Re: [PATCH net-next 08/13] net: phylink: Represent PHY-less SFP modules with phy_port
From: Romain Gantois <romain.gantois@bootlin.com>
Date: 2026-01-29 08:58:50
Also in:
lkml
On Tuesday, 27 January 2026 14:41:56 CET Maxime Chevallier wrote: ...
quoted hunk ↗ jump to hunk
@@ -1786,13 +1787,31 @@ static int phylink_create_sfp_port(struct phylink*pl) else pl->sfp_bus_port = port; + if (pl->mod_port) { + ret = phy_link_topo_add_port(pl->netdev, pl->mod_port); + if (ret) + goto out_bus_port; + } + + return 0; +out_bus_port: + phy_link_topo_del_port(pl->netdev, port);
This seems strange to me. Why clean up after phy_link_topo_add_port() if it returned an error code? Presumably phy_link_topo_add_port() cleans up after itself if it encounters an error doesn't it?
quoted hunk ↗ jump to hunk
+ phy_port_destroy(port); return ret; } static void phylink_destroy_sfp_port(struct phylink *pl) { - if (pl->netdev && pl->sfp_bus_port) - phy_link_topo_del_port(pl->netdev, pl->sfp_bus_port); + if (pl->netdev) { + if (pl->sfp_bus_port) + phy_link_topo_del_port(pl->netdev, pl->sfp_bus_port); + + /* Only remove it from the topology, it will be destroyed at + * module removal. + */ + if (pl->mod_port) + phy_link_topo_del_port(pl->netdev, pl->mod_port); + } if (pl->sfp_bus_port) phy_port_destroy(pl->sfp_bus_port);@@ -3998,6 +4017,49 @@ static void phylink_sfp_disconnect_phy(void*upstream, phylink_disconnect_phy(upstream); } +static int phylink_sfp_connect_nophy(void *upstream)
I'd name this "phylink_sfp_connect_no_phy" just to keep the name formatting consistent. Thanks, -- Romain Gantois, Bootlin Embedded Linux and Kernel engineering https://bootlin.com
Attachments
- signature.asc [application/pgp-signature] 833 bytes