On Tue, Dec 17, 2024 at 06:26:26PM -0800, Luis Chamberlain wrote:
> Now that we can read asynchronously buffer heads from a folio in
> chunks, we can chop up bh_read_iter() with a smaller array size.
> Use an array of 8 to avoid stack growth warnings on systems with
> huge base page sizes.
>
> Reviewed-by: Hannes Reinecke <hare@xxxxxxx>
> Signed-off-by: Luis Chamberlain <mcgrof@xxxxxxxxxx>
> ---
> fs/buffer.c | 15 ++++++++++-----
> 1 file changed, 10 insertions(+), 5 deletions(-)
>
> diff --git a/fs/buffer.c b/fs/buffer.c
> index b8ba72f2f211..bfa9c09b8597 100644
> --- a/fs/buffer.c
> +++ b/fs/buffer.c
> @@ -2415,7 +2415,10 @@ static void bh_read_batch_async(struct folio *folio,
> (__tmp); \
> (__tmp) = bh_next(__tmp, __head))
>
> +#define MAX_BUF_CHUNK 8
> +
> struct bh_iter {
> + int chunk_number;
> sector_t iblock;
> get_block_t *get_block;
> bool any_get_block_error;
Oh... this can be cleaned up even further, chunk_idx and be removed now,
and we can also remove the unused i variable... I'll wait for more
feedback for a v2.
Luis
