we can actually remove the clear_buffer_new() and mark_buffer_dirty() bits because they will be done by block_commit_write() or folio_zero_new_buffers() and they are superfluous and somewhat odd here anyway. And the call to folio_zero_new_buffers() from ext4_block_write_begin() needs to call ext4_journalled_zero_new_buffers() for inodes where data is journalled. Will you try to implement this or should I look into it? Honza -- Jan Kara <jack@xxxxxxxx> SUSE Labs, CR
