Re: [PATCH] RDMA/mlx5: Use clamp() in _mlx5r_umr_zap_mkey()

[David Laight]

On Tue, 10 Mar 2026 15:57:28 +0100
Thorsten Blum <thorsten.blum@xxxxxxxxx> wrote:

> Replace min(max()) with clamp(). No functional change.
> 
> Signed-off-by: Thorsten Blum <thorsten.blum@xxxxxxxxx>
> ---
>  drivers/infiniband/hw/mlx5/umr.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/infiniband/hw/mlx5/umr.c b/drivers/infiniband/hw/mlx5/umr.c
> index 4e562e0dd9e1..1a6b0ac5c24d 100644
> --- a/drivers/infiniband/hw/mlx5/umr.c
> +++ b/drivers/infiniband/hw/mlx5/umr.c
> @@ -1013,7 +1013,7 @@ static int _mlx5r_umr_zap_mkey(struct mlx5_ib_mr *mr,
>  		 MLX5_IB_UPD_XLT_ATOMIC;
>  	max_log_size = get_max_log_entity_size_cap(dev, access_mode);
>  	max_page_shift = order_base_2(mr->ibmr.length);
> -	max_page_shift = min(max(max_page_shift, page_shift), max_log_size);
> +	max_page_shift = clamp(max_page_shift, page_shift, max_log_size);

Is page_shift absolutely guaranteed to be less than (or equal to)
max_log_size?
clamp() doesn't guarantee the order of the comparisons.
If I read the code correctly it changed a few years back - on a commit
that said it didn't change it!
I think clamp() currently uses the other order, I looked at proposing to
swap it so that clamp(int_var, 0, unsigned_limit) could be valid (returning
0 for negative values).

	David


>  	/* Count blocks in units of max_page_shift, we will zap exactly this
>  	 * many to make the whole MR non-present.
>  	 * Block size must be aligned to MLX5_UMR_FLEX_ALIGNMENT since it may
>