of_phy_connect(9)

💖 of_phy_connect(9) 💖

okay bestie, so this is a page from the Kernel API Manual (section 9, the kernel developer one, iykyk) and it's literally about wiring your network device up to a PHY that the device tree told you about. it's giving "matchmaker but for hardware" 🎀

---

✨ NAME ✨

of_phy_connect — connect to the phy described in the device tree 🦋

---

🌸 SYNOPSIS 🌸

struct phy_device * of_phy_connect (struct net_device *dev , struct
       device_node *phy_np , void (*hndlr)(struct net_device *) , u32 flags ,
       phy_interface_t iface );

---

💅 ARGUMENTS 💅

• dev — pointer to the net_device that's claiming the phy. like, this is the one putting a ring on it 💍
• phy_np — pointer to the device tree node for the PHY. your PHY's little birth certificate, fr fr 🌳
• hndlr — the link state callback for the network device. so it can hit you up when the link status changes 📞
• flags — flags to pass to the PHY 🚩
• iface — the PHY data interface type. the whole vibe check of how data flows 🔌

---

👑 DESCRIPTION 👑

so here's the tea: if it works, you get back a pointer to the phy_device, and the embedded struct device refcount is incremented by one ✨

if it doesn't work? you get NULL. no cap, that's the failure case 😭

⚠️ bestie listen — you must drop that refcount by calling phy_disconnect or phy_detach. don't just leave it hanging, that's leaked refcount energy and it's not cute 💀

---

🔥 SEE ALSO 🔥

• Kernel file ./drivers/net/mdio/of_mdio.c
• __of_mdiobus_register(9)
• of_mdio_find_device(9)
• of_phy_find_device(9)
• of_phy_get_and_connect(9)

---

and that's the whole page, periodt 💜 this lil icon lives over in ./drivers/net/mdio, dated October 2026 — go off bestie, may your refcounts always balance 🫶🌸