netdev - Re: [PATCH] rfkill: allocate static minor

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, 24 Oct 2019 22:23:57 +0200
From:   Marcel Holtmann <marcel@...tmann.org>
To:     Greg KH <gregkh@...uxfoundation.org>
Cc:     Arnd Bergmann <arnd@...db.de>,
        Johannes Berg <johannes@...solutions.net>,
        "David S. Miller" <davem@...emloft.net>,
        linux-kernel@...r.kernel.org, linux-wireless@...r.kernel.org,
        netdev@...r.kernel.org
Subject: Re: [PATCH] rfkill: allocate static minor

Hi Greg,

>> udev has a feature of creating /dev/<node> device-nodes if it finds
>> a devnode:<node> modalias. This allows for auto-loading of modules that
>> provide the node. This requires to use a statically allocated minor
>> number for misc character devices.
>> 
>> However, rfkill uses dynamic minor numbers and prevents auto-loading
>> of the module. So allocate the next static misc minor number and use
>> it for rfkill.
> 
> As rfkill has been around for a long time, what new use case is needing
> to auto-load this based on a major number?

we have bug reports from iwd users where it fails opening /dev/rfkill. Since iwd can be actually started before the WiFi hardware is fully probed and all its drivers are loaded, we have a race-condition here if rfkill is not capable of auto-loading.

The difference is really that iwd is a fully self-contained WiFi daemon compared to wpa_supplicant which is just some sort of helper. iwd is fully hot plug capable as well compared to wpa_supplicant. It looks like this is exposing the race condition for our users. Frankly, we should have fixed rfkill a long time ago when we fixed uinput, uhid etc, but seems we forgot it. I assume mainly because it magically got loaded in time by some module dependencies.

Regards

Marcel