linux-kernel - Re: Race condition in kobject_get

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-next>] [day] [month] [year] [list]

Message-ID: <YkQhc5ZjWe35T8fU@kroah.com>
Date:   Wed, 30 Mar 2022 11:22:59 +0200
From:   Greg KH <gregkh@...uxfoundation.org>
To:     Mukesh Ojha <quic_mojha@...cinc.com>
Cc:     "Rafael J. Wysocki" <rafael@...nel.org>,
        linux-kernel@...r.kernel.org
Subject: Re: Race condition in kobject_get_path()

On Wed, Mar 30, 2022 at 02:47:20PM +0530, Mukesh Ojha wrote:
> Hi All,
> 
> We are facing one issue where one driver is trying to register its device
> from driver probe
> and from another path  dev_set_name() done from driver probe of the added
> device whose
> length can be more than earlier which could result in redzone overwritten
> issue.

What drivers are doing this so we can see the code?  This should be
happening in order and not at the same time.

> 
> Can we get your suggestion here ? is this case of a race here ?
> 
> 
> p1 p2
> 
> *kobject_get_path* <https://opengrok.qualcomm.com/source/s?refs=kobject_get_path&project=KERNEL.PLATFORM.1.0>
> 
> 
>                             =>  get_kobj_path_length  (length is calculated
> from this path)
>                                dev_set_name(same dev name)
>                                  => fill_kobj_path  (path is copied here)

I do not understand this trace, sorry.  What code are you referring to?

confused,

greg k-h