Re: [PATCH net-next] net: ethernet: mtk_wed: do not assume offload callbacks are always set

[Simon Horman <horms@kernel.org>]

On Tue, Sep 12, 2023 at 10:26:07AM +0200, Lorenzo Bianconi wrote:
> Check if wlan.offload_enable and wlan.offload_disable callbacks are set
> in mtk_wed_flow_add/mtk_wed_flow_remove since mt7996 will not rely
> on them.
> 
> Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>

Hi Lorenzo,

It's not not a big deal from my perspective, but
I do wonder if these mediatek patches could have been a series.

> ---
>  drivers/net/ethernet/mediatek/mtk_wed.c | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/drivers/net/ethernet/mediatek/mtk_wed.c b/drivers/net/ethernet/mediatek/mtk_wed.c
> index 94376aa2b34c..d8cd59f44401 100644
> --- a/drivers/net/ethernet/mediatek/mtk_wed.c
> +++ b/drivers/net/ethernet/mediatek/mtk_wed.c
> @@ -1718,6 +1718,9 @@ int mtk_wed_flow_add(int index)
>  	if (!hw || !hw->wed_dev)
>  		return -ENODEV;
>  
> +	if (!hw->wed_dev->wlan.offload_enable)
> +		return 0;

A little further down in this function it is assumed that hw->wed_dev may
be NULL, a check made under a lock no less. But it is dereferenced
unconditionally here without a lock. This doesn't seem right one way or
another.

As flagged by Smatch.

> +
>  	if (hw->num_flows) {
>  		hw->num_flows++;
>  		return 0;
> @@ -1747,6 +1750,9 @@ void mtk_wed_flow_remove(int index)
>  	if (!hw)
>  		return;
>  
> +	if (!hw->wed_dev->wlan.offload_disable)
> +		return;
> +
>  	if (--hw->num_flows)
>  		return;
>  
> -- 
> 2.41.0
> 
>