netdev - Re: [PATCH] net: ethernet: Add missing depends on MDIO

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Wed, 12 Apr 2023 15:52:52 -0700
From:   Jacob Keller <jacob.e.keller@...el.com>
To:     Andrew Lunn <andrew@...n.ch>,
        Simon Horman <simon.horman@...igine.com>
CC:     David Miller <davem@...emloft.net>,
        Jakub Kicinski <kuba@...nel.org>,
        netdev <netdev@...r.kernel.org>,
        Florian Fainelli <f.fainelli@...il.com>,
        Heiner Kallweit <hkallweit1@...il.com>,
        Russell King <rmk+kernel@...linux.org.uk>
Subject: Re: [PATCH] net: ethernet: Add missing depends on MDIO_DEVRES



On 4/10/2023 5:11 AM, Andrew Lunn wrote:
> On Mon, Apr 10, 2023 at 11:07:58AM +0200, Simon Horman wrote:
>> On Sun, Apr 09, 2023 at 05:02:04PM +0200, Andrew Lunn wrote:
>>> A number of MDIO drivers make use of devm_mdiobus_alloc_size(). This
>>> is only available when CONFIG_MDIO_DEVRES is enabled. Add missing
>>> depends or selects, depending on if there are circular dependencies or
>>> not. This avoids linker errors, especially for randconfig builds.
>>>
>>> Signed-off-by: Andrew Lunn <andrew@...n.ch>
>>> ---
>>>  drivers/net/ethernet/freescale/Kconfig       | 1 +
>>>  drivers/net/ethernet/freescale/enetc/Kconfig | 1 +
>>>  drivers/net/ethernet/marvell/Kconfig         | 1 +
>>>  drivers/net/ethernet/qualcomm/Kconfig        | 1 +
>>>  drivers/net/mdio/Kconfig                     | 3 +++
>>>  5 files changed, 7 insertions(+)
>>>
>>> diff --git a/drivers/net/ethernet/freescale/Kconfig b/drivers/net/ethernet/freescale/Kconfig
>>> index f1e80d6996ef..1c78f66a89da 100644
>>> --- a/drivers/net/ethernet/freescale/Kconfig
>>> +++ b/drivers/net/ethernet/freescale/Kconfig
>>> @@ -71,6 +71,7 @@ config FSL_XGMAC_MDIO
>>>  	tristate "Freescale XGMAC MDIO"
>>>  	select PHYLIB
>>>  	depends on OF
>>> +	select MDIO_DEVRES
>>>  	select OF_MDIO
>>>  	help
>>>  	  This driver supports the MDIO bus on the Fman 10G Ethernet MACs, and
>>
>> Perhaps this is a good idea, but I'd like to mention that I don't think
>> it is strictly necessary as:
>>
>> 1. FSL_XGMAC_MDIO selects PHYLIB.
>> 2. And PHYLIB selects MDIO_DEVRES.
>>
>> Likewise for FSL_ENETC, MV643XX_ETH, QCOM_EMAC.
>>
>> Is there some combination of N/y/m that defeats my logic here?
>> I feel like I am missing something obvious.
> 
> I keep getting 0-day randconfig build warning about kernel
> configuration which don't link. It seems to get worse when we add in
> support of MAC and PHY LEDs. My guess is, the additional dependencies
> for LEDs upsets the conflict resolution engine, and it comes out with
> a different solution. `select` is a soft dependency. It is more a
> hint, and can be ignored. And when a randconfig kernel fails to build,
> MDIO_DEVRES is disabled.
> 
> Where possible, i've added a `depends on`, which is a much stronger
> dependency. But that can lead to circular dependencies, which kconfig
> cannot handle. In such cases, i've added selects. Maybe having more
> selects for a config option will influence it to find a solution which
> has MDIO_DEVRES enabled?
> 

I think this approach is ok. Other alternatives to solve the circular
dependency are much more invasive.

> I've had this patch in a github tree for a week or more, and 0-day has
> not yet returned any randconfig build errors. But i've not combined it
> with the LED code.
> 
>      Andrew