On 2014-02-24 11:32:43 +0100, Diego Biurrun wrote:
> On Sat, Feb 22, 2014 at 07:59:17PM +0100, Janne Grunau wrote:
> > --- a/libavcodec/x86/dcadsp.asm
> > +++ b/libavcodec/x86/dcadsp.asm
> > @@ -88,3 +88,90 @@ INT8X8_FMUL_INT32
> >  
> >  INIT_XMM sse4
> >  INT8X8_FMUL_INT32
> > +
> > +; %1=v0/v1  %2=in1  %3=in2
> > +%macro FIR_LOOP 2-3
> > +.loop%1:
> > +%define va          m1
> > +%define vb          m2
> > +; for v0, incrementint and for v1, decrementing
> 
> incrementin_g_

fixed locally

> > +    add      count, 8
> > +    jl   .loop%1
> 
> Is there a reason for adding %1 to the jump label?

yes, the macro is expanded twice with different %1 parameter and the %1
is appended so that it would jump to the correct label.

> > +; dca_lfe_fir(float *out, float *in, float *coefs)
> 
> Please add the return type to the comment for completeness.

fixed locally

Janne
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to