[FFmpeg-devel] [PATCH] x86: dsputilenc: convert ff_sse{8, 16}_mmx() to yasm

Christophe Gisquet christophe.gisquet at gmail.com
Sun May 25 19:34:59 CEST 2014


2014-05-24 2:18 GMT+02:00 Timothy Gu <timothygu99 at gmail.com>:
> +int ff_sse8_mmx(MpegEncContext *v, uint8_t *pix1, uint8_t *pix2,
> +                int line_size, int h);
> +int ff_sse16_mmx(MpegEncContext *v, uint8_t *pix1, uint8_t *pix2,
> +                 int line_size, int h);

this does not compile with --disable-yasm:
libavcodec/libavcodec.a(dsputilenc_mmx.o): In function `nsse8_mmx':
libavcodec/x86/dsputilenc_mmx.c:318: undefined reference to `ff_sse8_mmx'
libavcodec/libavcodec.a(dsputilenc_mmx.o): In function `nsse16_mmx':
libavcodec/x86/dsputilenc_mmx.c:305: undefined reference to `ff_sse16_mmx'

This also results in kind of messy code because some functions now
rely on both inline asm and yasm. To fix this, ideally hf_noise* would
be ported to yasm too.

Meanwhile, it may be good to revert this change.


More information about the ffmpeg-devel mailing list