linux-kernel - Re: [PATCH] gpio: Add device link support

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] [day] [month] [year] [list]

Message-ID: <CACRpkdbKCOgJDYWx0hJ+tYyimXbShMNV00oHsVAskeJm1kT0Jg@mail.gmail.com>
Date:   Tue, 25 Jun 2019 00:34:30 +0200
From:   Linus Walleij <linus.walleij@...aro.org>
To:     Thierry Reding <thierry.reding@...il.com>
Cc:     Bartosz Golaszewski <bgolaszewski@...libre.com>,
        "open list:GPIO SUBSYSTEM" <linux-gpio@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] gpio: Add device link support

On Fri, Jun 21, 2019 at 5:23 PM Thierry Reding <thierry.reding@...il.com> wrote:

> From: Thierry Reding <treding@...dia.com>
>
> Create a device link between a GPIO consumer and the GPIO supplier (i.e.
> the GPIO chip's parent). This makes the driver core aware of the
> dependency between the two devices so that it can properly order the
> suspend/resume sequence.
>
> As a side-effect the GPIO consumer will be force unloaded when the GPIO
> supplier goes away, which prevents the consumer from accessing dangling
> GPIOs.
>
> Signed-off-by: Thierry Reding <treding@...dia.com>

I was thinking we were gonna see this soon :D

> +void gpiod_add_device_link(struct gpio_desc *desc, struct device *consumer)
> +{
> +       struct device_link *link;
> +
> +       link = device_link_add(consumer, desc->gdev->dev.parent,
> +                              DL_FLAG_AUTOREMOVE_CONSUMER);

I think this needs a NULL check.

Some GPIO chips still have NULL as parent here, they just
register a gpio_chip right out of boardfile code without any
corresponding device. MIPS does this for example IIRC.

> +               device_link_remove(desc->consumer, desc->gdev->dev.parent);

And same on remove.

But why not use desc->gdev->dev as link, simply?
That device has the same lifetime as the gpio_chip, sometimes
even longer I think, if e.g. there is userspace holding the
GPIOs.

Yours,
Linus Walleij