Re: [PATCH 1/1] Fixed to BUG_ON to WARN_ON def

[Leon Romanovsky]

On Mon, Dec 12, 2016 at 12:58:59PM +0200, Ozgur Karatas wrote:
> Hello all,
> I think should be use to "WARN_ON" and checkpatch script give to error, I fixed and I think  should don't use "BUG_ON".
> Regards,
>
> Signed-off-by: Ozgur Karatas <okaratas@xxxxxxxxxxxxxx>

NAK, Leon Romanovsky <leonro@xxxxxxxxxxxx>

If we put aside commit message issue, which was pointed to you by Stefan, your
proposed change is incorrect. By chnaging BUG_ONs to be WARN_ONs, you
will left the driver in improper state.

Thanks

> ---
> drivers/net/ethernet/mellanox/mlx4/icm.c |  4 ++--
>
> diff --git a/drivers/net/ethernet/mellanox/mlx4/icm.c b/drivers/net/ethernet/mellanox/mlx4/icm.c
> index 2a9dd46..3fde535 100644
> --- a/drivers/net/ethernet/mellanox/mlx4/icm.c
> +++ b/drivers/net/ethernet/mellanox/mlx4/icm.c
> @@ -119,7 +119,7 @@ static int mlx4_alloc_icm_coherent(struct device *dev, struct scatterlist *mem,
>  		return -ENOMEM;
>
>  	sg_set_buf(mem, buf, PAGE_SIZE << order);
> -	BUG_ON(mem->offset);
> +	WARN_ON(mem->offset);
>  	sg_dma_len(mem) = PAGE_SIZE << order;
>  	return 0;
>  }
> @@ -133,7 +133,7 @@ struct mlx4_icm *mlx4_alloc_icm(struct mlx4_dev *dev, int npages,
>  	int ret;
>
>  	/* We use sg_set_buf for coherent allocs, which assumes low memory */
> -	BUG_ON(coherent && (gfp_mask & __GFP_HIGHMEM));
> +	WARN_ON(coherent && (gfp_mask & __GFP_HIGHMEM));
>
>  	icm = kmalloc_node(sizeof(*icm),
>  			   gfp_mask & ~(__GFP_HIGHMEM | __GFP_NOWARN),
> --
> 2.1.4

Attachment: signature.asc
Description: PGP signature