The mmxext optimizations should be at least equally fast if available and
amd3dnow optimizations are being deprecated. Thus the former should
override the latter, not the other way around.

Simplifies host/target libc detection splitting.

Fixes an assertion failure in vdpau_get_buffer() when -hwaccel auto is
used, but hwaccel init fails.

Otherwise the expression will be evaluated as unsigned, which will break
when the result should be negative.
CC:libav-stable@libav.org

The table is MMX-specific and used nowhere else.

Based on a patch by Michael Niedermayer.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>

Certain instrumentation addons leads to a false positive in configure
and link failures at the end of the build phase.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>

Original decoder seems to always use 176x128 tiles anyway and this helps
avoiding lots of issues with odd tile sizes in fuzzed files.
Signed-off-by: Anton Khirnov <anton@khirnov.net>

The field still remains in MpegEncContext because it is used by the
mpeg4 decoder.

It is only used in one place immediately after it is set.

That field is never set for encoding.

The field remains in MpegEncContext for use in the encoder.