Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
Merge branch 'netdev_features-start-cleaning-netdev_features_t-up'
Alexander Lobakin says:
====================
netdev_features: start cleaning netdev_features_t up
NETDEV_FEATURE_COUNT is currently 64, which means we can't add any new
features as netdev_features_t is u64.
As per several discussions, instead of converting netdev_features_t to
a bitmap, which would mean A LOT of changes, we can try cleaning up
netdev feature bits.
There's a bunch of bits which don't really mean features, rather device
attributes/properties that can't be changed via Ethtool in any of the
drivers. Such attributes can be moved to netdev private flags without
losing any functionality.
Start converting some read-only netdev features to private flags from
the ones that are most obvious, like lockless Tx, inability to change
network namespace etc. I was able to reduce NETDEV_FEATURE_COUNT from
64 to 60, which mean 4 free slots for new features. There are obviously
more read-only features to convert, such as highDMA, "challenged VLAN",
HSR (4 bits) - this will be done in subsequent series.
Please note that currently netdev features are not uAPI/ABI by any means.
Ethtool passes their names and bits to the userspace separately and there
are no hardcoded names/bits in the userspace, so that new Ethtool could
work on older kernels and vice versa.
This, however, isn't true for Ethtools < 3.4. I haven't changed the bit
positions of the already existing features and instead replaced the freed
bits with stubs. But it's anyway theoretically possible that Ethtools
older than 2011 will break. I hope no currently supported distros supply
such an ancient version.
Shell scripts also most likely won't break since the removed bits were
always read-only, meaning nobody would try touching them from a script.
====================
Link: https://patch.msgid.link/20240829123340.789395-1-aleksander.lobakin@intel.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
+208
-194
+6
-1
Documentation/networking/net_cachelines/net_device.rst
+6
-1
Documentation/networking/net_cachelines/net_device.rst
···
7
7
8
8
Type Name fastpath_tx_access fastpath_rx_access Comments
9
9
..struct ..net_device
10
+
unsigned_long:32 priv_flags read_mostly - __dev_queue_xmit(tx)
11
+
unsigned_long:1 lltx read_mostly - HARD_TX_LOCK,HARD_TX_TRYLOCK,HARD_TX_UNLOCK(tx)
10
12
char name[16] - -
11
13
struct_netdev_name_node* name_node
12
14
struct_dev_ifalias* ifalias
···
25
23
struct adj_list
26
24
unsigned_int flags read_mostly read_mostly __dev_queue_xmit,__dev_xmit_skb,ip6_output,__ip6_finish_output(tx);ip6_rcv_core(rx)
27
25
xdp_features_t xdp_features
28
-
unsigned_long_long priv_flags read_mostly - __dev_queue_xmit(tx)
29
26
struct_net_device_ops* netdev_ops read_mostly - netdev_core_pick_tx,netdev_start_xmit(tx)
30
27
struct_xdp_metadata_ops* xdp_metadata_ops
31
28
int ifindex - read_mostly ip6_rcv_core
···
164
163
bool proto_down
165
164
unsigned:1 wol_enabled
166
165
unsigned:1 threaded - - napi_poll(napi_enable,dev_set_threaded)
166
+
unsigned_long:1 see_all_hwtstamp_requests
167
+
unsigned_long:1 change_proto_down
168
+
unsigned_long:1 netns_local
169
+
unsigned_long:1 fcoe_mtu
167
170
struct_list_head net_notifier_list
168
171
struct_macsec_ops* macsec_ops
169
172
struct_udp_tunnel_nic_info* udp_tunnel_nic_info
-15
Documentation/networking/netdev-features.rst
-15
Documentation/networking/netdev-features.rst
···
139
139
Features contained in NETIF_F_SOFT_FEATURES are features of networking
140
140
stack. Driver should not change behaviour based on them.
141
141
142
-
* LLTX driver (deprecated for hardware drivers)
143
-
144
-
NETIF_F_LLTX is meant to be used by drivers that don't need locking at all,
145
-
e.g. software tunnels.
146
-
147
-
This is also used in a few legacy drivers that implement their
148
-
own locking, don't use it for new (hardware) drivers.
149
-
150
-
* netns-local device
151
-
152
-
NETIF_F_NETNS_LOCAL is set for devices that are not allowed to move between
153
-
network namespaces (e.g. loopback).
154
-
155
-
Don't use it in drivers.
156
-
157
142
* VLAN challenged
158
143
159
144
NETIF_F_VLAN_CHALLENGED should be set for devices which can't cope with VLAN
+2
-2
Documentation/networking/netdevices.rst
+2
-2
Documentation/networking/netdevices.rst
···
258
258
ndo_start_xmit:
259
259
Synchronization: __netif_tx_lock spinlock.
260
260
261
-
When the driver sets NETIF_F_LLTX in dev->features this will be
261
+
When the driver sets dev->lltx this will be
262
262
called without holding netif_tx_lock. In this case the driver
263
263
has to lock by itself when needed.
264
264
The locking there should also properly protect against
265
-
set_rx_mode. WARNING: use of NETIF_F_LLTX is deprecated.
265
+
set_rx_mode. WARNING: use of dev->lltx is deprecated.
266
266
Don't use it for new drivers.
267
267
268
268
Context: Process with BHs disabled or BH (timer),
+2
-2
Documentation/networking/switchdev.rst
+2
-2
Documentation/networking/switchdev.rst
···
137
137
Port Features
138
138
^^^^^^^^^^^^^
139
139
140
-
NETIF_F_NETNS_LOCAL
140
+
dev->netns_local
141
141
142
142
If the switchdev driver (and device) only supports offloading of the default
143
-
network namespace (netns), the driver should set this feature flag to prevent
143
+
network namespace (netns), the driver should set this private flag to prevent
144
144
the port netdev from being moved out of the default netns. A netns-aware
145
145
driver/device would not set this flag and be responsible for partitioning
146
146
hardware to preserve netns containment. This means hardware cannot forward
+2
-2
drivers/net/amt.c
+2
-2
drivers/net/amt.c
···
3098
3098
dev->hard_header_len = 0;
3099
3099
dev->addr_len = 0;
3100
3100
dev->priv_flags |= IFF_NO_QUEUE;
3101
-
dev->features |= NETIF_F_LLTX;
3101
+
dev->lltx = true;
3102
+
dev->netns_local = true;
3102
3103
dev->features |= NETIF_F_GSO_SOFTWARE;
3103
-
dev->features |= NETIF_F_NETNS_LOCAL;
3104
3104
dev->hw_features |= NETIF_F_SG | NETIF_F_HW_CSUM;
3105
3105
dev->hw_features |= NETIF_F_FRAGLIST | NETIF_F_RXCSUM;
3106
3106
dev->hw_features |= NETIF_F_GSO_SOFTWARE;
+1
-1
drivers/net/bareudp.c
+1
-1
drivers/net/bareudp.c
···
553
553
SET_NETDEV_DEVTYPE(dev, &bareudp_type);
554
554
dev->features |= NETIF_F_SG | NETIF_F_HW_CSUM | NETIF_F_FRAGLIST;
555
555
dev->features |= NETIF_F_RXCSUM;
556
-
dev->features |= NETIF_F_LLTX;
557
556
dev->features |= NETIF_F_GSO_SOFTWARE;
558
557
dev->hw_features |= NETIF_F_SG | NETIF_F_HW_CSUM | NETIF_F_FRAGLIST;
559
558
dev->hw_features |= NETIF_F_RXCSUM;
···
565
566
dev->type = ARPHRD_NONE;
566
567
netif_keep_dst(dev);
567
568
dev->priv_flags |= IFF_NO_QUEUE;
569
+
dev->lltx = true;
568
570
dev->flags = IFF_POINTOPOINT | IFF_NOARP | IFF_MULTICAST;
569
571
dev->pcpu_stat_type = NETDEV_PCPU_STAT_TSTATS;
570
572
}
+4
-4
drivers/net/bonding/bond_main.c
+4
-4
drivers/net/bonding/bond_main.c
···
5928
5928
#endif /* CONFIG_XFRM_OFFLOAD */
5929
5929
5930
5930
/* don't acquire bond device's netif_tx_lock when transmitting */
5931
-
bond_dev->features |= NETIF_F_LLTX;
5931
+
bond_dev->lltx = true;
5932
+
5933
+
/* Don't allow bond devices to change network namespaces. */
5934
+
bond_dev->netns_local = true;
5932
5935
5933
5936
/* By default, we declare the bond to be fully
5934
5937
* VLAN hardware accelerated capable. Special
···
5939
5936
* when there are slaves that are not hw accel
5940
5937
* capable
5941
5938
*/
5942
-
5943
-
/* Don't allow bond devices to change network namespaces. */
5944
-
bond_dev->features |= NETIF_F_NETNS_LOCAL;
5945
5939
5946
5940
bond_dev->hw_features = BOND_VLAN_FEATURES |
5947
5941
NETIF_F_HW_VLAN_CTAG_RX |
+2
-1
drivers/net/dummy.c
+2
-1
drivers/net/dummy.c
···
109
109
dev->flags |= IFF_NOARP;
110
110
dev->flags &= ~IFF_MULTICAST;
111
111
dev->priv_flags |= IFF_LIVE_ADDR_CHANGE | IFF_NO_QUEUE;
112
+
dev->lltx = true;
112
113
dev->features |= NETIF_F_SG | NETIF_F_FRAGLIST;
113
114
dev->features |= NETIF_F_GSO_SOFTWARE;
114
-
dev->features |= NETIF_F_HW_CSUM | NETIF_F_HIGHDMA | NETIF_F_LLTX;
115
+
dev->features |= NETIF_F_HW_CSUM | NETIF_F_HIGHDMA;
115
116
dev->features |= NETIF_F_GSO_ENCAP_ALL;
116
117
dev->hw_features |= dev->features;
117
118
dev->hw_enc_features |= dev->features;
+1
-1
drivers/net/ethernet/adi/adin1110.c
+1
-1
drivers/net/ethernet/adi/adin1110.c
···
1599
1599
netdev->netdev_ops = &adin1110_netdev_ops;
1600
1600
netdev->ethtool_ops = &adin1110_ethtool_ops;
1601
1601
netdev->priv_flags |= IFF_UNICAST_FLT;
1602
-
netdev->features |= NETIF_F_NETNS_LOCAL;
1602
+
netdev->netns_local = true;
1603
1603
1604
1604
port_priv->phydev = get_phy_device(priv->mii_bus, i + 1, false);
1605
1605
if (IS_ERR(port_priv->phydev)) {
+2
-1
drivers/net/ethernet/chelsio/cxgb/cxgb2.c
+2
-1
drivers/net/ethernet/chelsio/cxgb/cxgb2.c
···
1034
1034
netdev->hw_features |= NETIF_F_SG | NETIF_F_IP_CSUM |
1035
1035
NETIF_F_RXCSUM;
1036
1036
netdev->features |= NETIF_F_SG | NETIF_F_IP_CSUM |
1037
-
NETIF_F_RXCSUM | NETIF_F_LLTX | NETIF_F_HIGHDMA;
1037
+
NETIF_F_RXCSUM | NETIF_F_HIGHDMA;
1038
+
netdev->lltx = true;
1038
1039
1039
1040
if (vlan_tso_capable(adapter)) {
1040
1041
netdev->features |=
+2
-4
drivers/net/ethernet/chelsio/cxgb4/cxgb4_fcoe.c
+2
-4
drivers/net/ethernet/chelsio/cxgb4/cxgb4_fcoe.c
···
81
81
82
82
netdev->features |= NETIF_F_FCOE_CRC;
83
83
netdev->vlan_features |= NETIF_F_FCOE_CRC;
84
-
netdev->features |= NETIF_F_FCOE_MTU;
85
-
netdev->vlan_features |= NETIF_F_FCOE_MTU;
84
+
netdev->fcoe_mtu = true;
86
85
87
86
netdev_features_change(netdev);
88
87
···
111
112
112
113
netdev->features &= ~NETIF_F_FCOE_CRC;
113
114
netdev->vlan_features &= ~NETIF_F_FCOE_CRC;
114
-
netdev->features &= ~NETIF_F_FCOE_MTU;
115
-
netdev->vlan_features &= ~NETIF_F_FCOE_MTU;
115
+
netdev->fcoe_mtu = false;
116
116
117
117
netdev_features_change(netdev);
118
118
+2
-1
drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
+2
-1
drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
···
229
229
net_dev->max_mtu = dpaa_get_max_mtu();
230
230
231
231
net_dev->hw_features |= (NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM |
232
-
NETIF_F_LLTX | NETIF_F_RXHASH);
232
+
NETIF_F_RXHASH);
233
233
234
234
net_dev->hw_features |= NETIF_F_SG | NETIF_F_HIGHDMA;
235
235
/* The kernels enables GSO automatically, if we declare NETIF_F_SG.
···
239
239
net_dev->features |= NETIF_F_RXCSUM;
240
240
241
241
net_dev->priv_flags |= IFF_LIVE_ADDR_CHANGE;
242
+
net_dev->lltx = true;
242
243
/* we do not want shared skbs on TX */
243
244
net_dev->priv_flags &= ~IFF_TX_SKB_SHARING;
244
245
+2
-1
drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c
+2
-1
drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c
···
4594
4594
4595
4595
net_dev->priv_flags |= supported;
4596
4596
net_dev->priv_flags &= ~not_supported;
4597
+
net_dev->lltx = true;
4597
4598
4598
4599
/* Features */
4599
4600
net_dev->features = NETIF_F_RXCSUM |
4600
4601
NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM |
4601
4602
NETIF_F_SG | NETIF_F_HIGHDMA |
4602
-
NETIF_F_LLTX | NETIF_F_HW_TC | NETIF_F_TSO;
4603
+
NETIF_F_HW_TC | NETIF_F_TSO;
4603
4604
net_dev->gso_max_segs = DPAA2_ETH_ENQUEUE_MAX_FDS;
4604
4605
net_dev->hw_features = net_dev->features;
4605
4606
net_dev->xdp_features = NETDEV_XDP_ACT_BASIC |
+1
-1
drivers/net/ethernet/intel/ixgbe/ixgbe_dcb_nl.c
+1
-1
drivers/net/ethernet/intel/ixgbe/ixgbe_dcb_nl.c
+2
-2
drivers/net/ethernet/intel/ixgbe/ixgbe_fcoe.c
+2
-2
drivers/net/ethernet/intel/ixgbe/ixgbe_fcoe.c
···
858
858
859
859
/* enable FCoE and notify stack */
860
860
adapter->flags |= IXGBE_FLAG_FCOE_ENABLED;
861
-
netdev->features |= NETIF_F_FCOE_MTU;
861
+
netdev->fcoe_mtu = true;
862
862
netdev_features_change(netdev);
863
863
864
864
/* release existing queues and reallocate them */
···
898
898
899
899
/* disable FCoE and notify stack */
900
900
adapter->flags &= ~IXGBE_FLAG_FCOE_ENABLED;
901
-
netdev->features &= ~NETIF_F_FCOE_MTU;
901
+
netdev->fcoe_mtu = false;
902
902
903
903
netdev_features_change(netdev);
904
904
+1
-1
drivers/net/ethernet/intel/ixgbe/ixgbe_lib.c
+1
-1
drivers/net/ethernet/intel/ixgbe/ixgbe_lib.c
···
981
981
set_bit(__IXGBE_RX_CSUM_UDP_ZERO_ERR, &ring->state);
982
982
983
983
#ifdef IXGBE_FCOE
984
-
if (adapter->netdev->features & NETIF_F_FCOE_MTU) {
984
+
if (adapter->netdev->fcoe_mtu) {
985
985
struct ixgbe_ring_feature *f;
986
986
f = &adapter->ring_feature[RING_F_FCOE];
987
987
if ((rxr_idx >= f->offset) &&
+4
-7
drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
+4
-7
drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
···
5079
5079
netif_set_tso_max_size(adapter->netdev, 32768);
5080
5080
5081
5081
#ifdef IXGBE_FCOE
5082
-
if (adapter->netdev->features & NETIF_F_FCOE_MTU)
5082
+
if (adapter->netdev->fcoe_mtu)
5083
5083
max_frame = max(max_frame, IXGBE_FCOE_JUMBO_FRAME_SIZE);
5084
5084
#endif
5085
5085
···
5136
5136
5137
5137
#ifdef IXGBE_FCOE
5138
5138
/* FCoE traffic class uses FCOE jumbo frames */
5139
-
if ((dev->features & NETIF_F_FCOE_MTU) &&
5140
-
(tc < IXGBE_FCOE_JUMBO_FRAME_SIZE) &&
5139
+
if (dev->fcoe_mtu && tc < IXGBE_FCOE_JUMBO_FRAME_SIZE &&
5141
5140
(pb == ixgbe_fcoe_get_tc(adapter)))
5142
5141
tc = IXGBE_FCOE_JUMBO_FRAME_SIZE;
5143
5142
#endif
···
5196
5197
5197
5198
#ifdef IXGBE_FCOE
5198
5199
/* FCoE traffic class uses FCOE jumbo frames */
5199
-
if ((dev->features & NETIF_F_FCOE_MTU) &&
5200
-
(tc < IXGBE_FCOE_JUMBO_FRAME_SIZE) &&
5200
+
if (dev->fcoe_mtu && tc < IXGBE_FCOE_JUMBO_FRAME_SIZE &&
5201
5201
(pb == netdev_get_prio_tc_map(dev, adapter->fcoe.up)))
5202
5202
tc = IXGBE_FCOE_JUMBO_FRAME_SIZE;
5203
5203
#endif
···
11094
11096
NETIF_F_FCOE_CRC;
11095
11097
11096
11098
netdev->vlan_features |= NETIF_F_FSO |
11097
-
NETIF_F_FCOE_CRC |
11098
-
NETIF_F_FCOE_MTU;
11099
+
NETIF_F_FCOE_CRC;
11099
11100
}
11100
11101
#endif /* IXGBE_FCOE */
11101
11102
if (adapter->flags2 & IXGBE_FLAG2_RSC_CAPABLE)
+2
-2
drivers/net/ethernet/intel/ixgbe/ixgbe_sriov.c
+2
-2
drivers/net/ethernet/intel/ixgbe/ixgbe_sriov.c
···
495
495
int err = 0;
496
496
497
497
#ifdef CONFIG_FCOE
498
-
if (dev->features & NETIF_F_FCOE_MTU)
498
+
if (dev->fcoe_mtu)
499
499
pf_max_frame = max_t(int, pf_max_frame,
500
500
IXGBE_FCOE_JUMBO_FRAME_SIZE);
501
501
···
857
857
int pf_max_frame = dev->mtu + ETH_HLEN;
858
858
859
859
#if IS_ENABLED(CONFIG_FCOE)
860
-
if (dev->features & NETIF_F_FCOE_MTU)
860
+
if (dev->fcoe_mtu)
861
861
pf_max_frame = max_t(int, pf_max_frame,
862
862
IXGBE_FCOE_JUMBO_FRAME_SIZE);
863
863
#endif /* CONFIG_FCOE */
+2
-1
drivers/net/ethernet/marvell/prestera/prestera_main.c
+2
-1
drivers/net/ethernet/marvell/prestera/prestera_main.c
···
633
633
if (err)
634
634
goto err_dl_port_register;
635
635
636
-
dev->features |= NETIF_F_NETNS_LOCAL | NETIF_F_HW_TC;
636
+
dev->features |= NETIF_F_HW_TC;
637
+
dev->netns_local = true;
637
638
dev->netdev_ops = &prestera_netdev_ops;
638
639
dev->ethtool_ops = &prestera_ethtool_ops;
639
640
SET_NETDEV_DEV(dev, sw->dev->dev);
+2
-2
drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+2
-2
drivers/net/ethernet/mellanox/mlx5/core/en_main.c
···
4414
4414
4415
4415
if (mlx5e_is_uplink_rep(priv)) {
4416
4416
features = mlx5e_fix_uplink_rep_features(netdev, features);
4417
-
features |= NETIF_F_NETNS_LOCAL;
4417
+
netdev->netns_local = true;
4418
4418
} else {
4419
-
features &= ~NETIF_F_NETNS_LOCAL;
4419
+
netdev->netns_local = false;
4420
4420
}
4421
4421
4422
4422
mutex_unlock(&priv->state_lock);
+2
-1
drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
+2
-1
drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
+4
-2
drivers/net/ethernet/mellanox/mlxsw/spectrum.c
+4
-2
drivers/net/ethernet/mellanox/mlxsw/spectrum.c
···
1676
1676
1677
1677
netif_carrier_off(dev);
1678
1678
1679
-
dev->features |= NETIF_F_NETNS_LOCAL | NETIF_F_LLTX | NETIF_F_SG |
1680
-
NETIF_F_HW_VLAN_CTAG_FILTER | NETIF_F_HW_TC;
1679
+
dev->features |= NETIF_F_SG | NETIF_F_HW_VLAN_CTAG_FILTER |
1680
+
NETIF_F_HW_TC;
1681
1681
dev->hw_features |= NETIF_F_HW_TC | NETIF_F_LOOPBACK;
1682
+
dev->lltx = true;
1683
+
dev->netns_local = true;
1682
1684
1683
1685
dev->min_mtu = ETH_MIN_MTU;
1684
1686
dev->max_mtu = MLXSW_PORT_MAX_MTU - MLXSW_PORT_ETH_FRAME_HDR;
+1
-1
drivers/net/ethernet/microchip/lan966x/lan966x_main.c
+1
-1
drivers/net/ethernet/microchip/lan966x/lan966x_main.c
···
816
816
NETIF_F_HW_VLAN_STAG_TX |
817
817
NETIF_F_HW_TC;
818
818
dev->hw_features |= NETIF_F_HW_TC;
819
-
dev->priv_flags |= IFF_SEE_ALL_HWTSTAMP_REQUESTS;
819
+
dev->see_all_hwtstamp_requests = true;
820
820
dev->needed_headroom = IFH_LEN_BYTES;
821
821
822
822
eth_hw_addr_gen(dev, lan966x->base_mac, p + 1);
+1
-2
drivers/net/ethernet/netronome/nfp/nfp_net_repr.c
+1
-2
drivers/net/ethernet/netronome/nfp/nfp_net_repr.c
···
248
248
249
249
features = netdev_intersect_features(features, lower_features);
250
250
features |= old_features & (NETIF_F_SOFT_FEATURES | NETIF_F_HW_TC);
251
-
features |= NETIF_F_LLTX;
252
251
253
252
return features;
254
253
}
···
385
386
netif_set_tso_max_segs(netdev, NFP_NET_LSO_MAX_SEGS);
386
387
387
388
netdev->priv_flags |= IFF_NO_QUEUE | IFF_DISABLE_NETPOLL;
388
-
netdev->features |= NETIF_F_LLTX;
389
+
netdev->lltx = true;
389
390
390
391
if (nfp_app_has_tc(app)) {
391
392
netdev->features |= NETIF_F_HW_TC;
+3
-2
drivers/net/ethernet/pasemi/pasemi_mac.c
+3
-2
drivers/net/ethernet/pasemi/pasemi_mac.c
···
1699
1699
1700
1700
netif_napi_add(dev, &mac->napi, pasemi_mac_poll);
1701
1701
1702
-
dev->features = NETIF_F_IP_CSUM | NETIF_F_LLTX | NETIF_F_SG |
1703
-
NETIF_F_HIGHDMA | NETIF_F_GSO;
1702
+
dev->features = NETIF_F_IP_CSUM | NETIF_F_SG | NETIF_F_HIGHDMA |
1703
+
NETIF_F_GSO;
1704
+
dev->lltx = true;
1704
1705
1705
1706
mac->dma_pdev = pci_get_device(PCI_VENDOR_ID_PASEMI, 0xa007, NULL);
1706
1707
if (!mac->dma_pdev) {
+1
-1
drivers/net/ethernet/qualcomm/rmnet/rmnet_vnd.c
+1
-1
drivers/net/ethernet/qualcomm/rmnet/rmnet_vnd.c
···
286
286
rmnet_dev->needs_free_netdev = true;
287
287
rmnet_dev->ethtool_ops = &rmnet_ethtool_ops;
288
288
289
-
rmnet_dev->features |= NETIF_F_LLTX;
289
+
rmnet_dev->lltx = true;
290
290
291
291
/* This perm addr will be used as interface identifier by IPv6 */
292
292
rmnet_dev->addr_assign_type = NET_ADDR_RANDOM;
+2
-1
drivers/net/ethernet/rocker/rocker_main.c
+2
-1
drivers/net/ethernet/rocker/rocker_main.c
···
2575
2575
netif_napi_add(dev, &rocker_port->napi_rx, rocker_port_poll_rx);
2576
2576
rocker_carrier_init(rocker_port);
2577
2577
2578
-
dev->features |= NETIF_F_NETNS_LOCAL | NETIF_F_SG;
2578
+
dev->features |= NETIF_F_SG;
2579
+
dev->netns_local = true;
2579
2580
2580
2581
/* MTU range: 68 - 9000 */
2581
2582
dev->min_mtu = ROCKER_PORT_MIN_MTU;
+2
-2
drivers/net/ethernet/sfc/ef100_rep.c
+2
-2
drivers/net/ethernet/sfc/ef100_rep.c
···
233
233
net_dev->ethtool_ops = &efx_ef100_rep_ethtool_ops;
234
234
net_dev->min_mtu = EFX_MIN_MTU;
235
235
net_dev->max_mtu = EFX_MAX_MTU;
236
-
net_dev->features |= NETIF_F_LLTX;
237
-
net_dev->hw_features |= NETIF_F_LLTX;
236
+
net_dev->lltx = true;
237
+
238
238
return efv;
239
239
fail1:
240
240
free_netdev(net_dev);
+2
-2
drivers/net/ethernet/tehuti/tehuti.c
+2
-2
drivers/net/ethernet/tehuti/tehuti.c
···
1671
1671
1672
1672
#endif
1673
1673
#ifdef BDX_LLTX
1674
-
netif_trans_update(ndev); /* NETIF_F_LLTX driver :( */
1674
+
netif_trans_update(ndev); /* dev->lltx driver :( */
1675
1675
#endif
1676
1676
ndev->stats.tx_packets++;
1677
1677
ndev->stats.tx_bytes += skb->len;
···
2019
2019
* set multicast list callback has to use priv->tx_lock.
2020
2020
*/
2021
2021
#ifdef BDX_LLTX
2022
-
ndev->features |= NETIF_F_LLTX;
2022
+
ndev->lltx = true;
2023
2023
#endif
2024
2024
/* MTU range: 60 - 16384 */
2025
2025
ndev->min_mtu = ETH_ZLEN;
+1
-1
drivers/net/ethernet/tehuti/tehuti.h
+1
-1
drivers/net/ethernet/tehuti/tehuti.h
+2
-1
drivers/net/ethernet/ti/cpsw_new.c
+2
-1
drivers/net/ethernet/ti/cpsw_new.c
···
1407
1407
cpsw->slaves[i].ndev = ndev;
1408
1408
1409
1409
ndev->features |= NETIF_F_HW_VLAN_CTAG_FILTER |
1410
-
NETIF_F_HW_VLAN_CTAG_RX | NETIF_F_NETNS_LOCAL | NETIF_F_HW_TC;
1410
+
NETIF_F_HW_VLAN_CTAG_RX | NETIF_F_HW_TC;
1411
+
ndev->netns_local = true;
1411
1412
1412
1413
ndev->xdp_features = NETDEV_XDP_ACT_BASIC |
1413
1414
NETDEV_XDP_ACT_REDIRECT |
+2
-1
drivers/net/ethernet/toshiba/spider_net.c
+2
-1
drivers/net/ethernet/toshiba/spider_net.c
···
2277
2277
netdev->hw_features = NETIF_F_RXCSUM | NETIF_F_IP_CSUM;
2278
2278
if (SPIDER_NET_RX_CSUM_DEFAULT)
2279
2279
netdev->features |= NETIF_F_RXCSUM;
2280
-
netdev->features |= NETIF_F_IP_CSUM | NETIF_F_LLTX;
2280
+
netdev->features |= NETIF_F_IP_CSUM;
2281
2281
/* some time: NETIF_F_HW_VLAN_CTAG_TX | NETIF_F_HW_VLAN_CTAG_RX |
2282
2282
* NETIF_F_HW_VLAN_CTAG_FILTER
2283
2283
*/
2284
+
netdev->lltx = true;
2284
2285
2285
2286
/* MTU range: 64 - 2294 */
2286
2287
netdev->min_mtu = SPIDER_NET_MIN_MTU;
+1
-1
drivers/net/geneve.c
+1
-1
drivers/net/geneve.c
···
1194
1194
1195
1195
SET_NETDEV_DEVTYPE(dev, &geneve_type);
1196
1196
1197
-
dev->features |= NETIF_F_LLTX;
1198
1197
dev->features |= NETIF_F_SG | NETIF_F_HW_CSUM | NETIF_F_FRAGLIST;
1199
1198
dev->features |= NETIF_F_RXCSUM;
1200
1199
dev->features |= NETIF_F_GSO_SOFTWARE;
···
1214
1215
netif_keep_dst(dev);
1215
1216
dev->priv_flags &= ~IFF_TX_SKB_SHARING;
1216
1217
dev->priv_flags |= IFF_LIVE_ADDR_CHANGE | IFF_NO_QUEUE;
1218
+
dev->lltx = true;
1217
1219
eth_hw_addr_random(dev);
1218
1220
}
1219
1221
+1
-1
drivers/net/gtp.c
+1
-1
drivers/net/gtp.c
+1
-1
drivers/net/hamradio/bpqether.c
+1
-1
drivers/net/hamradio/bpqether.c
+2
-1
drivers/net/ipvlan/ipvlan_main.c
+2
-1
drivers/net/ipvlan/ipvlan_main.c
···
114
114
NETIF_F_GSO_ROBUST | NETIF_F_GSO_SOFTWARE | NETIF_F_GSO_ENCAP_ALL)
115
115
116
116
#define IPVLAN_ALWAYS_ON \
117
-
(IPVLAN_ALWAYS_ON_OFLOADS | NETIF_F_LLTX | NETIF_F_VLAN_CHALLENGED)
117
+
(IPVLAN_ALWAYS_ON_OFLOADS | NETIF_F_VLAN_CHALLENGED)
118
118
119
119
#define IPVLAN_FEATURES \
120
120
(NETIF_F_SG | NETIF_F_HW_CSUM | NETIF_F_HIGHDMA | NETIF_F_FRAGLIST | \
···
141
141
dev->vlan_features = phy_dev->vlan_features & IPVLAN_FEATURES;
142
142
dev->vlan_features |= IPVLAN_ALWAYS_ON_OFLOADS;
143
143
dev->hw_enc_features |= dev->features;
144
+
dev->lltx = true;
144
145
netif_inherit_tso_max(dev, phy_dev);
145
146
dev->hard_header_len = phy_dev->hard_header_len;
146
147
+2
-2
drivers/net/loopback.c
+2
-2
drivers/net/loopback.c
···
171
171
dev->type = ARPHRD_LOOPBACK; /* 0x0001*/
172
172
dev->flags = IFF_LOOPBACK;
173
173
dev->priv_flags |= IFF_LIVE_ADDR_CHANGE | IFF_NO_QUEUE;
174
+
dev->lltx = true;
175
+
dev->netns_local = true;
174
176
netif_keep_dst(dev);
175
177
dev->hw_features = NETIF_F_GSO_SOFTWARE;
176
178
dev->features = NETIF_F_SG | NETIF_F_FRAGLIST
···
181
179
| NETIF_F_RXCSUM
182
180
| NETIF_F_SCTP_CRC
183
181
| NETIF_F_HIGHDMA
184
-
| NETIF_F_LLTX
185
-
| NETIF_F_NETNS_LOCAL
186
182
| NETIF_F_VLAN_CHALLENGED
187
183
| NETIF_F_LOOPBACK;
188
184
dev->ethtool_ops = eth_ops;
+2
-2
drivers/net/macsec.c
+2
-2
drivers/net/macsec.c
···
3550
3550
return err;
3551
3551
3552
3552
dev->features = real_dev->features & MACSEC_FEATURES;
3553
-
dev->features |= NETIF_F_LLTX | NETIF_F_GSO_SOFTWARE;
3553
+
dev->features |= NETIF_F_GSO_SOFTWARE;
3554
+
dev->lltx = true;
3554
3555
dev->pcpu_stat_type = NETDEV_PCPU_STAT_TSTATS;
3555
3556
3556
3557
macsec_set_head_tail_room(dev);
···
3582
3581
3583
3582
features &= (real_dev->features & MACSEC_FEATURES) |
3584
3583
NETIF_F_GSO_SOFTWARE | NETIF_F_SOFT_FEATURES;
3585
-
features |= NETIF_F_LLTX;
3586
3584
3587
3585
return features;
3588
3586
}
+4
-2
drivers/net/macvlan.c
+4
-2
drivers/net/macvlan.c
···
900
900
(NETIF_F_SG | NETIF_F_HW_CSUM | NETIF_F_GSO_SOFTWARE | \
901
901
NETIF_F_GSO_ROBUST | NETIF_F_GSO_ENCAP_ALL)
902
902
903
-
#define ALWAYS_ON_FEATURES (ALWAYS_ON_OFFLOADS | NETIF_F_LLTX)
903
+
#define ALWAYS_ON_FEATURES ALWAYS_ON_OFFLOADS
904
904
905
905
#define MACVLAN_FEATURES \
906
906
(NETIF_F_SG | NETIF_F_HW_CSUM | NETIF_F_HIGHDMA | NETIF_F_FRAGLIST | \
···
932
932
dev->vlan_features = lowerdev->vlan_features & MACVLAN_FEATURES;
933
933
dev->vlan_features |= ALWAYS_ON_OFFLOADS;
934
934
dev->hw_enc_features |= dev->features;
935
+
dev->lltx = true;
935
936
netif_inherit_tso_max(dev, lowerdev);
936
937
dev->hard_header_len = lowerdev->hard_header_len;
937
938
macvlan_set_lockdep_class(dev);
···
1214
1213
dev->max_mtu = ETH_MAX_MTU;
1215
1214
dev->priv_flags &= ~IFF_TX_SKB_SHARING;
1216
1215
netif_keep_dst(dev);
1217
-
dev->priv_flags |= IFF_UNICAST_FLT | IFF_CHANGE_PROTO_DOWN;
1216
+
dev->priv_flags |= IFF_UNICAST_FLT;
1217
+
dev->change_proto_down = true;
1218
1218
dev->netdev_ops = &macvlan_netdev_ops;
1219
1219
dev->needs_free_netdev = true;
1220
1220
dev->priv_destructor = macvlan_dev_free;
+2
-2
drivers/net/net_failover.c
+2
-2
drivers/net/net_failover.c
···
731
731
IFF_TX_SKB_SHARING);
732
732
733
733
/* don't acquire failover netdev's netif_tx_lock when transmitting */
734
-
failover_dev->features |= NETIF_F_LLTX;
734
+
failover_dev->lltx = true;
735
735
736
736
/* Don't allow failover devices to change network namespaces. */
737
-
failover_dev->features |= NETIF_F_NETNS_LOCAL;
737
+
failover_dev->netns_local = true;
738
738
739
739
failover_dev->hw_features = FAILOVER_VLAN_FEATURES |
740
740
NETIF_F_HW_VLAN_CTAG_TX |
+2
-1
drivers/net/netkit.c
+2
-1
drivers/net/netkit.c
···
255
255
dev->priv_flags |= IFF_LIVE_ADDR_CHANGE;
256
256
dev->priv_flags |= IFF_PHONY_HEADROOM;
257
257
dev->priv_flags |= IFF_NO_QUEUE;
258
+
dev->lltx = true;
258
259
259
260
dev->ethtool_ops = &netkit_ethtool_ops;
260
261
dev->netdev_ops = &netkit_netdev_ops;
261
262
262
-
dev->features |= netkit_features | NETIF_F_LLTX;
263
+
dev->features |= netkit_features;
263
264
dev->hw_features = netkit_features;
264
265
dev->hw_enc_features = netkit_features;
265
266
dev->mpls_features = NETIF_F_HW_CSUM | NETIF_F_GSO_SOFTWARE;
+2
-2
drivers/net/nlmon.c
+2
-2
drivers/net/nlmon.c
···
63
63
{
64
64
dev->type = ARPHRD_NETLINK;
65
65
dev->priv_flags |= IFF_NO_QUEUE;
66
+
dev->lltx = true;
66
67
67
68
dev->netdev_ops = &nlmon_ops;
68
69
dev->ethtool_ops = &nlmon_ethtool_ops;
69
70
dev->needs_free_netdev = true;
70
71
71
-
dev->features = NETIF_F_SG | NETIF_F_FRAGLIST |
72
-
NETIF_F_HIGHDMA | NETIF_F_LLTX;
72
+
dev->features = NETIF_F_SG | NETIF_F_FRAGLIST | NETIF_F_HIGHDMA;
73
73
dev->flags = IFF_NOARP;
74
74
dev->pcpu_stat_type = NETDEV_PCPU_STAT_LSTATS;
75
75
+1
-1
drivers/net/ppp/ppp_generic.c
+1
-1
drivers/net/ppp/ppp_generic.c
+1
-1
drivers/net/rionet.c
+1
-1
drivers/net/rionet.c
+4
-4
drivers/net/team/team_core.c
+4
-4
drivers/net/team/team_core.c
···
2189
2189
* Let this up to underlay drivers.
2190
2190
*/
2191
2191
dev->priv_flags |= IFF_UNICAST_FLT | IFF_LIVE_ADDR_CHANGE;
2192
-
2193
-
dev->features |= NETIF_F_LLTX;
2194
-
dev->features |= NETIF_F_GRO;
2192
+
dev->lltx = true;
2195
2193
2196
2194
/* Don't allow team devices to change network namespaces. */
2197
-
dev->features |= NETIF_F_NETNS_LOCAL;
2195
+
dev->netns_local = true;
2196
+
2197
+
dev->features |= NETIF_F_GRO;
2198
2198
2199
2199
dev->hw_features = TEAM_VLAN_FEATURES |
2200
2200
NETIF_F_HW_VLAN_CTAG_RX |
+3
-2
drivers/net/tun.c
+3
-2
drivers/net/tun.c
···
990
990
dev->hw_features = NETIF_F_SG | NETIF_F_FRAGLIST |
991
991
TUN_USER_FEATURES | NETIF_F_HW_VLAN_CTAG_TX |
992
992
NETIF_F_HW_VLAN_STAG_TX;
993
-
dev->features = dev->hw_features | NETIF_F_LLTX;
993
+
dev->features = dev->hw_features;
994
994
dev->vlan_features = dev->features &
995
995
~(NETIF_F_HW_VLAN_CTAG_TX |
996
996
NETIF_F_HW_VLAN_STAG_TX);
997
+
dev->lltx = true;
997
998
998
999
tun->flags = (tun->flags & ~TUN_FEATURES) |
999
1000
(ifr->ifr_flags & TUN_FEATURES);
···
1130
1129
goto drop;
1131
1130
}
1132
1131
1133
-
/* NETIF_F_LLTX requires to do our own update of trans_start */
1132
+
/* dev->lltx requires to do our own update of trans_start */
1134
1133
queue = netdev_get_tx_queue(dev, txq);
1135
1134
txq_trans_cond_update(queue);
1136
1135
+1
-1
drivers/net/veth.c
+1
-1
drivers/net/veth.c
···
1697
1697
dev->priv_flags |= IFF_NO_QUEUE;
1698
1698
dev->priv_flags |= IFF_PHONY_HEADROOM;
1699
1699
dev->priv_flags |= IFF_DISABLE_NETPOLL;
1700
+
dev->lltx = true;
1700
1701
1701
1702
dev->netdev_ops = &veth_netdev_ops;
1702
1703
dev->xdp_metadata_ops = &veth_xdp_metadata_ops;
1703
1704
dev->ethtool_ops = &veth_ethtool_ops;
1704
-
dev->features |= NETIF_F_LLTX;
1705
1705
dev->features |= VETH_FEATURES;
1706
1706
dev->vlan_features = dev->features &
1707
1707
~(NETIF_F_HW_VLAN_CTAG_TX |
+2
-2
drivers/net/vrf.c
+2
-2
drivers/net/vrf.c
···
1635
1635
eth_hw_addr_random(dev);
1636
1636
1637
1637
/* don't acquire vrf device's netif_tx_lock when transmitting */
1638
-
dev->features |= NETIF_F_LLTX;
1638
+
dev->lltx = true;
1639
1639
1640
1640
/* don't allow vrf devices to change network namespaces. */
1641
-
dev->features |= NETIF_F_NETNS_LOCAL;
1641
+
dev->netns_local = true;
1642
1642
1643
1643
/* does not make sense for a VLAN to be added to a vrf device */
1644
1644
dev->features |= NETIF_F_VLAN_CHALLENGED;
+2
-2
drivers/net/vsockmon.c
+2
-2
drivers/net/vsockmon.c
···
83
83
{
84
84
dev->type = ARPHRD_VSOCKMON;
85
85
dev->priv_flags |= IFF_NO_QUEUE;
86
+
dev->lltx = true;
86
87
87
88
dev->netdev_ops = &vsockmon_ops;
88
89
dev->ethtool_ops = &vsockmon_ethtool_ops;
89
90
dev->needs_free_netdev = true;
90
91
91
-
dev->features = NETIF_F_SG | NETIF_F_FRAGLIST |
92
-
NETIF_F_HIGHDMA | NETIF_F_LLTX;
92
+
dev->features = NETIF_F_SG | NETIF_F_FRAGLIST | NETIF_F_HIGHDMA;
93
93
94
94
dev->flags = IFF_NOARP;
95
95
+3
-2
drivers/net/vxlan/vxlan_core.c
+3
-2
drivers/net/vxlan/vxlan_core.c
···
3321
3321
dev->needs_free_netdev = true;
3322
3322
SET_NETDEV_DEVTYPE(dev, &vxlan_type);
3323
3323
3324
-
dev->features |= NETIF_F_LLTX;
3325
3324
dev->features |= NETIF_F_SG | NETIF_F_HW_CSUM | NETIF_F_FRAGLIST;
3326
3325
dev->features |= NETIF_F_RXCSUM;
3327
3326
dev->features |= NETIF_F_GSO_SOFTWARE;
···
3330
3331
dev->hw_features |= NETIF_F_RXCSUM;
3331
3332
dev->hw_features |= NETIF_F_GSO_SOFTWARE;
3332
3333
netif_keep_dst(dev);
3333
-
dev->priv_flags |= IFF_NO_QUEUE | IFF_CHANGE_PROTO_DOWN;
3334
+
dev->priv_flags |= IFF_NO_QUEUE;
3335
+
dev->change_proto_down = true;
3336
+
dev->lltx = true;
3334
3337
3335
3338
/* MTU range: 68 - 65535 */
3336
3339
dev->min_mtu = ETH_MIN_MTU;
+1
-1
drivers/net/wireguard/device.c
+1
-1
drivers/net/wireguard/device.c
···
289
289
dev->type = ARPHRD_NONE;
290
290
dev->flags = IFF_POINTOPOINT | IFF_NOARP;
291
291
dev->priv_flags |= IFF_NO_QUEUE;
292
-
dev->features |= NETIF_F_LLTX;
292
+
dev->lltx = true;
293
293
dev->features |= WG_NETDEV_FEATURES;
294
294
dev->hw_features |= WG_NETDEV_FEATURES;
295
295
dev->hw_enc_features |= WG_NETDEV_FEATURES;
+2
-2
drivers/scsi/fcoe/fcoe.c
+2
-2
drivers/scsi/fcoe/fcoe.c
···
722
722
* will return 0, so do this first.
723
723
*/
724
724
mfs = netdev->mtu;
725
-
if (netdev->features & NETIF_F_FCOE_MTU) {
725
+
if (netdev->fcoe_mtu) {
726
726
mfs = FCOE_MTU;
727
727
FCOE_NETDEV_DBG(netdev, "Supports FCOE_MTU of %d bytes\n", mfs);
728
728
}
···
1863
1863
case NETDEV_CHANGE:
1864
1864
break;
1865
1865
case NETDEV_CHANGEMTU:
1866
-
if (netdev->features & NETIF_F_FCOE_MTU)
1866
+
if (netdev->fcoe_mtu)
1867
1867
break;
1868
1868
mfs = netdev->mtu - (sizeof(struct fcoe_hdr) +
1869
1869
sizeof(struct fcoe_crc_eof));
+1
-1
drivers/staging/octeon/ethernet.c
+1
-1
drivers/staging/octeon/ethernet.c
+4
-12
include/linux/netdev_features.h
+4
-12
include/linux/netdev_features.h
···
24
24
NETIF_F_HW_VLAN_CTAG_FILTER_BIT,/* Receive filtering on VLAN CTAGs */
25
25
NETIF_F_VLAN_CHALLENGED_BIT, /* Device cannot handle VLAN packets */
26
26
NETIF_F_GSO_BIT, /* Enable software GSO. */
27
-
NETIF_F_LLTX_BIT, /* LockLess TX - deprecated. Please */
28
-
/* do not use LLTX in new drivers */
29
-
NETIF_F_NETNS_LOCAL_BIT, /* Does not change network namespaces */
27
+
__UNUSED_NETIF_F_12,
28
+
__UNUSED_NETIF_F_13,
30
29
NETIF_F_GRO_BIT, /* Generic receive offload */
31
30
NETIF_F_LRO_BIT, /* large receive offload */
32
31
···
58
59
59
60
NETIF_F_FCOE_CRC_BIT, /* FCoE CRC32 */
60
61
NETIF_F_SCTP_CRC_BIT, /* SCTP checksum offload */
61
-
NETIF_F_FCOE_MTU_BIT, /* Supports max FCoE MTU, 2158 bytes*/
62
+
__UNUSED_NETIF_F_37,
62
63
NETIF_F_NTUPLE_BIT, /* N-tuple filters supported */
63
64
NETIF_F_RXHASH_BIT, /* Receive hashing offload */
64
65
NETIF_F_RXCSUM_BIT, /* Receive checksumming offload */
···
105
106
#define __NETIF_F(name) __NETIF_F_BIT(NETIF_F_##name##_BIT)
106
107
107
108
#define NETIF_F_FCOE_CRC __NETIF_F(FCOE_CRC)
108
-
#define NETIF_F_FCOE_MTU __NETIF_F(FCOE_MTU)
109
109
#define NETIF_F_FRAGLIST __NETIF_F(FRAGLIST)
110
110
#define NETIF_F_FSO __NETIF_F(FSO)
111
111
#define NETIF_F_GRO __NETIF_F(GRO)
···
118
120
#define NETIF_F_HW_VLAN_CTAG_TX __NETIF_F(HW_VLAN_CTAG_TX)
119
121
#define NETIF_F_IP_CSUM __NETIF_F(IP_CSUM)
120
122
#define NETIF_F_IPV6_CSUM __NETIF_F(IPV6_CSUM)
121
-
#define NETIF_F_LLTX __NETIF_F(LLTX)
122
123
#define NETIF_F_LOOPBACK __NETIF_F(LOOPBACK)
123
124
#define NETIF_F_LRO __NETIF_F(LRO)
124
-
#define NETIF_F_NETNS_LOCAL __NETIF_F(NETNS_LOCAL)
125
125
#define NETIF_F_NOCACHE_COPY __NETIF_F(NOCACHE_COPY)
126
126
#define NETIF_F_NTUPLE __NETIF_F(NTUPLE)
127
127
#define NETIF_F_RXCSUM __NETIF_F(RXCSUM)
···
188
192
189
193
/* Features valid for ethtool to change */
190
194
/* = all defined minus driver/device-class-related */
191
-
#define NETIF_F_NEVER_CHANGE (NETIF_F_VLAN_CHALLENGED | \
192
-
NETIF_F_LLTX | NETIF_F_NETNS_LOCAL)
195
+
#define NETIF_F_NEVER_CHANGE NETIF_F_VLAN_CHALLENGED
193
196
194
197
/* remember that ((t)1 << t_BITS) is undefined in C99 */
195
198
#define NETIF_F_ETHTOOL_BITS ((__NETIF_F_BIT(NETDEV_FEATURE_COUNT - 1) | \
···
208
213
209
214
#define NETIF_F_ALL_TSO (NETIF_F_TSO | NETIF_F_TSO6 | \
210
215
NETIF_F_TSO_ECN | NETIF_F_TSO_MANGLEID)
211
-
212
-
#define NETIF_F_ALL_FCOE (NETIF_F_FCOE_CRC | NETIF_F_FCOE_MTU | \
213
-
NETIF_F_FSO)
214
216
215
217
/* List of features with software fallbacks. */
216
218
#define NETIF_F_GSO_SOFTWARE (NETIF_F_ALL_TSO | NETIF_F_GSO_SCTP | \
+29
-13
include/linux/netdevice.h
+29
-13
include/linux/netdevice.h
···
1613
1613
* userspace; this means that the order of these flags can change
1614
1614
* during any kernel release.
1615
1615
*
1616
-
* You should have a pretty good reason to be extending these flags.
1616
+
* You should add bitfield booleans after either net_device::priv_flags
1617
+
* (hotpath) or ::threaded (slowpath) instead of extending these flags.
1617
1618
*
1618
1619
* @IFF_802_1Q_VLAN: 802.1Q VLAN device
1619
1620
* @IFF_EBRIDGE: Ethernet bridging device
···
1653
1652
* @IFF_NO_ADDRCONF: prevent ipv6 addrconf
1654
1653
* @IFF_TX_SKB_NO_LINEAR: device/driver is capable of xmitting frames with
1655
1654
* skb_headlen(skb) == 0 (data starts from frag0)
1656
-
* @IFF_CHANGE_PROTO_DOWN: device supports setting carrier via IFLA_PROTO_DOWN
1657
-
* @IFF_SEE_ALL_HWTSTAMP_REQUESTS: device wants to see calls to
1658
-
* ndo_hwtstamp_set() for all timestamp requests regardless of source,
1659
-
* even if those aren't HWTSTAMP_SOURCE_NETDEV.
1660
1655
*/
1661
1656
enum netdev_priv_flags {
1662
1657
IFF_802_1Q_VLAN = 1<<0,
···
1687
1690
IFF_L3MDEV_RX_HANDLER = 1<<29,
1688
1691
IFF_NO_ADDRCONF = BIT_ULL(30),
1689
1692
IFF_TX_SKB_NO_LINEAR = BIT_ULL(31),
1690
-
IFF_CHANGE_PROTO_DOWN = BIT_ULL(32),
1691
-
IFF_SEE_ALL_HWTSTAMP_REQUESTS = BIT_ULL(33),
1692
1693
};
1693
1694
1694
1695
/* Specifies the type of the struct net_device::ml_priv pointer */
···
1717
1722
* Actually, this whole structure is a big mistake. It mixes I/O
1718
1723
* data with strictly "high-level" data, and it has to know about
1719
1724
* almost every data structure used in the INET module.
1725
+
*
1726
+
* @priv_flags: flags invisible to userspace defined as bits, see
1727
+
* enum netdev_priv_flags for the definitions
1728
+
* @lltx: device supports lockless Tx. Deprecated for real HW
1729
+
* drivers. Mainly used by logical interfaces, such as
1730
+
* bonding and tunnels
1720
1731
*
1721
1732
* @name: This is the first field of the "visible" part of this structure
1722
1733
* (i.e. as seen by users in the "Space.c" file). It is the name
···
1790
1789
*
1791
1790
* @flags: Interface flags (a la BSD)
1792
1791
* @xdp_features: XDP capability supported by the device
1793
-
* @priv_flags: Like 'flags' but invisible to userspace,
1794
-
* see if.h for the definitions
1795
1792
* @gflags: Global flags ( kept as legacy )
1796
1793
* @priv_len: Size of the ->priv flexible array
1797
1794
* @priv: Flexible array containing private data
···
1963
1964
*
1964
1965
* @threaded: napi threaded mode is enabled
1965
1966
*
1967
+
* @see_all_hwtstamp_requests: device wants to see calls to
1968
+
* ndo_hwtstamp_set() for all timestamp requests
1969
+
* regardless of source, even if those aren't
1970
+
* HWTSTAMP_SOURCE_NETDEV
1971
+
* @change_proto_down: device supports setting carrier via IFLA_PROTO_DOWN
1972
+
* @netns_local: interface can't change network namespaces
1973
+
* @fcoe_mtu: device supports maximum FCoE MTU, 2158 bytes
1974
+
*
1966
1975
* @net_notifier_list: List of per-net netdev notifier block
1967
1976
* that follow this device when it is moved
1968
1977
* to another network namespace.
···
2021
2014
2022
2015
/* TX read-mostly hotpath */
2023
2016
__cacheline_group_begin(net_device_read_tx);
2024
-
unsigned long long priv_flags;
2017
+
struct_group(priv_flags_fast,
2018
+
unsigned long priv_flags:32;
2019
+
unsigned long lltx:1;
2020
+
);
2025
2021
const struct net_device_ops *netdev_ops;
2026
2022
const struct header_ops *header_ops;
2027
2023
struct netdev_queue *_tx;
···
2359
2349
struct lock_class_key *qdisc_tx_busylock;
2360
2350
bool proto_down;
2361
2351
bool threaded;
2352
+
2353
+
/* priv_flags_slow, ungrouped to save space */
2354
+
unsigned long see_all_hwtstamp_requests:1;
2355
+
unsigned long change_proto_down:1;
2356
+
unsigned long netns_local:1;
2357
+
unsigned long fcoe_mtu:1;
2362
2358
2363
2359
struct list_head net_notifier_list;
2364
2360
···
4441
4425
}
4442
4426
4443
4427
#define HARD_TX_LOCK(dev, txq, cpu) { \
4444
-
if ((dev->features & NETIF_F_LLTX) == 0) { \
4428
+
if (!(dev)->lltx) { \
4445
4429
__netif_tx_lock(txq, cpu); \
4446
4430
} else { \
4447
4431
__netif_tx_acquire(txq); \
···
4449
4433
}
4450
4434
4451
4435
#define HARD_TX_TRYLOCK(dev, txq) \
4452
-
(((dev->features & NETIF_F_LLTX) == 0) ? \
4436
+
(!(dev)->lltx ? \
4453
4437
__netif_tx_trylock(txq) : \
4454
4438
__netif_tx_acquire(txq))
4455
4439
4456
4440
#define HARD_TX_UNLOCK(dev, txq) { \
4457
-
if ((dev->features & NETIF_F_LLTX) == 0) { \
4441
+
if (!(dev)->lltx) { \
4458
4442
__netif_tx_unlock(txq); \
4459
4443
} else { \
4460
4444
__netif_tx_release(txq); \
+1
-2
lib/test_bpf.c
+1
-2
lib/test_bpf.c
···
15077
15077
.build_skb = build_test_skb_linear_no_head_frag,
15078
15078
.features = NETIF_F_SG | NETIF_F_FRAGLIST |
15079
15079
NETIF_F_HW_VLAN_CTAG_TX | NETIF_F_GSO |
15080
-
NETIF_F_LLTX | NETIF_F_GRO |
15081
-
NETIF_F_IPV6_CSUM | NETIF_F_RXCSUM |
15080
+
NETIF_F_GRO | NETIF_F_IPV6_CSUM | NETIF_F_RXCSUM |
15082
15081
NETIF_F_HW_VLAN_STAG_TX
15083
15082
}
15084
15083
};
+6
-4
net/8021q/vlan_dev.c
+6
-4
net/8021q/vlan_dev.c
···
564
564
NETIF_F_FRAGLIST | NETIF_F_GSO_SOFTWARE |
565
565
NETIF_F_GSO_ENCAP_ALL |
566
566
NETIF_F_HIGHDMA | NETIF_F_SCTP_CRC |
567
-
NETIF_F_ALL_FCOE;
567
+
NETIF_F_FCOE_CRC | NETIF_F_FSO;
568
568
569
569
if (real_dev->vlan_features & NETIF_F_HW_MACSEC)
570
570
dev->hw_features |= NETIF_F_HW_MACSEC;
571
571
572
-
dev->features |= dev->hw_features | NETIF_F_LLTX;
572
+
dev->features |= dev->hw_features;
573
+
dev->lltx = true;
574
+
dev->fcoe_mtu = true;
573
575
netif_inherit_tso_max(dev, real_dev);
574
576
if (dev->features & NETIF_F_VLAN_FEATURES)
575
577
netdev_warn(real_dev, "VLAN features are set incorrectly. Q-in-Q configurations may not work correctly.\n");
576
578
577
-
dev->vlan_features = real_dev->vlan_features & ~NETIF_F_ALL_FCOE;
579
+
dev->vlan_features = real_dev->vlan_features &
580
+
~(NETIF_F_FCOE_CRC | NETIF_F_FSO);
578
581
dev->hw_enc_features = vlan_tnl_features(real_dev);
579
582
dev->mpls_features = real_dev->mpls_features;
580
583
···
658
655
lower_features |= NETIF_F_HW_CSUM;
659
656
features = netdev_intersect_features(features, lower_features);
660
657
features |= old_features & (NETIF_F_SOFT_FEATURES | NETIF_F_GSO_SOFTWARE);
661
-
features |= NETIF_F_LLTX;
662
658
663
659
return features;
664
660
}
+2
-2
net/8021q/vlanproc.c
+2
-2
net/8021q/vlanproc.c
···
238
238
239
239
stats = dev_get_stats(vlandev, &temp);
240
240
seq_printf(seq,
241
-
"%s VID: %d REORDER_HDR: %i dev->priv_flags: %llx\n",
241
+
"%s VID: %d REORDER_HDR: %i dev->priv_flags: %x\n",
242
242
vlandev->name, vlan->vlan_id,
243
-
(int)(vlan->flags & 1), vlandev->priv_flags);
243
+
(int)(vlan->flags & 1), (u32)vlandev->priv_flags);
244
244
245
245
seq_printf(seq, fmt64, "total frames received", stats->rx_packets);
246
246
seq_printf(seq, fmt64, "total bytes received", stats->rx_bytes);
+3
-2
net/batman-adv/soft-interface.c
+3
-2
net/batman-adv/soft-interface.c
···
1020
1020
dev->netdev_ops = &batadv_netdev_ops;
1021
1021
dev->needs_free_netdev = true;
1022
1022
dev->priv_destructor = batadv_softif_free;
1023
-
dev->features |= NETIF_F_HW_VLAN_CTAG_FILTER | NETIF_F_NETNS_LOCAL;
1024
-
dev->features |= NETIF_F_LLTX;
1023
+
dev->features |= NETIF_F_HW_VLAN_CTAG_FILTER;
1025
1024
dev->priv_flags |= IFF_NO_QUEUE;
1025
+
dev->lltx = true;
1026
+
dev->netns_local = true;
1026
1027
1027
1028
/* can't call min_mtu, because the needed variables
1028
1029
* have not been initialized yet
+4
-2
net/bridge/br_device.c
+4
-2
net/bridge/br_device.c
···
487
487
dev->ethtool_ops = &br_ethtool_ops;
488
488
SET_NETDEV_DEVTYPE(dev, &br_type);
489
489
dev->priv_flags = IFF_EBRIDGE | IFF_NO_QUEUE;
490
+
dev->lltx = true;
491
+
dev->netns_local = true;
490
492
491
-
dev->features = COMMON_FEATURES | NETIF_F_LLTX | NETIF_F_NETNS_LOCAL |
492
-
NETIF_F_HW_VLAN_CTAG_TX | NETIF_F_HW_VLAN_STAG_TX;
493
+
dev->features = COMMON_FEATURES | NETIF_F_HW_VLAN_CTAG_TX |
494
+
NETIF_F_HW_VLAN_STAG_TX;
493
495
dev->hw_features = COMMON_FEATURES | NETIF_F_HW_VLAN_CTAG_TX |
494
496
NETIF_F_HW_VLAN_STAG_TX;
495
497
dev->vlan_features = COMMON_FEATURES;
+4
-4
net/core/dev.c
+4
-4
net/core/dev.c
···
9274
9274
*/
9275
9275
int dev_change_proto_down(struct net_device *dev, bool proto_down)
9276
9276
{
9277
-
if (!(dev->priv_flags & IFF_CHANGE_PROTO_DOWN))
9277
+
if (!dev->change_proto_down)
9278
9278
return -EOPNOTSUPP;
9279
9279
if (!netif_device_present(dev))
9280
9280
return -ENODEV;
···
11487
11487
11488
11488
/* Don't allow namespace local devices to be moved. */
11489
11489
err = -EINVAL;
11490
-
if (dev->features & NETIF_F_NETNS_LOCAL)
11490
+
if (dev->netns_local)
11491
11491
goto out;
11492
11492
11493
11493
/* Ensure the device has been registered */
···
11869
11869
char fb_name[IFNAMSIZ];
11870
11870
11871
11871
/* Ignore unmoveable devices (i.e. loopback) */
11872
-
if (dev->features & NETIF_F_NETNS_LOCAL)
11872
+
if (dev->netns_local)
11873
11873
continue;
11874
11874
11875
11875
/* Leave virtual devices for the generic cleanup */
···
11930
11930
static void __init net_dev_struct_check(void)
11931
11931
{
11932
11932
/* TX read-mostly hotpath */
11933
-
CACHELINE_ASSERT_GROUP_MEMBER(struct net_device, net_device_read_tx, priv_flags);
11933
+
CACHELINE_ASSERT_GROUP_MEMBER(struct net_device, net_device_read_tx, priv_flags_fast);
11934
11934
CACHELINE_ASSERT_GROUP_MEMBER(struct net_device, net_device_read_tx, netdev_ops);
11935
11935
CACHELINE_ASSERT_GROUP_MEMBER(struct net_device, net_device_read_tx, header_ops);
11936
11936
CACHELINE_ASSERT_GROUP_MEMBER(struct net_device, net_device_read_tx, _tx);
+4
-5
net/core/dev_ioctl.c
+4
-5
net/core/dev_ioctl.c
···
317
317
* should take precedence in front of hardware timestamping provided by the
318
318
* netdev. If the netdev driver needs to perform specific actions even for PHY
319
319
* timestamping to work properly (a switch port must trap the timestamped
320
-
* frames and not forward them), it must set IFF_SEE_ALL_HWTSTAMP_REQUESTS in
321
-
* dev->priv_flags.
320
+
* frames and not forward them), it must set dev->see_all_hwtstamp_requests.
322
321
*/
323
322
int dev_set_hwtstamp_phylib(struct net_device *dev,
324
323
struct kernel_hwtstamp_config *cfg,
···
331
332
332
333
cfg->source = phy_ts ? HWTSTAMP_SOURCE_PHYLIB : HWTSTAMP_SOURCE_NETDEV;
333
334
334
-
if (phy_ts && (dev->priv_flags & IFF_SEE_ALL_HWTSTAMP_REQUESTS)) {
335
+
if (phy_ts && dev->see_all_hwtstamp_requests) {
335
336
err = ops->ndo_hwtstamp_get(dev, &old_cfg);
336
337
if (err)
337
338
return err;
338
339
}
339
340
340
-
if (!phy_ts || (dev->priv_flags & IFF_SEE_ALL_HWTSTAMP_REQUESTS)) {
341
+
if (!phy_ts || dev->see_all_hwtstamp_requests) {
341
342
err = ops->ndo_hwtstamp_set(dev, cfg, extack);
342
343
if (err) {
343
344
if (extack->_msg)
···
346
347
}
347
348
}
348
349
349
-
if (phy_ts && (dev->priv_flags & IFF_SEE_ALL_HWTSTAMP_REQUESTS))
350
+
if (phy_ts && dev->see_all_hwtstamp_requests)
350
351
changed = kernel_hwtstamp_config_changed(&old_cfg, cfg);
351
352
352
353
if (phy_ts) {
+1
-2
net/core/net-sysfs.c
+1
-2
net/core/net-sysfs.c
···
1764
1764
1765
1765
static bool netdev_uses_bql(const struct net_device *dev)
1766
1766
{
1767
-
if (dev->features & NETIF_F_LLTX ||
1768
-
dev->priv_flags & IFF_NO_QUEUE)
1767
+
if (dev->lltx || (dev->priv_flags & IFF_NO_QUEUE))
1769
1768
return false;
1770
1769
1771
1770
return IS_ENABLED(CONFIG_BQL);
+1
-1
net/core/rtnetlink.c
+1
-1
net/core/rtnetlink.c
+2
-1
net/dsa/user.c
+2
-1
net/dsa/user.c
···
2642
2642
2643
2643
user->features = conduit->vlan_features | NETIF_F_HW_TC;
2644
2644
user->hw_features |= NETIF_F_HW_TC;
2645
-
user->features |= NETIF_F_LLTX;
2646
2645
if (user->needed_tailroom)
2647
2646
user->features &= ~(NETIF_F_SG | NETIF_F_FRAGLIST);
2648
2647
if (ds->needs_standalone_vlan_filtering)
2649
2648
user->features |= NETIF_F_HW_VLAN_CTAG_FILTER;
2649
+
2650
+
user->lltx = true;
2650
2651
}
2651
2652
2652
2653
int dsa_user_suspend(struct net_device *user_dev)
-3
net/ethtool/common.c
-3
net/ethtool/common.c
···
25
25
[NETIF_F_HW_VLAN_STAG_FILTER_BIT] = "rx-vlan-stag-filter",
26
26
[NETIF_F_VLAN_CHALLENGED_BIT] = "vlan-challenged",
27
27
[NETIF_F_GSO_BIT] = "tx-generic-segmentation",
28
-
[NETIF_F_LLTX_BIT] = "tx-lockless",
29
-
[NETIF_F_NETNS_LOCAL_BIT] = "netns-local",
30
28
[NETIF_F_GRO_BIT] = "rx-gro",
31
29
[NETIF_F_GRO_HW_BIT] = "rx-gro-hw",
32
30
[NETIF_F_LRO_BIT] = "rx-lro",
···
50
52
51
53
[NETIF_F_FCOE_CRC_BIT] = "tx-checksum-fcoe-crc",
52
54
[NETIF_F_SCTP_CRC_BIT] = "tx-checksum-sctp",
53
-
[NETIF_F_FCOE_MTU_BIT] = "fcoe-mtu",
54
55
[NETIF_F_NTUPLE_BIT] = "rx-ntuple-filter",
55
56
[NETIF_F_RXHASH_BIT] = "rx-hashing",
56
57
[NETIF_F_RXCSUM_BIT] = "rx-checksum",
+6
-6
net/hsr/hsr_device.c
+6
-6
net/hsr/hsr_device.c
···
554
554
dev->netdev_ops = &hsr_device_ops;
555
555
SET_NETDEV_DEVTYPE(dev, &hsr_type);
556
556
dev->priv_flags |= IFF_NO_QUEUE | IFF_DISABLE_NETPOLL;
557
+
/* Prevent recursive tx locking */
558
+
dev->lltx = true;
559
+
/* Not sure about this. Taken from bridge code. netdevice.h says
560
+
* it means "Does not change network namespaces".
561
+
*/
562
+
dev->netns_local = true;
557
563
558
564
dev->needs_free_netdev = true;
559
565
···
569
563
570
564
dev->features = dev->hw_features;
571
565
572
-
/* Prevent recursive tx locking */
573
-
dev->features |= NETIF_F_LLTX;
574
566
/* VLAN on top of HSR needs testing and probably some work on
575
567
* hsr_header_create() etc.
576
568
*/
577
569
dev->features |= NETIF_F_VLAN_CHALLENGED;
578
-
/* Not sure about this. Taken from bridge code. netdev_features.h says
579
-
* it means "Does not change network namespaces".
580
-
*/
581
-
dev->features |= NETIF_F_NETNS_LOCAL;
582
570
}
583
571
584
572
/* Return true if dev is a HSR master; return false otherwise.
+1
-1
net/ieee802154/6lowpan/core.c
+1
-1
net/ieee802154/6lowpan/core.c
···
116
116
ldev->netdev_ops = &lowpan_netdev_ops;
117
117
ldev->header_ops = &lowpan_header_ops;
118
118
ldev->needs_free_netdev = true;
119
-
ldev->features |= NETIF_F_NETNS_LOCAL;
119
+
ldev->netns_local = true;
120
120
}
121
121
122
122
static int lowpan_validate(struct nlattr *tb[], struct nlattr *data[],
+5
-5
net/ieee802154/core.c
+5
-5
net/ieee802154/core.c
···
226
226
list_for_each_entry(wpan_dev, &rdev->wpan_dev_list, list) {
227
227
if (!wpan_dev->netdev)
228
228
continue;
229
-
wpan_dev->netdev->features &= ~NETIF_F_NETNS_LOCAL;
229
+
wpan_dev->netdev->netns_local = false;
230
230
err = dev_change_net_namespace(wpan_dev->netdev, net, "wpan%d");
231
231
if (err)
232
232
break;
233
-
wpan_dev->netdev->features |= NETIF_F_NETNS_LOCAL;
233
+
wpan_dev->netdev->netns_local = true;
234
234
}
235
235
236
236
if (err) {
···
242
242
list) {
243
243
if (!wpan_dev->netdev)
244
244
continue;
245
-
wpan_dev->netdev->features &= ~NETIF_F_NETNS_LOCAL;
245
+
wpan_dev->netdev->netns_local = false;
246
246
err = dev_change_net_namespace(wpan_dev->netdev, net,
247
247
"wpan%d");
248
248
WARN_ON(err);
249
-
wpan_dev->netdev->features |= NETIF_F_NETNS_LOCAL;
249
+
wpan_dev->netdev->netns_local = true;
250
250
}
251
251
252
252
return err;
···
291
291
switch (state) {
292
292
/* TODO NETDEV_DEVTYPE */
293
293
case NETDEV_REGISTER:
294
-
dev->features |= NETIF_F_NETNS_LOCAL;
294
+
dev->netns_local = true;
295
295
wpan_dev->identifier = ++rdev->wpan_dev_id;
296
296
list_add_rcu(&wpan_dev->list, &rdev->wpan_dev_list);
297
297
rdev->devlist_generation++;
+3
-1
net/ipv4/ip_gre.c
+3
-1
net/ipv4/ip_gre.c
···
996
996
tunnel->hlen = tunnel->tun_hlen + tunnel->encap_hlen;
997
997
dev->needed_headroom = tunnel->hlen + sizeof(tunnel->parms.iph);
998
998
999
-
dev->features |= GRE_FEATURES | NETIF_F_LLTX;
999
+
dev->features |= GRE_FEATURES;
1000
1000
dev->hw_features |= GRE_FEATURES;
1001
1001
1002
1002
/* TCP offload with GRE SEQ is not supported, nor can we support 2
···
1010
1010
1011
1011
dev->features |= NETIF_F_GSO_SOFTWARE;
1012
1012
dev->hw_features |= NETIF_F_GSO_SOFTWARE;
1013
+
1014
+
dev->lltx = true;
1013
1015
}
1014
1016
1015
1017
static int ipgre_tunnel_init(struct net_device *dev)
+1
-1
net/ipv4/ip_tunnel.c
+1
-1
net/ipv4/ip_tunnel.c
···
1161
1161
* Allowing to move it to another netns is clearly unsafe.
1162
1162
*/
1163
1163
if (!IS_ERR(itn->fb_tunnel_dev)) {
1164
-
itn->fb_tunnel_dev->features |= NETIF_F_NETNS_LOCAL;
1164
+
itn->fb_tunnel_dev->netns_local = true;
1165
1165
itn->fb_tunnel_dev->mtu = ip_tunnel_bind_dev(itn->fb_tunnel_dev);
1166
1166
ip_tunnel_add(itn, netdev_priv(itn->fb_tunnel_dev));
1167
1167
itn->type = itn->fb_tunnel_dev->type;
+1
-1
net/ipv4/ip_vti.c
+1
-1
net/ipv4/ip_vti.c
+1
-1
net/ipv4/ipip.c
+1
-1
net/ipv4/ipip.c
+1
-1
net/ipv4/ipmr.c
+1
-1
net/ipv4/ipmr.c
···
537
537
dev->flags = IFF_NOARP;
538
538
dev->netdev_ops = ®_vif_netdev_ops;
539
539
dev->needs_free_netdev = true;
540
-
dev->features |= NETIF_F_NETNS_LOCAL;
540
+
dev->netns_local = true;
541
541
}
542
542
543
543
static struct net_device *ipmr_reg_vif(struct net *net, struct mr_table *mrt)
+4
-3
net/ipv6/ip6_gre.c
+4
-3
net/ipv6/ip6_gre.c
···
1471
1471
{
1472
1472
struct ip6_tnl *nt = netdev_priv(dev);
1473
1473
1474
-
dev->features |= GRE6_FEATURES | NETIF_F_LLTX;
1474
+
dev->features |= GRE6_FEATURES;
1475
1475
dev->hw_features |= GRE6_FEATURES;
1476
1476
1477
1477
/* TCP offload with GRE SEQ is not supported, nor can we support 2
···
1485
1485
1486
1486
dev->features |= NETIF_F_GSO_SOFTWARE;
1487
1487
dev->hw_features |= NETIF_F_GSO_SOFTWARE;
1488
+
1489
+
dev->lltx = true;
1488
1490
}
1489
1491
1490
1492
static int ip6gre_tunnel_init_common(struct net_device *dev)
···
1621
1619
/* FB netdevice is special: we have one, and only one per netns.
1622
1620
* Allowing to move it to another netns is clearly unsafe.
1623
1621
*/
1624
-
ign->fb_tunnel_dev->features |= NETIF_F_NETNS_LOCAL;
1625
-
1622
+
ign->fb_tunnel_dev->netns_local = true;
1626
1623
1627
1624
ip6gre_fb_tunnel_init(ign->fb_tunnel_dev);
1628
1625
ign->fb_tunnel_dev->rtnl_link_ops = &ip6gre_link_ops;
+2
-2
net/ipv6/ip6_tunnel.c
+2
-2
net/ipv6/ip6_tunnel.c
···
1849
1849
dev->type = ARPHRD_TUNNEL6;
1850
1850
dev->flags |= IFF_NOARP;
1851
1851
dev->addr_len = sizeof(struct in6_addr);
1852
-
dev->features |= NETIF_F_LLTX;
1852
+
dev->lltx = true;
1853
1853
dev->pcpu_stat_type = NETDEV_PCPU_STAT_TSTATS;
1854
1854
netif_keep_dst(dev);
1855
1855
···
2258
2258
/* FB netdevice is special: we have one, and only one per netns.
2259
2259
* Allowing to move it to another netns is clearly unsafe.
2260
2260
*/
2261
-
ip6n->fb_tnl_dev->features |= NETIF_F_NETNS_LOCAL;
2261
+
ip6n->fb_tnl_dev->netns_local = true;
2262
2262
2263
2263
err = ip6_fb_tnl_dev_init(ip6n->fb_tnl_dev);
2264
2264
if (err < 0)
+1
-1
net/ipv6/ip6mr.c
+1
-1
net/ipv6/ip6mr.c
···
640
640
dev->flags = IFF_NOARP;
641
641
dev->netdev_ops = ®_vif_netdev_ops;
642
642
dev->needs_free_netdev = true;
643
-
dev->features |= NETIF_F_NETNS_LOCAL;
643
+
dev->netns_local = true;
644
644
}
645
645
646
646
static struct net_device *ip6mr_reg_vif(struct net *net, struct mr_table *mrt)
+2
-2
net/ipv6/sit.c
+2
-2
net/ipv6/sit.c
···
1436
1436
dev->flags = IFF_NOARP;
1437
1437
netif_keep_dst(dev);
1438
1438
dev->addr_len = 4;
1439
-
dev->features |= NETIF_F_LLTX;
1439
+
dev->lltx = true;
1440
1440
dev->features |= SIT_FEATURES;
1441
1441
dev->hw_features |= SIT_FEATURES;
1442
1442
dev->pcpu_stat_type = NETDEV_PCPU_STAT_TSTATS;
···
1856
1856
/* FB netdevice is special: we have one, and only one per netns.
1857
1857
* Allowing to move it to another netns is clearly unsafe.
1858
1858
*/
1859
-
sitn->fb_tunnel_dev->features |= NETIF_F_NETNS_LOCAL;
1859
+
sitn->fb_tunnel_dev->netns_local = true;
1860
1860
1861
1861
err = register_netdev(sitn->fb_tunnel_dev);
1862
1862
if (err)
+1
-1
net/l2tp/l2tp_eth.c
+1
-1
net/l2tp/l2tp_eth.c
···
97
97
SET_NETDEV_DEVTYPE(dev, &l2tpeth_type);
98
98
ether_setup(dev);
99
99
dev->priv_flags &= ~IFF_TX_SKB_SHARING;
100
-
dev->features |= NETIF_F_LLTX;
100
+
dev->lltx = true;
101
101
dev->netdev_ops = &l2tp_eth_netdev_ops;
102
102
dev->needs_free_netdev = true;
103
103
dev->pcpu_stat_type = NETDEV_PCPU_STAT_TSTATS;
+6
-5
net/openvswitch/vport-internal_dev.c
+6
-5
net/openvswitch/vport-internal_dev.c
···
102
102
netdev->priv_flags &= ~IFF_TX_SKB_SHARING;
103
103
netdev->priv_flags |= IFF_LIVE_ADDR_CHANGE | IFF_OPENVSWITCH |
104
104
IFF_NO_QUEUE;
105
+
netdev->lltx = true;
105
106
netdev->needs_free_netdev = true;
106
107
netdev->priv_destructor = NULL;
107
108
netdev->ethtool_ops = &internal_dev_ethtool_ops;
108
109
netdev->rtnl_link_ops = &internal_dev_link_ops;
109
110
110
-
netdev->features = NETIF_F_LLTX | NETIF_F_SG | NETIF_F_FRAGLIST |
111
-
NETIF_F_HIGHDMA | NETIF_F_HW_CSUM |
112
-
NETIF_F_GSO_SOFTWARE | NETIF_F_GSO_ENCAP_ALL;
111
+
netdev->features = NETIF_F_SG | NETIF_F_FRAGLIST | NETIF_F_HIGHDMA |
112
+
NETIF_F_HW_CSUM | NETIF_F_GSO_SOFTWARE |
113
+
NETIF_F_GSO_ENCAP_ALL;
113
114
114
115
netdev->vlan_features = netdev->features;
115
116
netdev->hw_enc_features = netdev->features;
116
117
netdev->features |= NETIF_F_HW_VLAN_CTAG_TX | NETIF_F_HW_VLAN_STAG_TX;
117
-
netdev->hw_features = netdev->features & ~NETIF_F_LLTX;
118
+
netdev->hw_features = netdev->features;
118
119
119
120
eth_hw_addr_random(netdev);
120
121
}
···
149
148
150
149
/* Restrict bridge port to current netns. */
151
150
if (vport->port_no == OVSP_LOCAL)
152
-
vport->dev->features |= NETIF_F_NETNS_LOCAL;
151
+
vport->dev->netns_local = true;
153
152
154
153
rtnl_lock();
155
154
err = register_netdevice(vport->dev);
+5
-5
net/wireless/core.c
+5
-5
net/wireless/core.c
···
165
165
list_for_each_entry(wdev, &rdev->wiphy.wdev_list, list) {
166
166
if (!wdev->netdev)
167
167
continue;
168
-
wdev->netdev->features &= ~NETIF_F_NETNS_LOCAL;
168
+
wdev->netdev->netns_local = false;
169
169
err = dev_change_net_namespace(wdev->netdev, net, "wlan%d");
170
170
if (err)
171
171
break;
172
-
wdev->netdev->features |= NETIF_F_NETNS_LOCAL;
172
+
wdev->netdev->netns_local = true;
173
173
}
174
174
175
175
if (err) {
···
181
181
list) {
182
182
if (!wdev->netdev)
183
183
continue;
184
-
wdev->netdev->features &= ~NETIF_F_NETNS_LOCAL;
184
+
wdev->netdev->netns_local = false;
185
185
err = dev_change_net_namespace(wdev->netdev, net,
186
186
"wlan%d");
187
187
WARN_ON(err);
188
-
wdev->netdev->features |= NETIF_F_NETNS_LOCAL;
188
+
wdev->netdev->netns_local = true;
189
189
}
190
190
191
191
return err;
···
1473
1473
SET_NETDEV_DEVTYPE(dev, &wiphy_type);
1474
1474
wdev->netdev = dev;
1475
1475
/* can only change netns with wiphy */
1476
-
dev->features |= NETIF_F_NETNS_LOCAL;
1476
+
dev->netns_local = true;
1477
1477
1478
1478
cfg80211_init_wdev(wdev);
1479
1479
break;
+1
-1
net/xfrm/xfrm_interface_core.c
+1
-1
net/xfrm/xfrm_interface_core.c
+1
-1
tools/testing/selftests/net/forwarding/README
+1
-1
tools/testing/selftests/net/forwarding/README
···
6
6
7
7
Unfortunately, these namespaces can not be used with actual switching
8
8
ASICs, as their ports can not be migrated to other network namespaces
9
-
(NETIF_F_NETNS_LOCAL) and most of them probably do not support the
9
+
(dev->netns_local) and most of them probably do not support the
10
10
L1-separation provided by namespaces.
11
11
12
12
However, a similar kind of flexibility can be achieved by using VRFs and