Re: [PATCH v8 04/10] iomap: don't get an reference on ZERO_PAGE for direct I/O block zeroing

[John Hubbard]

On 1/23/23 09:30, David Howells wrote:
> From: Christoph Hellwig <hch@xxxxxx>
>
> ZERO_PAGE can't go away, no need to hold an extra reference.

That statement is true, but...

> Signed-off-by: Christoph Hellwig <hch@xxxxxx>
> Signed-off-by: David Howells <dhowells@xxxxxxxxxx>
> ---
>   fs/iomap/direct-io.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/fs/iomap/direct-io.c b/fs/iomap/direct-io.c
> index 9804714b1751..47db4ead1e74 100644
> --- a/fs/iomap/direct-io.c
> +++ b/fs/iomap/direct-io.c
> @@ -202,7 +202,7 @@ static void iomap_dio_zero(const struct iomap_iter *iter, struct iomap_dio *dio,
>   	bio->bi_private = dio;
>   	bio->bi_end_io = iomap_dio_bio_end_io;
>   
> -	get_page(page);
> +	bio_set_flag(bio, BIO_NO_PAGE_REF);

...is it accurate to assume that the entire bio is pointing to the zero
page? I recall working through this area earlier last year, and ended up
just letting the zero page get pinned, and then unpinning upon release,
which is harmless.

I like your approach, if it is possible. I'm just not sure that it's
correct given that bio's can have more than one page.

thanks,
--
John Hubbard
NVIDIA