Re: [PATCH] drm: refuse ADDFB2 ioctl for broken bigendian drivers

From: Daniel Vetter <daniel@ffwll.ch>
To: Gerd Hoffmann <kraxel@redhat.com>
Cc: dri-devel@lists.freedesktop.org, David Airlie <airlied@linux.ie>,
	open list <linux-kernel@vger.kernel.org>,
	Sean Paul <sean@poorly.run>
Subject: Re: [PATCH] drm: refuse ADDFB2 ioctl for broken bigendian drivers
Date: Thu, 6 Sep 2018 09:32:36 +0200	[thread overview]
Message-ID: <20180906073236.GL4929@phenom.ffwll.local> (raw)
In-Reply-To: <20180906072850.11150-1-kraxel@redhat.com>

On Thu, Sep 06, 2018 at 09:28:50AM +0200, Gerd Hoffmann wrote:
> Drivers must set the quirk_addfb_prefer_host_byte_order quirk to make
> the drm_mode_addfb() compat code work correctly on bigendian machines.
> 
> If they don't they interpret pixel_format values incorrectly for bug
> compatibility, which in turn implies the ADDFB2 ioctl does not work
> correctly then.  So block it to make userspace fallback to ADDFB.
> 
> Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
> ---
>  drivers/gpu/drm/drm_ioctl.c | 25 ++++++++++++++++++++++++-
>  1 file changed, 24 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/drm_ioctl.c b/drivers/gpu/drm/drm_ioctl.c
> index ea10e9a26a..08e48bda23 100644
> --- a/drivers/gpu/drm/drm_ioctl.c
> +++ b/drivers/gpu/drm/drm_ioctl.c
> @@ -469,6 +469,29 @@ static int drm_copy_field(char __user *buf, size_t *buf_len, const char *value)
>  	return 0;
>  }
>  
> +static int drm_mode_addfb2_check(struct drm_device *dev,
> +				 void *data, struct drm_file *file_priv)
> +{
> +#ifdef __BIG_ENDIAN
> +	if (!dev->mode_config.quirk_addfb_prefer_host_byte_order) {
> +		/*
> +		 * Drivers must set the
> +		 * quirk_addfb_prefer_host_byte_order quirk to make
> +		 * the drm_mode_addfb() compat code work correctly on
> +		 * bigendian machines.
> +		 *
> +		 * If they don't they interpret pixel_format values
> +		 * incorrectly for bug compatibility, which in turn
> +		 * implies the ADDFB2 ioctl does not work correctly
> +		 * then.  So block it to make userspace fallback to
> +		 * ADDFB.
> +		 */

DRM_DEBUG_KMS would be good here I think, just to help userspace people
figure out why they got an -EINVAL.

> +		return -EINVAL;
> +	}
> +#endif
> +	return drm_mode_addfb2(dev, data, file_priv);

I think it'd be better if we have all the addfb code in drm_framebuffer.c.
So leave the function name, and extract the common code into
__drm_mode_addfb2 helper.

With these two nits addressed:

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>

> +}
> +
>  /*
>   * Get version information
>   *
> @@ -645,7 +668,7 @@ static const struct drm_ioctl_desc drm_ioctls[] = {
>  	DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETPROPBLOB, drm_mode_getblob_ioctl, DRM_UNLOCKED),
>  	DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETFB, drm_mode_getfb, DRM_UNLOCKED),
>  	DRM_IOCTL_DEF(DRM_IOCTL_MODE_ADDFB, drm_mode_addfb_ioctl, DRM_UNLOCKED),
> -	DRM_IOCTL_DEF(DRM_IOCTL_MODE_ADDFB2, drm_mode_addfb2, DRM_UNLOCKED),
> +	DRM_IOCTL_DEF(DRM_IOCTL_MODE_ADDFB2, drm_mode_addfb2_check, DRM_UNLOCKED),
>  	DRM_IOCTL_DEF(DRM_IOCTL_MODE_RMFB, drm_mode_rmfb_ioctl, DRM_UNLOCKED),
>  	DRM_IOCTL_DEF(DRM_IOCTL_MODE_PAGE_FLIP, drm_mode_page_flip_ioctl, DRM_MASTER|DRM_UNLOCKED),
>  	DRM_IOCTL_DEF(DRM_IOCTL_MODE_DIRTYFB, drm_mode_dirtyfb_ioctl, DRM_MASTER|DRM_UNLOCKED),
> -- 
> 2.9.3
> 
> _______________________________________________
> dri-devel mailing list
> dri-devel@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/dri-devel

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