linux-kernel - Re: [PATCH v2] gpio: prevent potential speculation leaks in gpio_device_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-prev] [day] [month] [year] [list]

Message-ID: <20240517093922.GA337309@rigel>
Date: Fri, 17 May 2024 17:39:22 +0800
From: Kent Gibson <warthog618@...il.com>
To: Hagar Hemdan <hagarhem@...zon.com>
Cc: Norbert Manthey <nmanthey@...zon.de>,
	Linus Walleij <linus.walleij@...aro.org>,
	Bartosz Golaszewski <brgl@...ev.pl>, linux-gpio@...r.kernel.org,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2] gpio: prevent potential speculation leaks in
 gpio_device_get_desc()

On Fri, May 17, 2024 at 09:09:04AM +0000, Hagar Hemdan wrote:
> Users can call the gpio_ioctl() interface to get information about gpio
> chip lines.
> Lines on the chip are identified by an offset in the range
> of [0,chip.lines).
> Offset is copied from user and then used as an array index to get
> the gpio descriptor without sanitization.
>
> This change ensures that the offset is sanitized by
> "using array_index_nospec" to mitigate any possibility of speculative
> information leaks.
>

This could better describe the problem.  I'm still not 100% sure I
understand it, so it would be great if the comment could clarify it,
specifically what "speculation leaks" means.

And when referencing functions use (), so array_index_nospec(), rather
than quotes.

> This bug was discovered and resolved using Coverity Static Analysis
> Security Testing (SAST) by Synopsys, Inc.
>
> Fixes: aad955842d1c ("gpiolib: cdev: support GPIO_V2_GET_LINEINFO_IOCTL and GPIO_V2_GET_LINEINFO_WATCH_IOCTL")

This is not the correct commit(s) - the bug would've been present in the
character device uAPI since it was first added.
In fact two out of three places you patched in v1 pre-date this commit.

> Signed-off-by: Hagar Hemdan <hagarhem@...zon.com>
> ---
> v2: call array_index_nospec() after the bounds check.
> ---
>  drivers/gpio/gpiolib.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/gpio/gpiolib.c b/drivers/gpio/gpiolib.c
> index fa50db0c3605..b58e4fe78cec 100644
> --- a/drivers/gpio/gpiolib.c
> +++ b/drivers/gpio/gpiolib.c
> @@ -17,6 +17,7 @@
>  #include <linux/list.h>
>  #include <linux/lockdep.h>
>  #include <linux/module.h>
> +#include <linux/nospec.h>
>  #include <linux/of.h>
>  #include <linux/pinctrl/consumer.h>
>  #include <linux/seq_file.h>
> @@ -201,7 +202,7 @@ gpio_device_get_desc(struct gpio_device *gdev, unsigned int hwnum)
>  	if (hwnum >= gdev->ngpio)
>  		return ERR_PTR(-EINVAL);
>
> -	return &gdev->descs[hwnum];
> +	return &gdev->descs[array_index_nospec(hwnum, gdev->ngpio)];
>  }
>  EXPORT_SYMBOL_GPL(gpio_device_get_desc);
>

That makes more sense to me, so I no problem with the code change.

Cheers,
Kent.