Hi,
On Mon, Feb 26, 2024 at 6:48 PM Andreas Rheinhardt
<andreas.rheinha...@outlook.com> wrote:
>
> Fixes the dxv3enc-dxt1 FATE test with UBSan.
>
> Signed-off-by: Andreas Rheinhardt <andreas.rheinha...@outlook.com>
> ---
> I have not actually tested whether the output is actually wrong
> on BE systems. Would be nice if someone could.
>
> libavcodec/dxvenc.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/libavcodec/dxvenc.c b/libavcodec/dxvenc.c
> index 1ce2b1d014..3f28fb696a 100644
> --- a/libavcodec/dxvenc.c
> +++ b/libavcodec/dxvenc.c
> @@ -149,7 +149,7 @@ typedef struct DXVEncContext {
> } else {
> \
> op = 0;
> \
> }
> \
> - *value |= (op << (state * 2));
> \
> + AV_WL32(value, AV_RL32(value) | (op << (state * 2)));
> \
> state++;
> \
> } while (0)
>
> @@ -157,7 +157,7 @@ static int dxv_compress_dxt1(AVCodecContext *avctx)
> {
> DXVEncContext *ctx = avctx->priv_data;
> PutByteContext *pbc = &ctx->pbc;
> - uint32_t *value;
> + void *value;
> uint32_t color, lut, idx, color_idx, lut_idx, prev_pos, state = 16, pos
> = 2, op = 0;
>
> ht_init(ctx->color_lookback_ht);
Confirming that this does fix the failing test on a big-endian PowerPC
+ Altivec virtual machine.
Thanks,
Sean McGovern
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
To unsubscribe, visit link above, or email
ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
