[FFmpeg-devel] [PATCH] Support limiting the number of pixels per image

[Marton Balint]

On Fri, 9 Dec 2016, Michael Niedermayer wrote:

> Adds av_image_check_size2() with max_pixels and pix_fmt parameters.
> pix_fmt is unused and is added to avoid a 2nd API change later
>
> The old function uses AVOptions to obtain the max_pixels value to simplify
> the transition.
>
> TODO: split into 2 patches (one per lib), docs & bump
>
> This allows preventing some OOM and "slow decoding" cases by limiting the maximum resolution
> this may be useful to avoid fuzzers getting stuck in boring cases
>
> Signed-off-by: Michael Niedermayer <michael at niedermayer.cc>
> ---
> libavcodec/avcodec.h                 |  8 ++++++++
> libavcodec/options_table.h           |  1 +
> libavutil/imgutils.c                 | 31 ++++++++++++++++++++++++++-----
> libavutil/imgutils.h                 | 14 ++++++++++++++
> tests/ref/fate/api-mjpeg-codec-param |  2 ++
> tests/ref/fate/api-png-codec-param   |  2 ++
> 6 files changed, 53 insertions(+), 5 deletions(-)
>
> diff --git a/libavcodec/avcodec.h b/libavcodec/avcodec.h
> index 7ac2adaf66..81052b10ef 100644
> --- a/libavcodec/avcodec.h
> +++ b/libavcodec/avcodec.h
> @@ -3570,6 +3570,14 @@ typedef struct AVCodecContext {
>      */
>     int trailing_padding;
> 
> +    /**
> +     * The number of pixels per image to maximally accept.
> +     *
> +     * - decoding: set by user
> +     * - encoding: unused
> +     */
> +    int max_pixels;

Don't we want to make this an int64, for future extendability?

> +
> } AVCodecContext;
> 
> AVRational av_codec_get_pkt_timebase         (const AVCodecContext *avctx);
> diff --git a/libavcodec/options_table.h b/libavcodec/options_table.h
> index ee79859953..2f5eb252fe 100644
> --- a/libavcodec/options_table.h
> +++ b/libavcodec/options_table.h
> @@ -570,6 +570,7 @@ static const AVOption avcodec_options[] = {
> {"codec_whitelist", "List of decoders that are allowed to be used", OFFSET(codec_whitelist), AV_OPT_TYPE_STRING, { .str = NULL },  CHAR_MIN, CHAR_MAX, A|V|S|D },
> {"pixel_format", "set pixel format", OFFSET(pix_fmt), AV_OPT_TYPE_PIXEL_FMT, {.i64=AV_PIX_FMT_NONE}, -1, INT_MAX, 0 },
> {"video_size", "set video size", OFFSET(width), AV_OPT_TYPE_IMAGE_SIZE, {.str=NULL}, 0, INT_MAX, 0 },
> +{"max_pixels", "Maximum number of pixels", OFFSET(max_pixels), AV_OPT_TYPE_INT, {.i64 = INT_MAX }, 0, INT_MAX, A|V|S|D },

INT_MAX is probably OK, as who knows what will crash for bigger sizes, 
right?

Thanks,
Marton