[Linuxarm] Re: [RFCv4 PATCH net-next] net: extend netdev_features_t

在 2021/11/10 6:32, Andrew Lunn 写道:

...

• if ((netdev->features & NETIF_F_HW_TC) > (features & NETIF_F_HW_TC) &&

• if ((netdev_active_features_test_bit(netdev, NETIF_F_HW_TC_BIT) >

•    netdev_features_test_bit(NETIF_F_NTUPLE_BIT, features)) &&
  

Using > is interesting.

will use

if (netdev_active_features_test_bit(netdev, NETIF_F_HW_TC_BIT) && !netdev_features_test_bit(netdev, NETIF_F_HW_TC_BIT))

instead.

But where did NETIF_F_NTUPLE_BIT come from?

Thanks for catching this!

...

• netdev->features |= NETIF_F_HW_VLAN_CTAG_FILTER |

• NETIF_F_HW_VLAN_CTAG_TX | NETIF_F_HW_VLAN_CTAG_RX |
  

• NETIF_F_RXCSUM | NETIF_F_SG | NETIF_F_GSO |
  

• NETIF_F_GRO | NETIF_F_TSO | NETIF_F_TSO6 | NETIF_F_GSO_GRE |
  

• NETIF_F_GSO_GRE_CSUM | NETIF_F_GSO_UDP_TUNNEL |
  

• NETIF_F_SCTP_CRC | NETIF_F_FRAGLIST;
  

• netdev_features_zero(&features);
• netdev_features_set_array(hns3_default_features_array,

• 		  ARRAY_SIZE(hns3_default_features_array),
  

• 		  &features);
  

The original code is netdev->features |= so it is appending these bits. Yet the first thing the new code does is zero features?

   Andrew

.

The features is a local variable, the change for netdev->active_features is later, by calling

netdev_active_features_direct_or(netdev, features);

By the way, have you reveiwed the rest of patch ? Is there anything else need to rework ?

I wonder whether to go ahead.

Thanks!

Jian