From mboxrd@z Thu Jan 1 00:00:00 1970 From: Daniel Mack Subject: [PATCH 0/3] mdio: Parse DT nodes for auto-probed PHYs Date: Sat, 10 May 2014 18:37:52 +0200 Message-ID: <1399739875-16434-1-git-send-email-zonque@gmail.com> Cc: netdev@vger.kernel.org, davem@davemloft.net, mugunthanvnm@ti.com, ujhelyi.m@gmail.com, Daniel Mack To: f.fainelli@gmail.com Return-path: Received: from mail-ee0-f43.google.com ([74.125.83.43]:42466 "EHLO mail-ee0-f43.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751132AbaEJQiE (ORCPT ); Sat, 10 May 2014 12:38:04 -0400 Received: by mail-ee0-f43.google.com with SMTP id d17so3474678eek.2 for ; Sat, 10 May 2014 09:38:02 -0700 (PDT) Sender: netdev-owner@vger.kernel.org List-ID: Hi Florian, here's another aproach of solving the issue with auto-probed buses and DT device nodes. In short, this patch set introduces a new hook which sets the of_node pointer of a phy device to a subnode of the bus, if there is any which matches the phy's address. This function is called from mdiobus_scan(), and is a no-op for !CONFIG_OF. Unlike with with the 1st version, dev->of_node is now available in the PHY driver's probe() callback already. Let me know what you think, I'm open to other ideas :) Daniel Daniel Mack (3): net: of_mdio: factor out code to parse a phy's 'reg' property net: of_mdio: add of_mdiobus_link_phydev() net: of_mdio: don't store the length of a property if we don't need to drivers/net/phy/mdio_bus.c | 6 ++++ drivers/of/of_mdio.c | 74 +++++++++++++++++++++++++++++++++++++--------- include/linux/of_mdio.h | 8 +++++ 3 files changed, 74 insertions(+), 14 deletions(-) -- 1.9.0