[FFmpeg-devel] [PATCH] Do not assume long is same width as x86 register in swscale.

Reimar Döffinger Reimar.Doeffinger
Wed Mar 18 09:12:57 CET 2009


On Wed, Mar 18, 2009 at 06:27:08AM +0100, Michael Niedermayer wrote:
> On Tue, Mar 17, 2009 at 09:30:06PM -0300, Ramiro Polla wrote:
> > Hi,
> > 
> > On Tue, Mar 17, 2009 at 8:51 PM, Diego Biurrun <diego at biurrun.de> wrote:
> > > On Tue, Mar 17, 2009 at 07:23:18PM -0300, Ramiro Polla wrote:
> > >>
> > >> swscale_funnycode2.diff checks for VirtualAlloc in windows.h (os/2
> > >> could also use something similar then).
> > >>
> > >> --- a/configure
> > >> +++ b/configure
> > >
> > > OK
> > >
> > >> --- a/swscale.c
> > >> +++ b/swscale.c
> > >> @@ -68,6 +68,10 @@ untested special converters
> > >> ?#define MAP_ANONYMOUS MAP_ANON
> > >> ?#endif
> > >> ?#endif
> > >> +#ifdef HAVE_VIRTUALALLOC
> > >
> > > #if
> > 
> > Ah, I had forgotten about the switch to #if
> > 
> > >> +#define WIN32_LEAN_AND_MEAN
> > >
> > > Is this necessary?
> > 
> > It's a good idea, like Reimar pointed out.
> > 
> > >> @@ -2513,6 +2517,9 @@ SwsContext *sws_getContext(int srcW, int srcH, enum PixelFormat srcFormat, int d
> > >> +#elif defined(HAVE_VIRTUALALLOC)
> > >
> > > Leave out the 'defined'.
> > >
> > >> @@ -3161,6 +3168,9 @@ void sws_freeContext(SwsContext *c){
> > >> +#elif defined(HAVE_VIRTUALALLOC)
> > >
> > > ditto
> > 
> > Updated patch attached.
> [...]
> 
> > @@ -2513,6 +2517,9 @@
> >  #ifdef MAP_ANONYMOUS
> >              c->funnyYCode = (uint8_t*)mmap(NULL, MAX_FUNNY_CODE_SIZE, PROT_EXEC | PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS, 0, 0);
> >              c->funnyUVCode = (uint8_t*)mmap(NULL, MAX_FUNNY_CODE_SIZE, PROT_EXEC | PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS, 0, 0);
> > +#elif HAVE_VIRTUALALLOC
> > +            c->funnyYCode = (uint8_t*)VirtualAlloc(NULL, MAX_FUNNY_CODE_SIZE, MEM_COMMIT, PAGE_EXECUTE_READWRITE);
> > +            c->funnyUVCode = (uint8_t*)VirtualAlloc(NULL, MAX_FUNNY_CODE_SIZE, MEM_COMMIT, PAGE_EXECUTE_READWRITE);
> 
> vertical align

That applies to the existing code, as well as the useless cast. IMO just
fix them all together in one cosmetic patch.




More information about the ffmpeg-devel mailing list