[libav-devel] [PATCH 022/132] dsputil.h: Clean up comments

Diego Biurrun diego at biurrun.de
Tue Mar 11 18:15:53 CET 2014


On Tue, Mar 11, 2014 at 05:44:15PM +0100, Luca Barbato wrote:
> On 11/03/14 17:27, Diego Biurrun wrote:
> > --- a/libavcodec/dsputil.h
> > +++ b/libavcodec/dsputil.h
> > @@ -23,8 +23,8 @@
> >  /**
> >   * @file
> >   * DSP utils.
> > - * note, many functions in here may use MMX which trashes the FPU state, it is
> > - * absolutely necessary to call emms_c() between dsp & float/double code
> > + * Note, many functions in here may use MMX which trashes the FPU state, it is
> @note ?
> > + * absolutely necessary to call emms_c() between DSP & float/double code.
> >   */
> >  
> > @@ -61,21 +61,22 @@ void ff_gmc_c(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy,
> >  
> >  /* minimum alignment rules ;)
> > -If you notice errors in the align stuff, need more alignment for some ASM code
> > -for some CPU or need to use a function with less aligned data then send a mail
> > -to the libav-devel mailing list, ...
> > -
> > -!warning These alignments might not match reality, (missing attribute((align))
> > -stuff somewhere possible).
> > -I (Michael) did not check them, these are just the alignments which I think
> > -could be reached easily ...
> > -
> > -!future video codecs might need functions with less strict alignment
> > -*/
> > + * If you notice errors in the align stuff, need more alignment for some ASM code
> > + * for some CPU or need to use a function with less aligned data then send a mail
> > + * to the libav-devel mailing list, ...
> > + *
> > + * !warning These alignments might not match reality, (missing attribute((align))
> @warning ?
> > + * stuff somewhere possible).
> > + * I (Michael) did not check them, these are just the alignments which I think
> > + * could be reached easily ...
> 
> Is it still valid?
> 
> Same questions for the rest where there is a tag available in doxy.

To be honest, all of these get thrown away in later patches, so I'd
just leave as-is to avoid rebase churn.

Diego


More information about the libav-devel mailing list