Thread (15 messages) 15 messages, 4 authors, 2025-01-08

Re: [PATCH net-next v5 2/8] net: ethtool: plumb PHY stats to PHY drivers

flat view

From: Jakub Kicinski <kuba@kernel.org>
Date: 2025-01-08 02:02:18
Also in: linux-doc, lkml

On Mon,  6 Jan 2025 09:32:55 +0100 Oleksij Rempel wrote:
quoted hunk ↗ jump to hunk
+	/**
+	 * @get_link_stats: Retrieve link statistics.
+	 * @dev: The PHY device for which the statistics are retrieved.
+	 * @link_stats: structure where link-specific stats will be stored.
+	 *
+	 * Retrieves link-related statistics for the given PHY device. The input
+	 * structure is pre-initialized with `ETHTOOL_STAT_NOT_SET`, and the
+	 * driver must only modify members corresponding to supported
+	 * statistics. Unmodified members will remain set to
+	 * `ETHTOOL_STAT_NOT_SET` and will not be returned to userspace.
+	 *
+	 * Return: 0 on success or a negative error code on failure.
These get callbacks return void
quoted hunk ↗ jump to hunk
+	 */
+	void (*get_link_stats)(struct phy_device *dev,
+			       struct ethtool_link_ext_stats *link_stats);
 	/** @get_sset_count: Number of statistic counters */
 	int (*get_sset_count)(struct phy_device *dev);
 	/** @get_strings: Names of the statistic counters */
@@ -1777,6 +1810,49 @@ static inline bool phy_is_pseudo_fixed_link(struct phy_device *phydev)
 	return phydev->is_pseudo_fixed_link;
 }
 
+/**
+ * phy_ethtool_get_phy_stats - Retrieve standardized PHY statistics
+ * @phydev: Pointer to the PHY device
+ * @phy_stats: Pointer to ethtool_eth_phy_stats structure
+ * @phydev_stats: Pointer to ethtool_phy_stats structure
+ *
+ * Fetches PHY statistics using a kernel-defined interface for consistent
+ * diagnostics. Unlike phy_ethtool_get_stats(), which allows custom stats,
+ * this function enforces a standardized format for better interoperability.
+ */
+static inline void phy_ethtool_get_phy_stats(struct phy_device *phydev,
+					struct ethtool_eth_phy_stats *phy_stats,
+					struct ethtool_phy_stats *phydev_stats)
+{
+	if (!phydev->drv || !phydev->drv->get_phy_stats)
+		return;
+
+	mutex_lock(&phydev->lock);
+	phydev->drv->get_phy_stats(phydev, phy_stats, phydev_stats);
+	mutex_unlock(&phydev->lock);
+}
+
+/**
+ * phy_ethtool_get_link_ext_stats - Retrieve extended link statistics for a PHY
+ * @phydev: Pointer to the PHY device
+ * @link_stats: Pointer to the structure to store extended link statistics
+ *
+ * Populates the ethtool_link_ext_stats structure with link down event counts
+ * and additional driver-specific link statistics, if available.
+ */
+static inline void phy_ethtool_get_link_ext_stats(struct phy_device *phydev,
+				    struct ethtool_link_ext_stats *link_stats)
+{
+	link_stats->link_down_events = READ_ONCE(phydev->link_down_events);
+
+	if (!phydev->drv || !phydev->drv->get_link_stats)
+		return;
+
+	mutex_lock(&phydev->lock);
+	phydev->drv->get_link_stats(phydev, link_stats);
+	mutex_unlock(&phydev->lock);
+}
Do these have to be static inlines?

Seemds like it will just bloat the header, and make alignment more
painful.
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help