[FFmpeg-devel] macos build error relating to yasm

[Adam Bennett]

I'm getting a yasm related build error that I didn't hit when building
ffmpeg on Windows.

I ran the following commands to build libav on a new Big Sur install after
installing xcode command line tools:

```
brew install yasm
brew install pkg-config
./configure --disable-everything --enable-libvpx
--enable-encoder=libvpx_vp8 --enable-encoder=libvpx_vp9 --enable-muxer=webm
make
```

This installed yasm 1.3.0 and pkg-config 0.29.2.

And I get this error from make:

```
libavutil/x86/tx_float.asm:842: error: undefined symbol `ff_cos_32_float'
(first use)
libavutil/x86/tx_float.asm:842: error:  (Each undefined symbol is reported
only once.)
libavutil/x86/tx_float.asm:1210: error: undefined symbol `ff_cos_64_float'
(first use)
libavutil/x86/tx_float.asm:1210: error: undefined symbol `ff_cos_128_float'
(first use)
libavutil/x86/tx_float.asm:1210: error: undefined symbol `ff_cos_256_float'
(first use)
libavutil/x86/tx_float.asm:1210: error: undefined symbol `ff_cos_512_float'
(first use)
libavutil/x86/tx_float.asm:1210: error: undefined symbol
`ff_cos_1024_float' (first use)
libavutil/x86/tx_float.asm:1210: error: undefined symbol
`ff_cos_2048_float' (first use)
libavutil/x86/tx_float.asm:1210: error: undefined symbol
`ff_cos_4096_float' (first use)
libavutil/x86/tx_float.asm:1210: error: undefined symbol
`ff_cos_8192_float' (first use)
libavutil/x86/tx_float.asm:1210: error: undefined symbol
`ff_cos_16384_float' (first use)
libavutil/x86/tx_float.asm:1210: error: undefined symbol
`ff_cos_32768_float' (first use)
libavutil/x86/tx_float.asm:1210: error: undefined symbol
`ff_cos_65536_float' (first use)
libavutil/x86/tx_float.asm:1210: error: undefined symbol
`ff_cos_131072_float' (first use)
make: *** [libavutil/x86/tx_float.o] Error 1
```

This line (42) in tx_float.asm seems to provide the undefined symbol but
truth be told I've never worked with a project using asm and not sure from
what I've seen what the problem is.

cextern cos_ %+ i %+ _float ; ff_cos_i_float...