[PATCH] RDMA/mlx4: Use bitmap_alloc() when applicable

[PATCH] RDMA/mlx4: Use bitmap_alloc() when applicable

[Christophe JAILLET]

Use 'bitmap_alloc()' to simplify code, improve the semantic and avoid some
open-coded arithmetic in allocator arguments.

Also change the corresponding 'kfree()' into 'bitmap_free()' to keep
consistency.

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
---
 drivers/infiniband/hw/mlx4/main.c | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index 0d2fa3338784..d66ce7694bbe 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -2784,10 +2784,8 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
 		if (err)
 			goto err_counter;
 
-		ibdev->ib_uc_qpns_bitmap =
-			kmalloc_array(BITS_TO_LONGS(ibdev->steer_qpn_count),
-				      sizeof(long),
-				      GFP_KERNEL);
+		ibdev->ib_uc_qpns_bitmap = bitmap_alloc(ibdev->steer_qpn_count,
+							GFP_KERNEL);
 		if (!ibdev->ib_uc_qpns_bitmap)
 			goto err_steer_qp_release;
 
@@ -2875,7 +2873,7 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
 	mlx4_ib_diag_cleanup(ibdev);
 
 err_steer_free_bitmap:
-	kfree(ibdev->ib_uc_qpns_bitmap);
+	bitmap_free(ibdev->ib_uc_qpns_bitmap);
 
 err_steer_qp_release:
 	mlx4_qp_release_range(dev, ibdev->steer_qpn_base,
@@ -2988,7 +2986,7 @@ static void mlx4_ib_remove(struct mlx4_dev *dev, void *ibdev_ptr)
 
 	mlx4_qp_release_range(dev, ibdev->steer_qpn_base,
 			      ibdev->steer_qpn_count);
-	kfree(ibdev->ib_uc_qpns_bitmap);
+	bitmap_free(ibdev->ib_uc_qpns_bitmap);
 
 	iounmap(ibdev->uar_map);
 	for (p = 0; p < ibdev->num_ports; ++p)
-- 
2.30.2

Re: [PATCH] RDMA/mlx4: Use bitmap_alloc() when applicable

[Joe Perches]

On Thu, 2021-11-25 at 20:42 +0100, Christophe JAILLET wrote:
> Use 'bitmap_alloc()' to simplify code, improve the semantic and avoid some
> open-coded arithmetic in allocator arguments.
> 
> Also change the corresponding 'kfree()' into 'bitmap_free()' to keep
> consistency.

Thanks.

> diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
[]
> @@ -2784,10 +2784,8 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
>  		if (err)
>  			goto err_counter;
>  
> -		ibdev->ib_uc_qpns_bitmap =
> -			kmalloc_array(BITS_TO_LONGS(ibdev->steer_qpn_count),
> -				      sizeof(long),
> -				      GFP_KERNEL);
> +		ibdev->ib_uc_qpns_bitmap = bitmap_alloc(ibdev->steer_qpn_count,
> +							GFP_KERNEL);

I wonder if it'd be simpler/smaller to change this to bitmap_zalloc and
remove the bitmap_zero in the if below.

Re: [PATCH] RDMA/mlx4: Use bitmap_alloc() when applicable

[Christophe JAILLET]

Le 25/11/2021 à 20:58, Joe Perches a écrit :
> On Thu, 2021-11-25 at 20:42 +0100, Christophe JAILLET wrote:
>> Use 'bitmap_alloc()' to simplify code, improve the semantic and avoid some
>> open-coded arithmetic in allocator arguments.
>>
>> Also change the corresponding 'kfree()' into 'bitmap_free()' to keep
>> consistency.
> 
> Thanks.
> 
>> diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
> []
>> @@ -2784,10 +2784,8 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
>>   		if (err)
>>   			goto err_counter;
>>   
>> -		ibdev->ib_uc_qpns_bitmap =
>> -			kmalloc_array(BITS_TO_LONGS(ibdev->steer_qpn_count),
>> -				      sizeof(long),
>> -				      GFP_KERNEL);
>> +		ibdev->ib_uc_qpns_bitmap = bitmap_alloc(ibdev->steer_qpn_count,
>> +							GFP_KERNEL);
> 
> I wonder if it'd be simpler/smaller to change this to bitmap_zalloc and
> remove the bitmap_zero in the if below.
> 

I asked myself the same question.
It is easy to see that the bitmap is either cleared or set.
So this is fine enough for me.

Let see if a maintainer has a preference on it.

CJ

Re: [PATCH] RDMA/mlx4: Use bitmap_alloc() when applicable

[Jason Gunthorpe]

On Thu, Nov 25, 2021 at 08:42:51PM +0100, Christophe JAILLET wrote:
> Use 'bitmap_alloc()' to simplify code, improve the semantic and avoid some
> open-coded arithmetic in allocator arguments.
> 
> Also change the corresponding 'kfree()' into 'bitmap_free()' to keep
> consistency.
> 
> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
> ---
>  drivers/infiniband/hw/mlx4/main.c | 10 ++++------
>  1 file changed, 4 insertions(+), 6 deletions(-)

Applied to for-next, thanks

Jason