- mutex_unlock(&inode->i_mutex);
+ inode_unlock(inode);
cl_io_fini(env, io);
return (bytes == pvec->ldp_size) ? 0 : (int)bytes;
}