> -----Original Message-----
> From: Dan Carpenter <dan.carpenter@xxxxxxxxxx>
> Sent: Saturday, August 17, 2024 2:53 AM
> To: David Thompson <davthompson@xxxxxxxxxx>
> Cc: kernel-janitors@xxxxxxxxxxxxxxx
> Subject: [bug report] mlxbf_gige: disable RX filters until RX path initialized
>
> Hello David Thompson,
>
> Commit df934abb185c ("mlxbf_gige: disable RX filters until RX path
> initialized") from Aug 9, 2024 (linux-next), leads to the following Smatch static
> checker warning:
>
> drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige_rx.c:30
> mlxbf_gige_disable_multicast_rx() warn: was expecting a 64 bit value instead of
> '((((1))) << (1))'
> drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige_rx.c:54
> mlxbf_gige_disable_mac_rx_filter() warn: was expecting a 64 bit value instead of
> '(((((1))) << (4)) << index)'
>
> drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige_rx.c
> 24 void mlxbf_gige_disable_multicast_rx(struct mlxbf_gige *priv)
> 25 {
> 26 void __iomem *base = priv->base;
> 27 u64 data;
> 28
> 29 data = readq(base + MLXBF_GIGE_RX_MAC_FILTER_GENERAL);
> --> 30 data &= ~MLXBF_GIGE_RX_MAC_FILTER_EN_MULTICAST;
>
> I don't know what's stored in "data", but it's a type u64. The type of
> MLXBF_GIGE_RX_MAC_FILTER_EN_MULTICAST is unsigned long. This means
> that on a 32bit system, the &= will zero out the top 32 bits as well as the
> MULICAST bit.
> Use BIT_ULL() to fix this bug?
>
> 31 writeq(data, base + MLXBF_GIGE_RX_MAC_FILTER_GENERAL);
> 32 }
>
> regards,
> dan carpenter
Thanks for the heads-up Dan.
I will review your proposed fix and verify with smatch locally.
- Dave
