diff options
author | Linus Torvalds <torvalds@linux-foundation.org> | 2021-09-01 02:43:06 +0300 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2021-09-01 02:43:06 +0300 |
commit | 9e9fb7655ed585da8f468e29221f0ba194a5f613 (patch) | |
tree | d2c51887389b8297635a5b90d5766897f00fe928 /net/switchdev | |
parent | 86ac54e79fe09b34c52691a780a6e31d12fa57f4 (diff) | |
parent | 29ce8f9701072fc221d9c38ad952de1a9578f95c (diff) | |
download | linux-9e9fb7655ed585da8f468e29221f0ba194a5f613.tar.xz |
Merge tag 'net-next-5.15' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next
Pull networking updates from Jakub Kicinski:
"Core:
- Enable memcg accounting for various networking objects.
BPF:
- Introduce bpf timers.
- Add perf link and opaque bpf_cookie which the program can read out
again, to be used in libbpf-based USDT library.
- Add bpf_task_pt_regs() helper to access user space pt_regs in
kprobes, to help user space stack unwinding.
- Add support for UNIX sockets for BPF sockmap.
- Extend BPF iterator support for UNIX domain sockets.
- Allow BPF TCP congestion control progs and bpf iterators to call
bpf_setsockopt(), e.g. to switch to another congestion control
algorithm.
Protocols:
- Support IOAM Pre-allocated Trace with IPv6.
- Support Management Component Transport Protocol.
- bridge: multicast: add vlan support.
- netfilter: add hooks for the SRv6 lightweight tunnel driver.
- tcp:
- enable mid-stream window clamping (by user space or BPF)
- allow data-less, empty-cookie SYN with TFO_SERVER_COOKIE_NOT_REQD
- more accurate DSACK processing for RACK-TLP
- mptcp:
- add full mesh path manager option
- add partial support for MP_FAIL
- improve use of backup subflows
- optimize option processing
- af_unix: add OOB notification support.
- ipv6: add IFLA_INET6_RA_MTU to expose MTU value advertised by the
router.
- mac80211: Target Wake Time support in AP mode.
- can: j1939: extend UAPI to notify about RX status.
Driver APIs:
- Add page frag support in page pool API.
- Many improvements to the DSA (distributed switch) APIs.
- ethtool: extend IRQ coalesce uAPI with timer reset modes.
- devlink: control which auxiliary devices are created.
- Support CAN PHYs via the generic PHY subsystem.
- Proper cross-chip support for tag_8021q.
- Allow TX forwarding for the software bridge data path to be
offloaded to capable devices.
Drivers:
- veth: more flexible channels number configuration.
- openvswitch: introduce per-cpu upcall dispatch.
- Add internet mix (IMIX) mode to pktgen.
- Transparently handle XDP operations in the bonding driver.
- Add LiteETH network driver.
- Renesas (ravb):
- support Gigabit Ethernet IP
- NXP Ethernet switch (sja1105):
- fast aging support
- support for "H" switch topologies
- traffic termination for ports under VLAN-aware bridge
- Intel 1G Ethernet
- support getcrosststamp() with PCIe PTM (Precision Time
Measurement) for better time sync
- support Credit-Based Shaper (CBS) offload, enabling HW traffic
prioritization and bandwidth reservation
- Broadcom Ethernet (bnxt)
- support pulse-per-second output
- support larger Rx rings
- Mellanox Ethernet (mlx5)
- support ethtool RSS contexts and MQPRIO channel mode
- support LAG offload with bridging
- support devlink rate limit API
- support packet sampling on tunnels
- Huawei Ethernet (hns3):
- basic devlink support
- add extended IRQ coalescing support
- report extended link state
- Netronome Ethernet (nfp):
- add conntrack offload support
- Broadcom WiFi (brcmfmac):
- add WPA3 Personal with FT to supported cipher suites
- support 43752 SDIO device
- Intel WiFi (iwlwifi):
- support scanning hidden 6GHz networks
- support for a new hardware family (Bz)
- Xen pv driver:
- harden netfront against malicious backends
- Qualcomm mobile
- ipa: refactor power management and enable automatic suspend
- mhi: move MBIM to WWAN subsystem interfaces
Refactor:
- Ambient BPF run context and cgroup storage cleanup.
- Compat rework for ndo_ioctl.
Old code removal:
- prism54 remove the obsoleted driver, deprecated by the p54 driver.
- wan: remove sbni/granch driver"
* tag 'net-next-5.15' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next: (1715 commits)
net: Add depends on OF_NET for LiteX's LiteETH
ipv6: seg6: remove duplicated include
net: hns3: remove unnecessary spaces
net: hns3: add some required spaces
net: hns3: clean up a type mismatch warning
net: hns3: refine function hns3_set_default_feature()
ipv6: remove duplicated 'net/lwtunnel.h' include
net: w5100: check return value after calling platform_get_resource()
net/mlxbf_gige: Make use of devm_platform_ioremap_resourcexxx()
net: mdio: mscc-miim: Make use of the helper function devm_platform_ioremap_resource()
net: mdio-ipq4019: Make use of devm_platform_ioremap_resource()
fou: remove sparse errors
ipv4: fix endianness issue in inet_rtm_getroute_build_skb()
octeontx2-af: Set proper errorcode for IPv4 checksum errors
octeontx2-af: Fix static code analyzer reported issues
octeontx2-af: Fix mailbox errors in nix_rss_flowkey_cfg
octeontx2-af: Fix loop in free and unmap counter
af_unix: fix potential NULL deref in unix_dgram_connect()
dpaa2-eth: Replace strlcpy with strscpy
octeontx2-af: Use NDC TX for transmit packet data
...
Diffstat (limited to 'net/switchdev')
-rw-r--r-- | net/switchdev/switchdev.c | 308 |
1 files changed, 308 insertions, 0 deletions
diff --git a/net/switchdev/switchdev.c b/net/switchdev/switchdev.c index 070698dd19bc..0b2c18efc079 100644 --- a/net/switchdev/switchdev.c +++ b/net/switchdev/switchdev.c @@ -378,6 +378,266 @@ int call_switchdev_blocking_notifiers(unsigned long val, struct net_device *dev, } EXPORT_SYMBOL_GPL(call_switchdev_blocking_notifiers); +struct switchdev_nested_priv { + bool (*check_cb)(const struct net_device *dev); + bool (*foreign_dev_check_cb)(const struct net_device *dev, + const struct net_device *foreign_dev); + const struct net_device *dev; + struct net_device *lower_dev; +}; + +static int switchdev_lower_dev_walk(struct net_device *lower_dev, + struct netdev_nested_priv *priv) +{ + struct switchdev_nested_priv *switchdev_priv = priv->data; + bool (*foreign_dev_check_cb)(const struct net_device *dev, + const struct net_device *foreign_dev); + bool (*check_cb)(const struct net_device *dev); + const struct net_device *dev; + + check_cb = switchdev_priv->check_cb; + foreign_dev_check_cb = switchdev_priv->foreign_dev_check_cb; + dev = switchdev_priv->dev; + + if (check_cb(lower_dev) && !foreign_dev_check_cb(lower_dev, dev)) { + switchdev_priv->lower_dev = lower_dev; + return 1; + } + + return 0; +} + +static struct net_device * +switchdev_lower_dev_find(struct net_device *dev, + bool (*check_cb)(const struct net_device *dev), + bool (*foreign_dev_check_cb)(const struct net_device *dev, + const struct net_device *foreign_dev)) +{ + struct switchdev_nested_priv switchdev_priv = { + .check_cb = check_cb, + .foreign_dev_check_cb = foreign_dev_check_cb, + .dev = dev, + .lower_dev = NULL, + }; + struct netdev_nested_priv priv = { + .data = &switchdev_priv, + }; + + netdev_walk_all_lower_dev_rcu(dev, switchdev_lower_dev_walk, &priv); + + return switchdev_priv.lower_dev; +} + +static int __switchdev_handle_fdb_add_to_device(struct net_device *dev, + const struct net_device *orig_dev, + const struct switchdev_notifier_fdb_info *fdb_info, + bool (*check_cb)(const struct net_device *dev), + bool (*foreign_dev_check_cb)(const struct net_device *dev, + const struct net_device *foreign_dev), + int (*add_cb)(struct net_device *dev, + const struct net_device *orig_dev, const void *ctx, + const struct switchdev_notifier_fdb_info *fdb_info), + int (*lag_add_cb)(struct net_device *dev, + const struct net_device *orig_dev, const void *ctx, + const struct switchdev_notifier_fdb_info *fdb_info)) +{ + const struct switchdev_notifier_info *info = &fdb_info->info; + struct net_device *br, *lower_dev; + struct list_head *iter; + int err = -EOPNOTSUPP; + + if (check_cb(dev)) + return add_cb(dev, orig_dev, info->ctx, fdb_info); + + if (netif_is_lag_master(dev)) { + if (!switchdev_lower_dev_find(dev, check_cb, foreign_dev_check_cb)) + goto maybe_bridged_with_us; + + /* This is a LAG interface that we offload */ + if (!lag_add_cb) + return -EOPNOTSUPP; + + return lag_add_cb(dev, orig_dev, info->ctx, fdb_info); + } + + /* Recurse through lower interfaces in case the FDB entry is pointing + * towards a bridge device. + */ + if (netif_is_bridge_master(dev)) { + if (!switchdev_lower_dev_find(dev, check_cb, foreign_dev_check_cb)) + return 0; + + /* This is a bridge interface that we offload */ + netdev_for_each_lower_dev(dev, lower_dev, iter) { + /* Do not propagate FDB entries across bridges */ + if (netif_is_bridge_master(lower_dev)) + continue; + + /* Bridge ports might be either us, or LAG interfaces + * that we offload. + */ + if (!check_cb(lower_dev) && + !switchdev_lower_dev_find(lower_dev, check_cb, + foreign_dev_check_cb)) + continue; + + err = __switchdev_handle_fdb_add_to_device(lower_dev, orig_dev, + fdb_info, check_cb, + foreign_dev_check_cb, + add_cb, lag_add_cb); + if (err && err != -EOPNOTSUPP) + return err; + } + + return 0; + } + +maybe_bridged_with_us: + /* Event is neither on a bridge nor a LAG. Check whether it is on an + * interface that is in a bridge with us. + */ + br = netdev_master_upper_dev_get_rcu(dev); + if (!br || !netif_is_bridge_master(br)) + return 0; + + if (!switchdev_lower_dev_find(br, check_cb, foreign_dev_check_cb)) + return 0; + + return __switchdev_handle_fdb_add_to_device(br, orig_dev, fdb_info, + check_cb, foreign_dev_check_cb, + add_cb, lag_add_cb); +} + +int switchdev_handle_fdb_add_to_device(struct net_device *dev, + const struct switchdev_notifier_fdb_info *fdb_info, + bool (*check_cb)(const struct net_device *dev), + bool (*foreign_dev_check_cb)(const struct net_device *dev, + const struct net_device *foreign_dev), + int (*add_cb)(struct net_device *dev, + const struct net_device *orig_dev, const void *ctx, + const struct switchdev_notifier_fdb_info *fdb_info), + int (*lag_add_cb)(struct net_device *dev, + const struct net_device *orig_dev, const void *ctx, + const struct switchdev_notifier_fdb_info *fdb_info)) +{ + int err; + + err = __switchdev_handle_fdb_add_to_device(dev, dev, fdb_info, + check_cb, + foreign_dev_check_cb, + add_cb, lag_add_cb); + if (err == -EOPNOTSUPP) + err = 0; + + return err; +} +EXPORT_SYMBOL_GPL(switchdev_handle_fdb_add_to_device); + +static int __switchdev_handle_fdb_del_to_device(struct net_device *dev, + const struct net_device *orig_dev, + const struct switchdev_notifier_fdb_info *fdb_info, + bool (*check_cb)(const struct net_device *dev), + bool (*foreign_dev_check_cb)(const struct net_device *dev, + const struct net_device *foreign_dev), + int (*del_cb)(struct net_device *dev, + const struct net_device *orig_dev, const void *ctx, + const struct switchdev_notifier_fdb_info *fdb_info), + int (*lag_del_cb)(struct net_device *dev, + const struct net_device *orig_dev, const void *ctx, + const struct switchdev_notifier_fdb_info *fdb_info)) +{ + const struct switchdev_notifier_info *info = &fdb_info->info; + struct net_device *br, *lower_dev; + struct list_head *iter; + int err = -EOPNOTSUPP; + + if (check_cb(dev)) + return del_cb(dev, orig_dev, info->ctx, fdb_info); + + if (netif_is_lag_master(dev)) { + if (!switchdev_lower_dev_find(dev, check_cb, foreign_dev_check_cb)) + goto maybe_bridged_with_us; + + /* This is a LAG interface that we offload */ + if (!lag_del_cb) + return -EOPNOTSUPP; + + return lag_del_cb(dev, orig_dev, info->ctx, fdb_info); + } + + /* Recurse through lower interfaces in case the FDB entry is pointing + * towards a bridge device. + */ + if (netif_is_bridge_master(dev)) { + if (!switchdev_lower_dev_find(dev, check_cb, foreign_dev_check_cb)) + return 0; + + /* This is a bridge interface that we offload */ + netdev_for_each_lower_dev(dev, lower_dev, iter) { + /* Do not propagate FDB entries across bridges */ + if (netif_is_bridge_master(lower_dev)) + continue; + + /* Bridge ports might be either us, or LAG interfaces + * that we offload. + */ + if (!check_cb(lower_dev) && + !switchdev_lower_dev_find(lower_dev, check_cb, + foreign_dev_check_cb)) + continue; + + err = __switchdev_handle_fdb_del_to_device(lower_dev, orig_dev, + fdb_info, check_cb, + foreign_dev_check_cb, + del_cb, lag_del_cb); + if (err && err != -EOPNOTSUPP) + return err; + } + + return 0; + } + +maybe_bridged_with_us: + /* Event is neither on a bridge nor a LAG. Check whether it is on an + * interface that is in a bridge with us. + */ + br = netdev_master_upper_dev_get_rcu(dev); + if (!br || !netif_is_bridge_master(br)) + return 0; + + if (!switchdev_lower_dev_find(br, check_cb, foreign_dev_check_cb)) + return 0; + + return __switchdev_handle_fdb_del_to_device(br, orig_dev, fdb_info, + check_cb, foreign_dev_check_cb, + del_cb, lag_del_cb); +} + +int switchdev_handle_fdb_del_to_device(struct net_device *dev, + const struct switchdev_notifier_fdb_info *fdb_info, + bool (*check_cb)(const struct net_device *dev), + bool (*foreign_dev_check_cb)(const struct net_device *dev, + const struct net_device *foreign_dev), + int (*del_cb)(struct net_device *dev, + const struct net_device *orig_dev, const void *ctx, + const struct switchdev_notifier_fdb_info *fdb_info), + int (*lag_del_cb)(struct net_device *dev, + const struct net_device *orig_dev, const void *ctx, + const struct switchdev_notifier_fdb_info *fdb_info)) +{ + int err; + + err = __switchdev_handle_fdb_del_to_device(dev, dev, fdb_info, + check_cb, + foreign_dev_check_cb, + del_cb, lag_del_cb); + if (err == -EOPNOTSUPP) + err = 0; + + return err; +} +EXPORT_SYMBOL_GPL(switchdev_handle_fdb_del_to_device); + static int __switchdev_handle_port_obj_add(struct net_device *dev, struct switchdev_notifier_port_obj_info *port_obj_info, bool (*check_cb)(const struct net_device *dev), @@ -549,3 +809,51 @@ int switchdev_handle_port_attr_set(struct net_device *dev, return err; } EXPORT_SYMBOL_GPL(switchdev_handle_port_attr_set); + +int switchdev_bridge_port_offload(struct net_device *brport_dev, + struct net_device *dev, const void *ctx, + struct notifier_block *atomic_nb, + struct notifier_block *blocking_nb, + bool tx_fwd_offload, + struct netlink_ext_ack *extack) +{ + struct switchdev_notifier_brport_info brport_info = { + .brport = { + .dev = dev, + .ctx = ctx, + .atomic_nb = atomic_nb, + .blocking_nb = blocking_nb, + .tx_fwd_offload = tx_fwd_offload, + }, + }; + int err; + + ASSERT_RTNL(); + + err = call_switchdev_blocking_notifiers(SWITCHDEV_BRPORT_OFFLOADED, + brport_dev, &brport_info.info, + extack); + return notifier_to_errno(err); +} +EXPORT_SYMBOL_GPL(switchdev_bridge_port_offload); + +void switchdev_bridge_port_unoffload(struct net_device *brport_dev, + const void *ctx, + struct notifier_block *atomic_nb, + struct notifier_block *blocking_nb) +{ + struct switchdev_notifier_brport_info brport_info = { + .brport = { + .ctx = ctx, + .atomic_nb = atomic_nb, + .blocking_nb = blocking_nb, + }, + }; + + ASSERT_RTNL(); + + call_switchdev_blocking_notifiers(SWITCHDEV_BRPORT_UNOFFLOADED, + brport_dev, &brport_info.info, + NULL); +} +EXPORT_SYMBOL_GPL(switchdev_bridge_port_unoffload); |