[FFmpeg-devel] [PATCH v2] avfilter/vf_lut3d: add x86-optimized tetrahedral interpolation

[Xiang, Haihao]

On Sat, 2021-10-09 at 15:24 -0700, Mark Reid wrote:
> On Sat, Oct 9, 2021 at 4:11 AM Paul B Mahol <onemda at gmail.com> wrote:
> 
> > will test and apply shortly, why 8bit is not covered?
> > 
> 
> Thanks for taking the time to test. I didn't do 8bit yet because I was
> trying to limit my testing matrix, and these happen to be my main use
> cases. I hope to try and incrementally add other pixel formats and
> interpolation methods.


This patch broke the linker with the following configuration:

  configuration: --prefix=/usr --libdir=/usr/lib/x86_64-linux-gnu --
shlibdir=/usr/lib/x86_64-linux-gnu --enable-opencl --enable-libglslang --enable-
vulkan --enable-libdrm --enable-shared --enable-pic --enable-gpl --disable-
stripping --disable-optimizations --disable-static --disable-mmx --disable-ssse3 
--enable-debug=3 --enable-libmfx --samples=fate-suite/ --enable-opengl

LD      ffprobe_g
libavfilter/libavfilter.so: undefined reference to
`ff_interp_tetrahedral_pf32_sse2'
libavfilter/libavfilter.so: undefined reference to
`ff_interp_tetrahedral_p16_sse2'
collect2: error: ld returned 1 exit status
Makefile:125: recipe for target 'ffprobe_g' failed
make: *** [ffprobe_g] Error 1

Thanks
Haihao