Re: [PATCH net-next] net/mlx5: check for allocation failure in mlx5_ft_pool_init()

From: Dan Carpenter <dan.carpenter@oracle.com>
To: Leon Romanovsky <leon@kernel.org>
Cc: Saeed Mahameed <saeedm@nvidia.com>,
	Paul Blakey <paulb@nvidia.com>,
	"David S. Miller" <davem@davemloft.net>,
	Jakub Kicinski <kuba@kernel.org>,
	netdev@vger.kernel.org, linux-rdma@vger.kernel.org,
	kernel-janitors@vger.kernel.org
Subject: Re: [PATCH net-next] net/mlx5: check for allocation failure in mlx5_ft_pool_init()
Date: Thu, 3 Jun 2021 18:46:25 +0300	[thread overview]
Message-ID: <20210603154625.GI1955@kadam> (raw)
In-Reply-To: <YLjVRjAyP3UpzgVr@unreal>

On Thu, Jun 03, 2021 at 04:12:38PM +0300, Leon Romanovsky wrote:
> On Thu, Jun 03, 2021 at 03:39:24PM +0300, Dan Carpenter wrote:
> > Add a check for if the kzalloc() fails.
> > 
> > Fixes: 4a98544d1827 ("net/mlx5: Move chains ft pool to be used by all firmware steering")
> > Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
> > ---
> >  drivers/net/ethernet/mellanox/mlx5/core/fs_ft_pool.c | 2 ++
> >  1 file changed, 2 insertions(+)
> > 
> > diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_ft_pool.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_ft_pool.c
> > index 526fbb669142..c14590acc772 100644
> > --- a/drivers/net/ethernet/mellanox/mlx5/core/fs_ft_pool.c
> > +++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_ft_pool.c
> > @@ -27,6 +27,8 @@ int mlx5_ft_pool_init(struct mlx5_core_dev *dev)
> >  	int i;
> >  
> >  	ft_pool = kzalloc(sizeof(*ft_pool), GFP_KERNEL);
> > +	if (!ft_pool)
> > +		return -ENOMEM;
> >  
> >  	for (i = ARRAY_SIZE(FT_POOLS) - 1; i >= 0; i--)
> >  		ft_pool->ft_left[i] = FT_SIZE / FT_POOLS[i];
> 
> 
> Dan thanks for your patch.
> 
> When reviewed your patch, I spotted another error in the patch from the Fixes line.
> 
>   2955         err = mlx5_ft_pool_init(dev);
>   2956         if (err)
>   2957                 return err;
>   2958
>   2959         steering = kzalloc(sizeof(*steering), GFP_KERNEL);
>   2960         if (!steering)
>   2961                 goto err;
>                        ^^^^^^^^ it will return success, while should return ENOMEM.

Smatch prints a static checker warning for this, but I never finished
going through the backlog of old "missing error code" warnings.  At one
point I was down to 38 warnings left but now I see that the backlog is
62 warnings so people are adding new bugs faster than I'm reviewing
them...  :P

I will take care of this tomorrow as a separate patch.  I will just
report or the other 61 warnings and get the backlog cleared out so that
I can start checking these better in the future.

regards,
dan carpenter