RDMA/mlx5: Use clamp() in _mlx5r_umr_zap_mkey()
Subsystems (2)
PATCH 1/1 RDMA/mlx5: Use clamp() in _mlx5r_umr_zap_mkey()
Replace min(max()) with clamp(). No functional change.
--- drivers/infiniband/hw/mlx5/umr.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/infiniband/hw/mlx5/umr.c +1 -1
@@ -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);/* 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
Quoted 1 line · Thorsten Blum
Replace min(max()) with
clamp(). No functional change.
Are you certain about that?
For the values to match, page_shift must be guaranteed to remain less or equal than max_log_size.
Thanks
Quoted 19 lines
Signed-off-by Thorsten Blum <[email protected]>--- drivers/infiniband/hw/mlx5/umr.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/infiniband/hw/mlx5/umr.c +1
-1@@ -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);/* 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
Quoted 6 lines · Leon Romanovsky
Quoted 1 line · Thorsten Blum
Replace min(max()) with
clamp(). No functional change.Are you certain about that?
For the values to match,page_shiftmust be guaranteed to remain less or equal thanmax_log_size.
Yeah sorry, I didn't realize this conversion isn't straightforward.
I also just realized that clamp() is defined with if/else semantics in linux/minmax.h, but uses min(max()) in tools/include/linux/kernel.h, which seems confusing to me.
Nonetheless, in this case it doesn't seem safe to replace min(max()) with clamp(), and I've learned something today. Thanks for catching it, and please drop this patch.
Quoted 12 lines · On Tue, 10 Mar 2026 15:57:28 +0100 Thorsten Blum <[email protected]>…
Replace min(max()) with
clamp(). No functional change.Signed-off-by Thorsten Blum <[email protected]>--- drivers/infiniband/hw/mlx5/umr.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/infiniband/hw/mlx5/umr.c +0
-0
drivers/infiniband/hw/mlx5/umr.c static int _mlx5r_umr_zap_mkey(struct mlx5_ib_mr *mr, · line 1016
Quoted 5 lines
drivers/infiniband/hw/mlx5/umr.c +1
-1@@ -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
Quoted 4 lines
drivers/infiniband/hw/mlx5/umr.c +0
-0/* 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
Quoted 10 lines · David Laight
Quoted 7 lines · On Tue, 10 Mar 2026 15:57:28 +0100 Thorsten Blum <[email protected]>…
Replace min(max()) with
clamp(). No functional change.Signed-off-by Thorsten Blum <[email protected]>--- drivers/infiniband/hw/mlx5/umr.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
Quoted 13 lines
drivers/infiniband/hw/mlx5/umr.c static int _mlx5r_umr_zap_mkey(struct mlx5_ib_mr *mr, · line 1016
Quoted 10 lines
drivers/infiniband/hw/mlx5/umr.c +1
-1@@ -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_shiftabsolutely guaranteed to be less than (or equal to)max_log_size?
I asked same question here.
https://lore.kernel.org/all/20260310184744.GI12611@unreal/
Thanks