Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Thu, 20 Dec 2018 08:14:05 +0100
From:   Daniel Vetter <daniel@ffwll.ch>
To:     "Gustavo A. R. Silva" <gustavo@embeddedor.com>
Cc:     Maarten Lankhorst <maarten.lankhorst@linux.intel.com>,
        Maxime Ripard <maxime.ripard@bootlin.com>,
        Sean Paul <sean@poorly.run>, David Airlie <airlied@linux.ie>,
        Daniel Vetter <daniel@ffwll.ch>,
        dri-devel@lists.freedesktop.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] drm/ioctl: Fix Spectre v1 vulnerabilities
Message-ID: <20181220071404.GD21184@phenom.ffwll.local>
Mail-Followup-To: "Gustavo A. R. Silva" <gustavo@embeddedor.com>,
        Maarten Lankhorst <maarten.lankhorst@linux.intel.com>,
        Maxime Ripard <maxime.ripard@bootlin.com>,
        Sean Paul <sean@poorly.run>, David Airlie <airlied@linux.ie>,
        dri-devel@lists.freedesktop.org, linux-kernel@vger.kernel.org
References: <20181220000015.GA18973@embeddedor>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20181220000015.GA18973@embeddedor>
User-Agent: Mutt/1.10.1 (2018-07-13)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Wed, Dec 19, 2018 at 06:00:15PM -0600, Gustavo A. R. Silva wrote:
> nr is indirectly controlled by user-space, hence leading to a
> potential exploitation of the Spectre variant 1 vulnerability.
> 
> This issue was detected with the help of Smatch:
> 
> drivers/gpu/drm/drm_ioctl.c:805 drm_ioctl() warn: potential spectre issue 'dev->driver->ioctls' [r]
> drivers/gpu/drm/drm_ioctl.c:810 drm_ioctl() warn: potential spectre issue 'drm_ioctls' [r] (local cap)
> drivers/gpu/drm/drm_ioctl.c:892 drm_ioctl_flags() warn: potential spectre issue 'drm_ioctls' [r] (local cap)
> 
> Fix this by sanitizing nr before using it to index dev->driver->ioctls
> and drm_ioctls.
> 
> Notice that given that speculation windows are large, the policy is
> to kill the speculation on the first load and not worry if it can be
> completed with a dependent load/store [1].
> 
> [1] https://marc.info/?l=linux-kernel&m=152449131114778&w=2
> 
> Cc: stable@vger.kernel.org
> Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>

lgtm and I think there's no other obvious place where we need
array_index_nospec in drm core. Applied to drm-misc-fixes.
-Daniel

> ---
>  drivers/gpu/drm/drm_ioctl.c | 10 ++++++++--
>  1 file changed, 8 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/gpu/drm/drm_ioctl.c b/drivers/gpu/drm/drm_ioctl.c
> index 94bd872d56c4..7e6746b2d704 100644
> --- a/drivers/gpu/drm/drm_ioctl.c
> +++ b/drivers/gpu/drm/drm_ioctl.c
> @@ -37,6 +37,7 @@
>  
>  #include <linux/pci.h>
>  #include <linux/export.h>
> +#include <linux/nospec.h>
>  
>  /**
>   * DOC: getunique and setversion story
> @@ -800,13 +801,17 @@ long drm_ioctl(struct file *filp,
>  
>  	if (is_driver_ioctl) {
>  		/* driver ioctl */
> -		if (nr - DRM_COMMAND_BASE >= dev->driver->num_ioctls)
> +		unsigned int index = nr - DRM_COMMAND_BASE;
> +
> +		if (index >= dev->driver->num_ioctls)
>  			goto err_i1;
> -		ioctl = &dev->driver->ioctls[nr - DRM_COMMAND_BASE];
> +		index = array_index_nospec(index, dev->driver->num_ioctls);
> +		ioctl = &dev->driver->ioctls[index];
>  	} else {
>  		/* core ioctl */
>  		if (nr >= DRM_CORE_IOCTL_COUNT)
>  			goto err_i1;
> +		nr = array_index_nospec(nr, DRM_CORE_IOCTL_COUNT);
>  		ioctl = &drm_ioctls[nr];
>  	}
>  
> @@ -888,6 +893,7 @@ bool drm_ioctl_flags(unsigned int nr, unsigned int *flags)
>  
>  	if (nr >= DRM_CORE_IOCTL_COUNT)
>  		return false;
> +	nr = array_index_nospec(nr, DRM_CORE_IOCTL_COUNT);
>  
>  	*flags = drm_ioctls[nr].flags;
>  	return true;
> -- 
> 2.20.1
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch