linux-kernel - Re: [PATCH v2 0/2] gpio-cdev: Release IRQ used by gpio-cdev on gpio chip removal

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: Sun, 3 Mar 2024 09:48:00 +0100
From: Linus Walleij <linus.walleij@...aro.org>
To: Bartosz Golaszewski <brgl@...ev.pl>
Cc: Herve Codina <herve.codina@...tlin.com>, Kent Gibson <warthog618@...il.com>, 
	Saravana Kannan <saravanak@...gle.com>, linux-gpio@...r.kernel.org, 
	linux-kernel@...r.kernel.org, Luca Ceresoli <luca.ceresoli@...tlin.com>, 
	Thomas Petazzoni <thomas.petazzoni@...tlin.com>
Subject: Re: [PATCH v2 0/2] gpio-cdev: Release IRQ used by gpio-cdev on gpio
 chip removal

On Sat, Mar 2, 2024 at 10:00 AM Bartosz Golaszewski <brgl@...ev.pl> wrote:
> On Fri, Mar 1, 2024 at 9:15 PM Linus Walleij <linus.walleij@...aro.org> wrote:

> > The problem is that I have never heard about a user unbinding/binding
> > a driver from sysfs for anything but debugging a drivers ability to
> > bind/unbind. Partly I feel that thing should just be moved
> > to debugfs given the usecase and because it just looks like a way for
> > attackers to provoke crashes given how some drivers look.
>
> That's not the only thing - device unbind can also be triggered by
> removing the module providing the driver which is a completely normal
> operation for user-space.

That is one thing, but every time I hear about "have you thought about
users going in and using bind/unbind in sysfs" it's for builtin bool
drivers.

I almost feel like bool drivers should have bind/unbind disabled by
default :/

The introduction of deferred probe has made the situation much better
because it tends to exercise the bind/unbind path a bit.

Yours,
Linus Walleij