From 5e51e50e2324c9374d06ab05e3d7d09123e1114f Mon Sep 17 00:00:00 2001 From: Kory Maincent Date: Thu, 12 Dec 2024 18:06:41 +0100 Subject: net: Make dev_get_hwtstamp_phylib accessible Make the dev_get_hwtstamp_phylib function accessible in prevision to use it from ethtool to read the hwtstamp current configuration. Reviewed-by: Florian Fainelli Reviewed-by: Jacob Keller Signed-off-by: Kory Maincent Signed-off-by: David S. Miller --- net/core/dev_ioctl.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'net/core/dev_ioctl.c') diff --git a/net/core/dev_ioctl.c b/net/core/dev_ioctl.c index 46d43b950471..67cf68817f23 100644 --- a/net/core/dev_ioctl.c +++ b/net/core/dev_ioctl.c @@ -266,8 +266,8 @@ static int dev_eth_ioctl(struct net_device *dev, * -EOPNOTSUPP for phylib for now, which is still more accurate than letting * the netdev handle the GET request. */ -static int dev_get_hwtstamp_phylib(struct net_device *dev, - struct kernel_hwtstamp_config *cfg) +int dev_get_hwtstamp_phylib(struct net_device *dev, + struct kernel_hwtstamp_config *cfg) { if (phy_is_default_hwtstamp(dev->phydev)) return phy_hwtstamp_get(dev->phydev, cfg); -- cgit v1.2.3 From b18fe47c0c093cc429f7e4d7694fdf0fc362aaf5 Mon Sep 17 00:00:00 2001 From: Kory Maincent Date: Thu, 12 Dec 2024 18:06:42 +0100 Subject: net: Make net_hwtstamp_validate accessible Make the net_hwtstamp_validate function accessible in prevision to use it from ethtool to validate the hwtstamp configuration before setting it. Reviewed-by: Florian Fainelli Reviewed-by: Jacob Keller Signed-off-by: Kory Maincent Signed-off-by: David S. Miller --- net/core/dev.h | 1 + net/core/dev_ioctl.c | 2 +- 2 files changed, 2 insertions(+), 1 deletion(-) (limited to 'net/core/dev_ioctl.c') diff --git a/net/core/dev.h b/net/core/dev.h index 357543cbde65..aa91eed55a40 100644 --- a/net/core/dev.h +++ b/net/core/dev.h @@ -312,5 +312,6 @@ int dev_set_hwtstamp_phylib(struct net_device *dev, struct netlink_ext_ack *extack); int dev_get_hwtstamp_phylib(struct net_device *dev, struct kernel_hwtstamp_config *cfg); +int net_hwtstamp_validate(const struct kernel_hwtstamp_config *cfg); #endif diff --git a/net/core/dev_ioctl.c b/net/core/dev_ioctl.c index 67cf68817f23..1f09930fca26 100644 --- a/net/core/dev_ioctl.c +++ b/net/core/dev_ioctl.c @@ -184,7 +184,7 @@ static int dev_ifsioc_locked(struct net *net, struct ifreq *ifr, unsigned int cm return err; } -static int net_hwtstamp_validate(const struct kernel_hwtstamp_config *cfg) +int net_hwtstamp_validate(const struct kernel_hwtstamp_config *cfg) { enum hwtstamp_tx_types tx_type; enum hwtstamp_rx_filters rx_filter; -- cgit v1.2.3 From 35f7cad1743e04bf2944a2aadb6b6a42adc57bca Mon Sep 17 00:00:00 2001 From: Kory Maincent Date: Thu, 12 Dec 2024 18:06:43 +0100 Subject: net: Add the possibility to support a selected hwtstamp in netdevice Introduce the description of a hwtstamp provider, mainly defined with a the hwtstamp source and the phydev pointer. Add a hwtstamp provider description within the netdev structure to allow saving the hwtstamp we want to use. This prepares for future support of an ethtool netlink command to select the desired hwtstamp provider. By default, the old API that does not support hwtstamp selectability is used, meaning the hwtstamp provider pointer is unset. Signed-off-by: Kory Maincent Signed-off-by: David S. Miller --- drivers/net/phy/phy_device.c | 10 ++++++++ include/linux/net_tstamp.h | 29 +++++++++++++++++++++++ include/linux/netdevice.h | 4 ++++ include/uapi/linux/net_tstamp.h | 11 +++++++++ net/core/dev_ioctl.c | 41 ++++++++++++++++++++++++++++++-- net/core/timestamping.c | 52 +++++++++++++++++++++++++++++++++++++---- 6 files changed, 140 insertions(+), 7 deletions(-) (limited to 'net/core/dev_ioctl.c') diff --git a/drivers/net/phy/phy_device.c b/drivers/net/phy/phy_device.c index b26bb33cd1d4..1a908af4175b 100644 --- a/drivers/net/phy/phy_device.c +++ b/drivers/net/phy/phy_device.c @@ -32,6 +32,7 @@ #include #include #include +#include #include #include #include @@ -1998,6 +1999,15 @@ void phy_detach(struct phy_device *phydev) phy_suspend(phydev); if (dev) { + struct hwtstamp_provider *hwprov; + + hwprov = rtnl_dereference(dev->hwprov); + /* Disable timestamp if it is the one selected */ + if (hwprov && hwprov->phydev == phydev) { + rcu_assign_pointer(dev->hwprov, NULL); + kfree_rcu(hwprov, rcu_head); + } + phydev->attached_dev->phydev = NULL; phydev->attached_dev = NULL; phy_link_topo_del_phy(dev, phydev); diff --git a/include/linux/net_tstamp.h b/include/linux/net_tstamp.h index 662074b08c94..ff0758e88ea1 100644 --- a/include/linux/net_tstamp.h +++ b/include/linux/net_tstamp.h @@ -19,6 +19,33 @@ enum hwtstamp_source { HWTSTAMP_SOURCE_PHYLIB, }; +/** + * struct hwtstamp_provider_desc - hwtstamp provider description + * + * @index: index of the hwtstamp provider. + * @qualifier: hwtstamp provider qualifier. + */ +struct hwtstamp_provider_desc { + int index; + enum hwtstamp_provider_qualifier qualifier; +}; + +/** + * struct hwtstamp_provider - hwtstamp provider object + * + * @rcu_head: RCU callback used to free the struct. + * @source: source of the hwtstamp provider. + * @phydev: pointer of the phydev source in case a PTP coming from phylib + * @desc: hwtstamp provider description. + */ + +struct hwtstamp_provider { + struct rcu_head rcu_head; + enum hwtstamp_source source; + struct phy_device *phydev; + struct hwtstamp_provider_desc desc; +}; + /** * struct kernel_hwtstamp_config - Kernel copy of struct hwtstamp_config * @@ -31,6 +58,7 @@ enum hwtstamp_source { * copied the ioctl request back to user space * @source: indication whether timestamps should come from the netdev or from * an attached phylib PHY + * @qualifier: qualifier of the hwtstamp provider * * Prefer using this structure for in-kernel processing of hardware * timestamping configuration, over the inextensible struct hwtstamp_config @@ -43,6 +71,7 @@ struct kernel_hwtstamp_config { struct ifreq *ifr; bool copied_to_user; enum hwtstamp_source source; + enum hwtstamp_provider_qualifier qualifier; }; static inline void hwtstamp_config_to_kernel(struct kernel_hwtstamp_config *kernel_cfg, diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h index d917949bba03..2593019ad5b1 100644 --- a/include/linux/netdevice.h +++ b/include/linux/netdevice.h @@ -82,6 +82,7 @@ struct xdp_metadata_ops; struct xdp_md; struct ethtool_netdev_state; struct phy_link_topology; +struct hwtstamp_provider; typedef u32 xdp_features_t; @@ -2045,6 +2046,7 @@ enum netdev_reg_state { * * @neighbours: List heads pointing to this device's neighbours' * dev_list, one per address-family. + * @hwprov: Tracks which PTP performs hardware packet time stamping. * * FIXME: cleanup struct net_device such that network protocol info * moves out. @@ -2457,6 +2459,8 @@ struct net_device { struct hlist_head neighbours[NEIGH_NR_TABLES]; + struct hwtstamp_provider __rcu *hwprov; + u8 priv[] ____cacheline_aligned __counted_by(priv_len); } ____cacheline_aligned; diff --git a/include/uapi/linux/net_tstamp.h b/include/uapi/linux/net_tstamp.h index 858339d1c1c4..55b0ab51096c 100644 --- a/include/uapi/linux/net_tstamp.h +++ b/include/uapi/linux/net_tstamp.h @@ -13,6 +13,17 @@ #include #include /* for SO_TIMESTAMPING */ +/* + * Possible type of hwtstamp provider. Mainly "precise" the default one + * is for IEEE 1588 quality and "approx" is for NICs DMA point. + */ +enum hwtstamp_provider_qualifier { + HWTSTAMP_PROVIDER_QUALIFIER_PRECISE, + HWTSTAMP_PROVIDER_QUALIFIER_APPROX, + + HWTSTAMP_PROVIDER_QUALIFIER_CNT, +}; + /* SO_TIMESTAMPING flags */ enum { SOF_TIMESTAMPING_TX_HARDWARE = (1<<0), diff --git a/net/core/dev_ioctl.c b/net/core/dev_ioctl.c index 1f09930fca26..087a57b7e4fa 100644 --- a/net/core/dev_ioctl.c +++ b/net/core/dev_ioctl.c @@ -6,6 +6,7 @@ #include #include #include +#include #include #include #include @@ -269,6 +270,21 @@ static int dev_eth_ioctl(struct net_device *dev, int dev_get_hwtstamp_phylib(struct net_device *dev, struct kernel_hwtstamp_config *cfg) { + struct hwtstamp_provider *hwprov; + + hwprov = rtnl_dereference(dev->hwprov); + if (hwprov) { + cfg->qualifier = hwprov->desc.qualifier; + if (hwprov->source == HWTSTAMP_SOURCE_PHYLIB && + hwprov->phydev) + return phy_hwtstamp_get(hwprov->phydev, cfg); + + if (hwprov->source == HWTSTAMP_SOURCE_NETDEV) + return dev->netdev_ops->ndo_hwtstamp_get(dev, cfg); + + return -EOPNOTSUPP; + } + if (phy_is_default_hwtstamp(dev->phydev)) return phy_hwtstamp_get(dev->phydev, cfg); @@ -324,11 +340,32 @@ int dev_set_hwtstamp_phylib(struct net_device *dev, struct netlink_ext_ack *extack) { const struct net_device_ops *ops = dev->netdev_ops; - bool phy_ts = phy_is_default_hwtstamp(dev->phydev); struct kernel_hwtstamp_config old_cfg = {}; + struct hwtstamp_provider *hwprov; + struct phy_device *phydev; bool changed = false; + bool phy_ts; int err; + hwprov = rtnl_dereference(dev->hwprov); + if (hwprov) { + if (hwprov->source == HWTSTAMP_SOURCE_PHYLIB && + hwprov->phydev) { + phy_ts = true; + phydev = hwprov->phydev; + } else if (hwprov->source == HWTSTAMP_SOURCE_NETDEV) { + phy_ts = false; + } else { + return -EOPNOTSUPP; + } + + cfg->qualifier = hwprov->desc.qualifier; + } else { + phy_ts = phy_is_default_hwtstamp(dev->phydev); + if (phy_ts) + phydev = dev->phydev; + } + cfg->source = phy_ts ? HWTSTAMP_SOURCE_PHYLIB : HWTSTAMP_SOURCE_NETDEV; if (phy_ts && dev->see_all_hwtstamp_requests) { @@ -350,7 +387,7 @@ int dev_set_hwtstamp_phylib(struct net_device *dev, changed = kernel_hwtstamp_config_changed(&old_cfg, cfg); if (phy_ts) { - err = phy_hwtstamp_set(dev->phydev, cfg, extack); + err = phy_hwtstamp_set(phydev, cfg, extack); if (err) { if (changed) ops->ndo_hwtstamp_set(dev, &old_cfg, NULL); diff --git a/net/core/timestamping.c b/net/core/timestamping.c index 3717fb152ecc..a50a7ef49ae8 100644 --- a/net/core/timestamping.c +++ b/net/core/timestamping.c @@ -9,6 +9,7 @@ #include #include #include +#include static unsigned int classify(const struct sk_buff *skb) { @@ -21,19 +22,39 @@ static unsigned int classify(const struct sk_buff *skb) void skb_clone_tx_timestamp(struct sk_buff *skb) { + struct hwtstamp_provider *hwprov; struct mii_timestamper *mii_ts; + struct phy_device *phydev; struct sk_buff *clone; unsigned int type; - if (!skb->sk || !skb->dev || - !phy_is_default_hwtstamp(skb->dev->phydev)) + if (!skb->sk || !skb->dev) return; + rcu_read_lock(); + hwprov = rcu_dereference(skb->dev->hwprov); + if (hwprov) { + if (hwprov->source != HWTSTAMP_SOURCE_PHYLIB || + !hwprov->phydev) { + rcu_read_unlock(); + return; + } + + phydev = hwprov->phydev; + } else { + phydev = skb->dev->phydev; + if (!phy_is_default_hwtstamp(phydev)) { + rcu_read_unlock(); + return; + } + } + rcu_read_unlock(); + type = classify(skb); if (type == PTP_CLASS_NONE) return; - mii_ts = skb->dev->phydev->mii_ts; + mii_ts = phydev->mii_ts; if (likely(mii_ts->txtstamp)) { clone = skb_clone_sk(skb); if (!clone) @@ -45,12 +66,33 @@ EXPORT_SYMBOL_GPL(skb_clone_tx_timestamp); bool skb_defer_rx_timestamp(struct sk_buff *skb) { + struct hwtstamp_provider *hwprov; struct mii_timestamper *mii_ts; + struct phy_device *phydev; unsigned int type; - if (!skb->dev || !phy_is_default_hwtstamp(skb->dev->phydev)) + if (!skb->dev) return false; + rcu_read_lock(); + hwprov = rcu_dereference(skb->dev->hwprov); + if (hwprov) { + if (hwprov->source != HWTSTAMP_SOURCE_PHYLIB || + !hwprov->phydev) { + rcu_read_unlock(); + return false; + } + + phydev = hwprov->phydev; + } else { + phydev = skb->dev->phydev; + if (!phy_is_default_hwtstamp(phydev)) { + rcu_read_unlock(); + return false; + } + } + rcu_read_unlock(); + if (skb_headroom(skb) < ETH_HLEN) return false; @@ -63,7 +105,7 @@ bool skb_defer_rx_timestamp(struct sk_buff *skb) if (type == PTP_CLASS_NONE) return false; - mii_ts = skb->dev->phydev->mii_ts; + mii_ts = phydev->mii_ts; if (likely(mii_ts->rxtstamp)) return mii_ts->rxtstamp(mii_ts, skb, type); -- cgit v1.2.3 From be94cfdb993ff091ce3b98a16b45795198957f95 Mon Sep 17 00:00:00 2001 From: Kuniyuki Iwashima Date: Wed, 15 Jan 2025 18:55:45 +0900 Subject: dev: Hold rtnl_net_lock() for dev_ifsioc(). Basically, dev_ifsioc() operates on the passed single netns (except for netdev notifier chains with lower/upper devices for which we will need more changes). Let's hold rtnl_net_lock() for dev_ifsioc(). Now that NETDEV_CHANGENAME is always triggered under rtnl_net_lock() of the device's netns. (do_setlink() and dev_ifsioc()) Signed-off-by: Kuniyuki Iwashima Reviewed-by: Eric Dumazet Link: https://patch.msgid.link/20250115095545.52709-4-kuniyu@amazon.com Signed-off-by: Jakub Kicinski --- net/core/dev.c | 7 ++----- net/core/dev_ioctl.c | 26 +++++++++++++++++--------- net/core/rtnl_net_debug.c | 15 +++------------ 3 files changed, 22 insertions(+), 26 deletions(-) (limited to 'net/core/dev_ioctl.c') diff --git a/net/core/dev.c b/net/core/dev.c index 32fcf54b779d..fe5f5855593d 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -1348,16 +1348,13 @@ static int dev_get_valid_name(struct net *net, struct net_device *dev, */ int dev_change_name(struct net_device *dev, const char *newname) { + struct net *net = dev_net(dev); unsigned char old_assign_type; char oldname[IFNAMSIZ]; int err = 0; int ret; - struct net *net; - - ASSERT_RTNL(); - BUG_ON(!dev_net(dev)); - net = dev_net(dev); + ASSERT_RTNL_NET(net); if (!strncmp(newname, dev->name, IFNAMSIZ)) return 0; diff --git a/net/core/dev_ioctl.c b/net/core/dev_ioctl.c index 087a57b7e4fa..4c2098ac9d72 100644 --- a/net/core/dev_ioctl.c +++ b/net/core/dev_ioctl.c @@ -543,7 +543,7 @@ static int dev_siocwandev(struct net_device *dev, struct if_settings *ifs) } /* - * Perform the SIOCxIFxxx calls, inside rtnl_lock() + * Perform the SIOCxIFxxx calls, inside rtnl_net_lock() */ static int dev_ifsioc(struct net *net, struct ifreq *ifr, void __user *data, unsigned int cmd) @@ -620,11 +620,14 @@ static int dev_ifsioc(struct net *net, struct ifreq *ifr, void __user *data, return -ENODEV; if (!netif_is_bridge_master(dev)) return -EOPNOTSUPP; + netdev_hold(dev, &dev_tracker, GFP_KERNEL); - rtnl_unlock(); + rtnl_net_unlock(net); + err = br_ioctl_call(net, netdev_priv(dev), cmd, ifr, NULL); + netdev_put(dev, &dev_tracker); - rtnl_lock(); + rtnl_net_lock(net); return err; case SIOCDEVPRIVATE ... SIOCDEVPRIVATE + 15: @@ -770,9 +773,11 @@ int dev_ioctl(struct net *net, unsigned int cmd, struct ifreq *ifr, dev_load(net, ifr->ifr_name); if (!ns_capable(net->user_ns, CAP_NET_ADMIN)) return -EPERM; - rtnl_lock(); + + rtnl_net_lock(net); ret = dev_ifsioc(net, ifr, data, cmd); - rtnl_unlock(); + rtnl_net_unlock(net); + if (colon) *colon = ':'; return ret; @@ -816,9 +821,11 @@ int dev_ioctl(struct net *net, unsigned int cmd, struct ifreq *ifr, case SIOCBONDSLAVEINFOQUERY: case SIOCBONDINFOQUERY: dev_load(net, ifr->ifr_name); - rtnl_lock(); + + rtnl_net_lock(net); ret = dev_ifsioc(net, ifr, data, cmd); - rtnl_unlock(); + rtnl_net_unlock(net); + if (need_copyout) *need_copyout = false; return ret; @@ -841,9 +848,10 @@ int dev_ioctl(struct net *net, unsigned int cmd, struct ifreq *ifr, (cmd >= SIOCDEVPRIVATE && cmd <= SIOCDEVPRIVATE + 15)) { dev_load(net, ifr->ifr_name); - rtnl_lock(); + + rtnl_net_lock(net); ret = dev_ifsioc(net, ifr, data, cmd); - rtnl_unlock(); + rtnl_net_unlock(net); return ret; } return -ENOTTY; diff --git a/net/core/rtnl_net_debug.c b/net/core/rtnl_net_debug.c index f406045cbd0e..7ecd28cc1c22 100644 --- a/net/core/rtnl_net_debug.c +++ b/net/core/rtnl_net_debug.c @@ -27,7 +27,6 @@ static int rtnl_net_debug_event(struct notifier_block *nb, case NETDEV_CHANGEADDR: case NETDEV_PRE_CHANGEADDR: case NETDEV_GOING_DOWN: - case NETDEV_CHANGENAME: case NETDEV_FEAT_CHANGE: case NETDEV_BONDING_FAILOVER: case NETDEV_PRE_UP: @@ -60,18 +59,10 @@ static int rtnl_net_debug_event(struct notifier_block *nb, ASSERT_RTNL(); break; - /* Once an event fully supports RTNL_NET, move it here - * and remove "if (0)" below. - * - * case NETDEV_XXX: - * ASSERT_RTNL_NET(net); - * break; - */ - } - - /* Just to avoid unused-variable error for dev and net. */ - if (0) + case NETDEV_CHANGENAME: ASSERT_RTNL_NET(net); + break; + } return NOTIFY_DONE; } -- cgit v1.2.3