[FFmpeg-devel] [PATCH] lavc/vvc: Fix derivation of inverse LMCS idx

[Frank Plowman]

Ping

On 02/02/2025 12:10, Frank Plowman wrote:
> The clamping of idxYInv from H.266(V3) section 8.8.2.3 was missing.
> This could lead to OOB reads from lmcs->pivot or input_pivot.
> 
> I also changed the derivation of the forward LMCS idx to use a shift
> rather than a division for speed and as this is actually how the
> variable is declared in the specification (8.7.5.2).
> 
> Signed-off-by: Frank Plowman <post at frankplowman.com>
> ---
>  libavcodec/vvc/ps.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/libavcodec/vvc/ps.c b/libavcodec/vvc/ps.c
> index 01b4615eda..fae6655cc0 100644
> --- a/libavcodec/vvc/ps.c
> +++ b/libavcodec/vvc/ps.c
> @@ -786,7 +786,7 @@ static int lmcs_derive_lut(VVCLMCS *lmcs, const H266RawAPS *rlmcs, const H266Raw
>  
>      //derive lmcs_fwd_lut
>      for (uint16_t sample = 0; sample < max; sample++) {
> -        const int idx_y = sample / org_cw;
> +        const int idx_y = sample >> shift;
>          const uint16_t fwd_sample = lmcs_derive_lut_sample(sample, lmcs->pivot,
>              input_pivot, scale_coeff, idx_y, max);
>          if (bit_depth > 8)
> @@ -802,6 +802,7 @@ static int lmcs_derive_lut(VVCLMCS *lmcs, const H266RawAPS *rlmcs, const H266Raw
>          uint16_t inv_sample;
>          while (i <= lmcs->max_bin_idx && sample >= lmcs->pivot[i + 1])
>              i++;
> +        i = FFMIN(i, LMCS_MAX_BIN_SIZE - 1);
>  
>          inv_sample = lmcs_derive_lut_sample(sample, input_pivot, lmcs->pivot,
>              inv_scale_coeff, i, max);