MMX2 code, because the HAVE_ defines do not have sane values here if
runtime cpudetection is used. Hopefully fixes runtime cpudetection
breakage from r19339.


git-svn-id: file:///var/local/repositories/mplayer/trunk/libswscale@19374 b3059339-0415-0410-9bf9-f77b7e298cf2

be protected by HAVE_MMX.


git-svn-id: file:///var/local/repositories/mplayer/trunk/libswscale@19367 b3059339-0415-0410-9bf9-f77b7e298cf2

git-svn-id: file:///var/local/repositories/mplayer/trunk/libswscale@19342 b3059339-0415-0410-9bf9-f77b7e298cf2

git-svn-id: file:///var/local/repositories/mplayer/trunk/libswscale@19339 b3059339-0415-0410-9bf9-f77b7e298cf2

git-svn-id: file:///var/local/repositories/mplayer/trunk/libswscale@19276 b3059339-0415-0410-9bf9-f77b7e298cf2

git-svn-id: file:///var/local/repositories/mplayer/trunk/libswscale@19211 b3059339-0415-0410-9bf9-f77b7e298cf2

update


git-svn-id: file:///var/local/repositories/mplayer/trunk/libswscale@19205 b3059339-0415-0410-9bf9-f77b7e298cf2

git-svn-id: file:///var/local/repositories/mplayer/trunk/libswscale@19186 b3059339-0415-0410-9bf9-f77b7e298cf2

git-svn-id: file:///var/local/repositories/mplayer/trunk/libswscale@19178 b3059339-0415-0410-9bf9-f77b7e298cf2

the +-1 issue is limited to >2tap vertical filters, so bilinear upscale was unaffected
the new code is sometime faster sometimes slower but the difference is significant (~20%) so its optional and enabled with arnd=1


git-svn-id: file:///var/local/repositories/mplayer/trunk/libswscale@19177 b3059339-0415-0410-9bf9-f77b7e298cf2

mv_malloc)


git-svn-id: file:///var/local/repositories/mplayer/trunk/libswscale@19174 b3059339-0415-0410-9bf9-f77b7e298cf2

git-svn-id: file:///var/local/repositories/mplayer/trunk/libswscale@19173 b3059339-0415-0410-9bf9-f77b7e298cf2

git-svn-id: file:///var/local/repositories/mplayer/trunk/libswscale@19148 b3059339-0415-0410-9bf9-f77b7e298cf2

git-svn-id: file:///var/local/repositories/mplayer/trunk/libswscale@18994 b3059339-0415-0410-9bf9-f77b7e298cf2

git-svn-id: file:///var/local/repositories/mplayer/trunk/libswscale@18880 b3059339-0415-0410-9bf9-f77b7e298cf2

git-svn-id: file:///var/local/repositories/mplayer/trunk/libswscale@18866 b3059339-0415-0410-9bf9-f77b7e298cf2

git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@18662 b3059339-0415-0410-9bf9-f77b7e298cf2

fixes crashes on AMD64.


git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@18576 b3059339-0415-0410-9bf9-f77b7e298cf2

git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@18556 b3059339-0415-0410-9bf9-f77b7e298cf2

also on 64 bit systems.


git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@18401 b3059339-0415-0410-9bf9-f77b7e298cf2

this is certainly not the most beautifull but it should work without -fomit-frame-pointer and PIC and gcc-4 vomits when ebp is given in a constraint directly even with -fomi-frame-pointer so theres not that much choice ...
if anyone knows a simpler/cleaner solution, dont hesitate to suggest it ...


git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@18393 b3059339-0415-0410-9bf9-f77b7e298cf2

stride!=width.


git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@18323 b3059339-0415-0410-9bf9-f77b7e298cf2

patch by Erik Auerswald <auerswal at unix-ag uni-kl de>


git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@18111 b3059339-0415-0410-9bf9-f77b7e298cf2

git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@18105 b3059339-0415-0410-9bf9-f77b7e298cf2

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27006

This bug caused certain movies (444P or RGB/BGR-based) to become brighter
and more purple when played on a YV12 vo.


git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@18047 b3059339-0415-0410-9bf9-f77b7e298cf2

consistent naming of functions


git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@17845 b3059339-0415-0410-9bf9-f77b7e298cf2

was fixed in rgb32to16 about a year ago: using only the first 8 bits of the
32-bit pixel.


git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@17671 b3059339-0415-0410-9bf9-f77b7e298cf2

how altivec_yuv2packedX is supposed to be used


git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@17643 b3059339-0415-0410-9bf9-f77b7e298cf2

instead fall back on yuv2packedXinC


git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@17642 b3059339-0415-0410-9bf9-f77b7e298cf2

to spec; also convert a few vec_splat's to vec_splat_{s,u}*


git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@17639 b3059339-0415-0410-9bf9-f77b7e298cf2

git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@17633 b3059339-0415-0410-9bf9-f77b7e298cf2

git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@17632 b3059339-0415-0410-9bf9-f77b7e298cf2

git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@17623 b3059339-0415-0410-9bf9-f77b7e298cf2

once when the scaler is initialized, instead of building them and freeing
them over and over. This gives massive performance improvements.
patch by Alan Curry, pacman*at*TheWorld*dot*com


git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@17589 b3059339-0415-0410-9bf9-f77b7e298cf2

git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@17588 b3059339-0415-0410-9bf9-f77b7e298cf2

patch by Alan Curry, pacman_at_TheWorld_dot_com


git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@17587 b3059339-0415-0410-9bf9-f77b7e298cf2

patch replaces '()' for the correct '(void)' in function
declarations/prototypes which have no parameters. The '()' syntax tell
thats there is a variable list of arguments, so that the compiler cannot
check this. The extra CFLAG '-Wstrict-declarations' shows those cases.

Comments about a similar patch applied to ffmpeg:

That in C++ these mean the same, but in ANSI C the semantics are
different; function() is an (obsolete) K&R C style forward declaration,
it basically means that the function can have any number and any types
of parameters, effectively completely preventing the compiler from doing
any sort of type checking. -- Erik Slagter

Defining functions with unspecified arguments is allowed but bad.
With arguments unspecified the compiler can't report an error/warning
if the function is called with incorrect arguments. -- Måns Rullgård


git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@17567 b3059339-0415-0410-9bf9-f77b7e298cf2

do the opposite of what they say.
The vec_mstrgb24 and vec_mstbgr24 macros should be opposites and not perform
the same things.
patch by Alan Curry, pacman_at_world_dot_std_dot_com


git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@17564 b3059339-0415-0410-9bf9-f77b7e298cf2

outputs RGBA. This patch supports 6 output formats and prints an error message
if it is asked to provide an output format it is not capable of.
patch by Alan Curry, pacman_at_world_dot_std_dot_com


git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@17561 b3059339-0415-0410-9bf9-f77b7e298cf2

patch by Alan Curry, pacman at world dot std dot com


git-svn-id: file:///var/local/repositories/mplayer/trunk/postproc@17559 b3059339-0415-0410-9bf9-f77b7e298cf2