From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753543AbdC1QjG (ORCPT ); Tue, 28 Mar 2017 12:39:06 -0400 Received: from mail-qt0-f193.google.com ([209.85.216.193]:34679 "EHLO mail-qt0-f193.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752238AbdC1QjE (ORCPT ); Tue, 28 Mar 2017 12:39:04 -0400 Subject: Re: [PATCH] ARM: orion5x: only call into phylib when available To: Arnd Bergmann , Jason Cooper , Andrew Lunn , Sebastian Hesselbarth , Gregory Clement References: <20170328101606.53008-1-arnd@arndb.de> Cc: "David S. Miller" , linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org From: Florian Fainelli Message-ID: <328515b4-e546-f66b-aecb-ab2c6168c56d@gmail.com> Date: Tue, 28 Mar 2017 09:38:35 -0700 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:45.0) Gecko/20100101 Thunderbird/45.7.0 MIME-Version: 1.0 In-Reply-To: <20170328101606.53008-1-arnd@arndb.de> Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Hi Arnd, On 03/28/2017 03:15 AM, Arnd Bergmann wrote: > Board code cannot call mdiobus_register_board_info() when phylib > or mdio_device is a loadable module: > > arch/arm/plat-orion/common.o: In function `orion_ge00_switch_init': > :(.init.text+0x474): undefined reference to `mdiobus_register_board_info' > > I had a number of ideas for how this could be solved, but after the MDIO > code got split out from PHYLIB it has gotten too hard, so I'm basically > giving up, and only call the mdiobus_register_board_info() function > if the MDIO layer is built-in to avoid the link error. This is similar > to how we handle PHY registration on other ARM platforms. Thanks for the fix, I still think we should be able to get mdio-boardinfo.o to be either built-in or modular, but more importantly, largely independent from MDIO_DEVICE and PHYLIB (and how they are selected). Your fix is definitively needed, but I agree with you that we should find something better (TBH, I was not expecting phy: Allow splitting MDIO bus/device support from PHYs to be merged that quickly). Thanks! > > Fixes: 90eff9096c01 ("net: phy: Allow splitting MDIO bus/device support from PHYs") > Fixes: 648ea0134069 ("net: phy: Allow pre-declaration of MDIO devices") > Signed-off-by: Arnd Bergmann > --- > arch/arm/mach-orion5x/Kconfig | 1 + > arch/arm/plat-orion/common.c | 5 +++++ > 2 files changed, 6 insertions(+) > > diff --git a/arch/arm/mach-orion5x/Kconfig b/arch/arm/mach-orion5x/Kconfig > index 468b8cb7fd5f..e3429c8c2e38 100644 > --- a/arch/arm/mach-orion5x/Kconfig > +++ b/arch/arm/mach-orion5x/Kconfig > @@ -6,6 +6,7 @@ menuconfig ARCH_ORION5X > select GPIOLIB > select MVEBU_MBUS > select PCI > + select PHYLIB if NETDEVICES > select PLAT_ORION_LEGACY > help > Support for the following Marvell Orion 5x series SoCs: > diff --git a/arch/arm/plat-orion/common.c b/arch/arm/plat-orion/common.c > index 9255b6d67ba5..aff6994950ba 100644 > --- a/arch/arm/plat-orion/common.c > +++ b/arch/arm/plat-orion/common.c > @@ -468,6 +468,7 @@ void __init orion_ge11_init(struct mv643xx_eth_platform_data *eth_data, > eth_data, &orion_ge11); > } > > +#ifdef CONFIG_ARCH_ORION5X > /***************************************************************************** > * Ethernet switch > ****************************************************************************/ > @@ -480,6 +481,9 @@ void __init orion_ge00_switch_init(struct dsa_chip_data *d) > struct mdio_board_info *bd; > unsigned int i; > > + if (!IS_BUILTIN(CONFIG_PHYLIB)) > + return; > + > for (i = 0; i < ARRAY_SIZE(d->port_names); i++) > if (!strcmp(d->port_names[i], "cpu")) > break; > @@ -493,6 +497,7 @@ void __init orion_ge00_switch_init(struct dsa_chip_data *d) > > mdiobus_register_board_info(&orion_ge00_switch_board_info, 1); > } > +#endif > > /***************************************************************************** > * I2C > -- Florian