On 12/9/2015 6:54 PM, Janne Grunau wrote: > --- > libavcodec/x86/fmtconvert.asm | 3 +++ > 1 file changed, 3 insertions(+) > > diff --git a/libavcodec/x86/fmtconvert.asm b/libavcodec/x86/fmtconvert.asm > index 727daa9..b9a78a5 100644 > --- a/libavcodec/x86/fmtconvert.asm > +++ b/libavcodec/x86/fmtconvert.asm > @@ -38,6 +38,9 @@ cglobal int32_to_float_fmul_scalar, 4, 4, %1, dst, src, > mul, len > %elif ARCH_X86_32 > movss m0, mulm > %endif > +%if ARCH_X86_64 > + movsxd lenq, lend
You should use the movsxdifnidn macro instead. > +%endif > SPLATD m0 > shl lenq, 2 In any case, changing this to "shl lend, 2" is a better solution than adding the above instruction. > add srcq, lenq > _______________________________________________ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
