linux-kernel - Re: [PATCH 2/2] media: coda: be more flexible wrt jpeg dimensions

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]

Message-ID: <c1fbc7912187302fdc27d9e333fab43b0c253507.camel@collabora.com>
Date:   Mon, 23 Mar 2020 11:27:15 -0300
From:   Ezequiel Garcia <ezequiel@...labora.com>
To:     Adrian Ratiu <adrian.ratiu@...labora.com>,
        Philipp Zabel <p.zabel@...gutronix.de>
Cc:     Mauro Carvalho Chehab <mchehab@...nel.org>,
        linux-media@...r.kernel.org, linux-kernel@...r.kernel.org,
        kernel@...labora.com, kernel@...gutronix.de,
        Tim Harvey <tharvey@...eworks.com>
Subject: Re: [PATCH 2/2] media: coda: be more flexible wrt jpeg dimensions

On Mon, 2020-03-23 at 15:09 +0200, Adrian Ratiu wrote:
> Don't require jpeg dimensions to exactly match format dimensions,
> so we are able to decode and display a wider range jpegs instead
> of outright rejecting the ones which don't match.
> 
> This is useful in applications which pass jpegs with arbitrary
> dimensions, where buffers can be reused to decode smaller jpegs
> without having to do expensive renegotiations.
> 
> Signed-off-by: Adrian Ratiu <adrian.ratiu@...labora.com>
> ---
>  drivers/media/platform/coda/coda-jpeg.c | 7 -------
>  1 file changed, 7 deletions(-)
> 
> diff --git a/drivers/media/platform/coda/coda-jpeg.c b/drivers/media/platform/coda/coda-jpeg.c
> index 162ba28a6b95..782a78dcaf4d 100644
> --- a/drivers/media/platform/coda/coda-jpeg.c
> +++ b/drivers/media/platform/coda/coda-jpeg.c
> @@ -302,13 +302,6 @@ int coda_jpeg_decode_header(struct coda_ctx *ctx, struct vb2_buffer *vb)
>  	}
>  
>  	q_data_src = get_q_data(ctx, V4L2_BUF_TYPE_VIDEO_OUTPUT);
> -	if (header.frame.height != q_data_src->height ||
> -	    header.frame.width != q_data_src->width) {
> -		v4l2_err(&dev->v4l2_dev,
> -			 "dimensions don't match format: %dx%d\n",
> -			 header.frame.width, header.frame.height);
> -		return -EINVAL;
> -	}

Since you are dropping this check, and allowing other sizes to be
decoded using, do you have any check to make sure you don't overrun
(in bytes, not in width x height) the CAPTURE (decoded) buffer?

Also, IIUC your application is negotiating W x H, but then
passing a different size: I wonder if this is not an abuse
of the spec?

Thanks,
Ezequiel

>  
>  	if (header.frame.num_components != 3) {
>  		v4l2_err(&dev->v4l2_dev,