Andreas: On Tue, Mar 12, 2024 at 9:14 PM Andreas Rheinhardt <andreas.rheinha...@outlook.com> wrote: > > h_subpel_filters_inner[i] and h_subpel_filters_outer[i / 2] > belong together and the former allows the range 0..6, > so the latter needs to support 0..3. But it has only three > elements. Add another one. > The value for the last element has been guesstimated > from subpel_filters in libavcodec/vp8dsp.c. > > This is also intended to fix FATE-failures with UBSan here: > https://fate.ffmpeg.org/report.cgi?time=20240312011016&slot=ppc-linux-gcc-13.2-ubsan-altivec-qemu > > Signed-off-by: Andreas Rheinhardt <andreas.rheinha...@outlook.com> > --- > libavcodec/ppc/vp8dsp_altivec.c | 3 ++- > 1 file changed, 2 insertions(+), 1 deletion(-) > > diff --git a/libavcodec/ppc/vp8dsp_altivec.c b/libavcodec/ppc/vp8dsp_altivec.c > index 12dac8b0a8..061914fc38 100644 > --- a/libavcodec/ppc/vp8dsp_altivec.c > +++ b/libavcodec/ppc/vp8dsp_altivec.c > @@ -50,11 +50,12 @@ static const vec_s8 h_subpel_filters_inner[7] = > // for 6tap filters, these are the outer two taps > // The zeros mask off pixels 4-7 when filtering 0-3 > // and vice-versa > -static const vec_s8 h_subpel_filters_outer[3] = > +static const vec_s8 h_subpel_filters_outer[4] = > { > REPT4(0, 0, 2, 1), > REPT4(0, 0, 3, 3), > REPT4(0, 0, 1, 2), > + REPT4(0, 0, 0, 0), > }; > > #define LOAD_H_SUBPEL_FILTER(i) \ > -- > 2.40.1 > > _______________________________________________ > 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".
Confirming this patch fixes fate-checkasm-vp8dsp (and presumably the other vp8 tests in GCC UBsan) on PowerPC QEMU, POWER7 (ppc64), and POWER9 (ppc64le). Thanks again, 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".