[FFmpeg-devel] [PATCH 4/5] avcodec/sga: Check non constant init_get_bits8()

Andreas Rheinhardt andreas.rheinhardt at outlook.com
Fri May 17 10:53:21 EEST 2024


Michael Niedermayer:
> Fixes: CID1473562 Unchecked return value
> Fixes: CID1473592 Unchecked return value
> 
> Sponsored-by: Sovereign Tech Fund
> Signed-off-by: Michael Niedermayer <michael at niedermayer.cc>
> ---
>  libavcodec/sga.c | 10 ++++++++--
>  1 file changed, 8 insertions(+), 2 deletions(-)
> 
> diff --git a/libavcodec/sga.c b/libavcodec/sga.c
> index 0f42cf912b2..aca941e057e 100644
> --- a/libavcodec/sga.c
> +++ b/libavcodec/sga.c
> @@ -254,11 +254,14 @@ static int decode_palmapdata(AVCodecContext *avctx)
>      const int bits = (s->nb_pal + 1) / 2;
>      GetByteContext *gb = &s->gb;
>      GetBitContext pm;
> +    int ret;
>  
>      bytestream2_seek(gb, s->palmapdata_offset, SEEK_SET);
>      if (bytestream2_get_bytes_left(gb) < s->palmapdata_size)
>          return AVERROR_INVALIDDATA;
> -    init_get_bits8(&pm, gb->buffer, s->palmapdata_size);
> +    ret = init_get_bits8(&pm, gb->buffer, s->palmapdata_size);
> +    if (ret < 0)
> +        return ret;
>  
>      for (int y = 0; y < s->tiles_h; y++) {
>          uint8_t *dst = s->palmapindex_data + y * s->tiles_w;
> @@ -277,11 +280,14 @@ static int decode_tiledata(AVCodecContext *avctx)
>      SGAVideoContext *s = avctx->priv_data;
>      GetByteContext *gb = &s->gb;
>      GetBitContext tm;
> +    int ret;
>  
>      bytestream2_seek(gb, s->tiledata_offset, SEEK_SET);
>      if (bytestream2_get_bytes_left(gb) < s->tiledata_size)
>          return AVERROR_INVALIDDATA;
> -    init_get_bits8(&tm, gb->buffer, s->tiledata_size);
> +    ret = init_get_bits8(&tm, gb->buffer, s->tiledata_size);
> +    if (ret < 0)
> +        return ret;
>  
>      for (int n = 0; n < s->nb_tiles; n++) {
>          uint8_t *dst = s->tileindex_data + n * 64;

Both of these can not fail and could be checked via av_assert1:
palmapdata_size is given by (s->tiles_w * s->tiles_h * ((s->nb_pal + 1)
/ 2) + 7) / 8 with tiles_w and tiles_h being in the 0..255 range and
nb_pal being in the 0..4 range.
tiledata_size is given by s->nb_tiles * 32; nb_tiles fits in 16 bits (it
is either read via AV_RB16 or is given as the product of tiles_h *
tiles_w, both of which are read from simple uint8_t.

- Andreas



More information about the ffmpeg-devel mailing list