net: Let the active time stamping layer be selectable.

Make the sysfs knob writable, and add checks in the ioctl and time
stamping paths to respect the currently selected time stamping layer.

Signed-off-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: Kory Maincent <kory.maincent@bootlin.com>

Documentation/ABI/testing/sysfs-class-net-timestamping

@@ -11,7 +11,10 @@ What:		/sys/class/net/<iface>/current_timestamping_provider
 Date:		January 2022
 Contact:	Richard Cochran <richardcochran@gmail.com>
 Description:
-		Show the current SO_TIMESTAMPING provider.
+		Shows or sets the current SO_TIMESTAMPING provider.
+		When changing the value, some packets in the kernel
+		networking stack may still be delivered with time
+		stamps from the previous provider.
 		The possible values are:
 		- "mac"  The MAC provides time stamping.
 		- "phy"  The PHY or MII device provides time stamping.

drivers/net/phy/phy_device.c

@@ -1451,6 +1451,11 @@ int phy_attach_direct(struct net_device *dev, struct phy_device *phydev,
 
 	phydev->phy_link_change = phy_link_change;
 	if (dev) {
+		if (phy_has_hwtstamp(phydev))
+			dev->selected_timestamping_layer = PHY_TIMESTAMPING;
+		else
+			dev->selected_timestamping_layer = MAC_TIMESTAMPING;
+
 		phydev->attached_dev = dev;
 		dev->phydev = phydev;
 
@@ -1774,6 +1779,7 @@ void phy_detach(struct phy_device *phydev)
 
 	phy_suspend(phydev);
 	if (dev) {
+		dev->selected_timestamping_layer = MAC_TIMESTAMPING;
 		phydev->attached_dev->phydev = NULL;
 		phydev->attached_dev = NULL;
 	}

include/linux/netdevice.h

@@ -1741,6 +1741,11 @@ enum netdev_ml_priv_type {
 	ML_PRIV_CAN,
 };
 
+enum timestamping_layer {
+	MAC_TIMESTAMPING,
+	PHY_TIMESTAMPING,
+};
+
 /**
  *	struct net_device - The DEVICE structure.
  *
@@ -1980,6 +1985,9 @@ enum netdev_ml_priv_type {
  *
  *	@threaded:	napi threaded mode is enabled
  *
+ *	@selected_timestamping_layer:	Tracks whether the MAC or the PHY
+ *					performs packet time stamping.
+ *
  *	@net_notifier_list:	List of per-net netdev notifier block
  *				that follow this device when it is moved
  *				to another network namespace.
@@ -2343,6 +2351,8 @@ struct net_device {
 	unsigned		wol_enabled:1;
 	unsigned		threaded:1;
 
+	enum timestamping_layer selected_timestamping_layer;
+
 	struct list_head	net_notifier_list;
 
 #if IS_ENABLED(CONFIG_MACSEC)

net/core/dev_ioctl.c

@@ -262,6 +262,43 @@ static int dev_eth_ioctl(struct net_device *dev,
 	return err;
 }
 
+static int dev_hwtstamp_ioctl(struct net_device *dev,
+			      struct ifreq *ifr, unsigned int cmd)
+{
+	const struct net_device_ops *ops = dev->netdev_ops;
+	int err;
+
+	err = dsa_ndo_eth_ioctl(dev, ifr, cmd);
+	if (err == 0 || err != -EOPNOTSUPP)
+		return err;
+
+	if (!netif_device_present(dev))
+		return -ENODEV;
+
+	switch (dev->selected_timestamping_layer) {
+
+	case MAC_TIMESTAMPING:
+		if (ops->ndo_do_ioctl == phy_do_ioctl) {
+			/* Some drivers set .ndo_do_ioctl to phy_do_ioctl. */
+			err = -EOPNOTSUPP;
+		} else {
+			err = ops->ndo_eth_ioctl(dev, ifr, cmd);
+		}
+		break;
+
+	case PHY_TIMESTAMPING:
+		if (phy_has_hwtstamp(dev->phydev)) {
+			err = phy_mii_ioctl(dev->phydev, ifr, cmd);
+		} else {
+			err = -ENODEV;
+			WARN_ON(1);
+		}
+		break;
+	}
+
+	return err;
+}
+
 static int dev_siocbond(struct net_device *dev,
 			struct ifreq *ifr, unsigned int cmd)
 {
@@ -397,6 +434,9 @@ static int dev_ifsioc(struct net *net, struct ifreq *ifr, void __user *data,
 			return err;
 		fallthrough;
 
+	case SIOCGHWTSTAMP:
+		return dev_hwtstamp_ioctl(dev, ifr, cmd);
+
 	/*
 	 *	Unknown or private ioctl
 	 */
@@ -407,9 +447,7 @@ static int dev_ifsioc(struct net *net, struct ifreq *ifr, void __user *data,
 
 		if (cmd == SIOCGMIIPHY ||
 		    cmd == SIOCGMIIREG ||
-		    cmd == SIOCSMIIREG ||
-		    cmd == SIOCSHWTSTAMP ||
-		    cmd == SIOCGHWTSTAMP) {
+		    cmd == SIOCSMIIREG) {
 			err = dev_eth_ioctl(dev, ifr, cmd);
 		} else if (cmd == SIOCBONDENSLAVE ||
 		    cmd == SIOCBONDRELEASE ||

net/core/net-sysfs.c

@@ -666,17 +666,59 @@ static ssize_t current_timestamping_provider_show(struct device *dev,
 	if (!rtnl_trylock())
 		return restart_syscall();
 
-	if (phy_has_tsinfo(phydev)) {
-		ret = sprintf(buf, "%s\n", "phy");
-	} else {
+	switch (netdev->selected_timestamping_layer) {
+	case MAC_TIMESTAMPING:
 		ret = sprintf(buf, "%s\n", "mac");
+		break;
+	case PHY_TIMESTAMPING:
+		ret = sprintf(buf, "%s\n", "phy");
+		break;
 	}
 
 	rtnl_unlock();
 
 	return ret;
 }
-static DEVICE_ATTR_RO(current_timestamping_provider);
+
+static ssize_t current_timestamping_provider_store(struct device *dev,
+						   struct device_attribute *attr,
+						   const char *buf, size_t len)
+{
+	struct net_device *netdev = to_net_dev(dev);
+	struct net *net = dev_net(netdev);
+	enum timestamping_layer flavor;
+
+	if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
+		return -EPERM;
+
+	if (sysfs_streq(buf, "mac"))
+		flavor = MAC_TIMESTAMPING;
+	else if (sysfs_streq(buf, "phy"))
+		flavor = PHY_TIMESTAMPING;
+	else
+		return -EINVAL;
+
+	if (!rtnl_trylock())
+		return restart_syscall();
+
+	if (!dev_isalive(netdev))
+		goto out;
+
+	if (netdev->selected_timestamping_layer != flavor) {
+		const struct net_device_ops *ops = netdev->netdev_ops;
+		struct ifreq ifr = {0};
+
+		/* Disable time stamping in the current layer. */
+		if (netif_device_present(netdev) && ops->ndo_eth_ioctl)
+			ops->ndo_eth_ioctl(netdev, &ifr, SIOCSHWTSTAMP);
+
+		netdev->selected_timestamping_layer = flavor;
+	}
+out:
+	rtnl_unlock();
+	return len;
+}
+static DEVICE_ATTR_RW(current_timestamping_provider);
 
 static struct attribute *net_class_attrs[] __ro_after_init = {
 	&dev_attr_netdev_group.attr,

net/core/timestamping.c

@@ -28,6 +28,9 @@ void skb_clone_tx_timestamp(struct sk_buff *skb)
 	if (!skb->sk)
 		return;
 
+	if (skb->dev->selected_timestamping_layer != PHY_TIMESTAMPING)
+		return;
+
 	type = classify(skb);
 	if (type == PTP_CLASS_NONE)
 		return;
@@ -50,6 +53,9 @@ bool skb_defer_rx_timestamp(struct sk_buff *skb)
 	if (!skb->dev || !skb->dev->phydev || !skb->dev->phydev->mii_ts)
 		return false;
 
+	if (skb->dev->selected_timestamping_layer != PHY_TIMESTAMPING)
+		return false;
+
 	if (skb_headroom(skb) < ETH_HLEN)
 		return false;
 

net/ethtool/common.c

@@ -629,10 +629,20 @@ int __ethtool_get_ts_info(struct net_device *dev, struct ethtool_ts_info *info)
 	memset(info, 0, sizeof(*info));
 	info->cmd = ETHTOOL_GET_TS_INFO;
 
-	if (phy_has_tsinfo(phydev))
-		return phy_ts_info(phydev, info);
-	if (ops->get_ts_info)
-		return ops->get_ts_info(dev, info);
+	switch (dev->selected_timestamping_layer) {
+
+	case MAC_TIMESTAMPING:
+		if (ops->get_ts_info)
+			return ops->get_ts_info(dev, info);
+		break;
+
+	case PHY_TIMESTAMPING:
+		if (phy_has_tsinfo(phydev)) {
+			return phy_ts_info(phydev, info);
+		}
+		WARN_ON(1);
+		return -ENODEV;
+	}
 
 	info->so_timestamping = SOF_TIMESTAMPING_RX_SOFTWARE |
 				SOF_TIMESTAMPING_SOFTWARE;