Re: [PATCH] bnx2x: fix memory leak in bnx2x_tpa_stop()

[Eric Dumazet]

On Fri, Jun 24, 2022 at 6:12 AM Jianglei Nie <niejianglei2021@xxxxxxx> wrote:
>
> bnx2x_tpa_stop() allocates a memory chunk from new_data with
> bnx2x_frag_alloc(). The new_data should be freed when some errors occur.
> But when "pad + len > fp->rx_buf_size" is true, bnx2x_tpa_stop() returns
> without releasing the new_data, which leads to a memory leak.
>
> We should free the new_data with bnx2x_frag_free() when "pad + len >
> fp->rx_buf_size" is true.
>
> Signed-off-by: Jianglei Nie <niejianglei2021@xxxxxxx>
> ---
>  drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c | 1 +
>  1 file changed, 1 insertion(+)
>
> diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
> index 5729a5ab059d..4cbd3ba5acb9 100644
> --- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
> +++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
> @@ -789,6 +789,7 @@ static void bnx2x_tpa_stop(struct bnx2x *bp, struct bnx2x_fastpath *fp,
>                         BNX2X_ERR("skb_put is about to fail...  pad %d  len %d  rx_buf_size %d\n",
>                                   pad, len, fp->rx_buf_size);
>                         bnx2x_panic();
> +                       bnx2x_frag_free(fp, new_data);

This will crash the host if new_data == NULL

Really, given that BNX2X_STOP_ON_ERROR is not defined, I am not sure
we really care about this ?

>                         return;
>                 }
>  #endif
> --
> 2.25.1
>