netdev - Re: [PATCH net 1/1] mdio_bus: fix mdio_register_device when RESET

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:   Thu, 21 Nov 2019 09:38:49 +0100
From:   David Bauer <mail@...id-bauer.net>
To:     Andrew Lunn <andrew@...n.ch>,
        Geert Uytterhoeven <geert@...ux-m68k.org>
Cc:     Andy Shevchenko <andriy.shevchenko@...ux.intel.com>,
        Marek Behún <marek.behun@....cz>,
        Dmitry Torokhov <dmitry.torokhov@...il.com>,
        Florian Fainelli <f.fainelli@...il.com>,
        "David S. Miller" <davem@...emloft.net>,
        Sergei Shtylyov <sergei.shtylyov@...entembedded.com>,
        netdev@...r.kernel.org, linux-renesas-soc@...r.kernel.org
Subject: Re: [PATCH net 1/1] mdio_bus: fix mdio_register_device when
 RESET_CONTROLLER is disabled

Hello Andrew,

On 11/21/19 3:08 AM, Andrew Lunn wrote:
>> The difference with the non-optional case is that
>> __devm_reset_control_get() registers a cleanup function if there's
>> no error condition, even for NULL (which is futile, will send a patch).
>>
>> However, more importantly, mdiobus_register_reset() calls a devm_*()
>> function on "&mdiodev->dev" ("mdio_bus ee700000.ethernet-ffffffff:01"),
>> which is a different device than the one being probed
>> (("ee700000.ethernet"), see also the callstack below).
>> In fact "&mdiodev->dev" hasn't been probed yet, leading to the WARNING
>> when it is probed later.
>>
>>     [<c0582de8>] (mdiobus_register_device) from [<c05810e0>] (phy_device_register+0xc/0x74)
>>     [<c05810e0>] (phy_device_register) from [<c0675ef4>] (of_mdiobus_register_phy+0x144/0x17c)
>>     [<c0675ef4>] (of_mdiobus_register_phy) from [<c06760f0>] (of_mdiobus_register+0x1c4/0x2d0)
>>     [<c06760f0>] (of_mdiobus_register) from [<c0589f0c>] (sh_eth_drv_probe+0x778/0x8ac)
>>     [<c0589f0c>] (sh_eth_drv_probe) from [<c0516ce8>] (platform_drv_probe+0x48/0x94)
>>
>> Has commit 71dd6c0dff51b5f1 ("net: phy: add support for
>> reset-controller") been tested with an actual reset present?

I'm using it on a AR9132 board, however the mdio bus is probed before the
ethernet driver, hence why i was not experiencing this misbehavior.

>>
>> Are Ethernet drivers not (no longer) allowed to register MDIO busses?
> 
> That is not good. The devm_reset_control_get() call need replaces with
> an unmanaged version, and a call to reset_control_put() added to
> mdiobus_unregister_device().
> 
> David, could you look at this, it was a patch from you that added
> this.

I will prepare patches to fix this bug.

Best wishes
David

> 
> 	Andrew
>