As a positive side-effect, this gets rid of the following warning:
libavcodec/flashsv.c:190:38: warning: assignment discards ‘const’ qualifier from pointer target type

An official specification now exists and it is both more detailed and easier to
understand than the incomplete notes available here.  Also remove ideas on how
to implement an encoder; these notes are already present in the encoder.

Fixes ticket #306.

Signed-off-by: Diego Biurrun <diego@biurrun.de>

* qatar/master:
  Add some missing mathematics.h #includes for av_rescale().
  opencore-amr: Add missing initializer braces to shut up gcc warning.
  ARM: workaround for bug in GNU assembler
  dv: fix comment wording mistake
  Rename libavcodec/high_bit_depth.h ---> libavcodec/bit_depth_template.c
  dv: fix valgrind use of uninitialised value warnings.
  mxfenc: fix ignored drop flag in binary timecode representation.
  PPC: use Altivec IMDCT only for supported sizes
  dv: fix comment spelling
  configure: simplify -rpath-link linker flag

Merged-by: Michael Niedermayer <michaelni@gmx.at>

This fixes the warning:
libavcodec/libopencore-amr.c:91: warning: missing braces around initializer

Some versions of the GNU assembler do not handle 64-bit
immediate operands containing arithmetic.  Writing the
value out in full works correctly.

Signed-off-by: Mans Rullgard <mans@mansr.com>

This naming scheme is used elsewhere, so it's sensible to be consistent.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>

The Altivec IMDCT works with size 32 and higher only.

Signed-off-by: Mans Rullgard <mans@mansr.com>

Signed-off-by: Diego Biurrun <diego@biurrun.de>

Lists of ':'-separated directories can be passed to -rpath-link directly.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

* qatar/master: (36 commits)
  ARM: allow unaligned buffer in fixed-point NEON FFT4
  fate: test more FFT etc sizes
  dca: set AVCodecContext frame_size for DTS audio
  YASM: Shut up unused variable compiler warning with --disable-yasm.
  x86_32: Fix build on x86_32 with --disable-yasm.
  iirfilter: add fate test
  doxygen: Add qmul docs.
  ogg: propagate return values and return more meaningful error values
  H.264: fix overreads of qscale_table
  Remove unused static tables and static inline functions.
  eval: clear Parser instances before using
  dct-test: remove 'ref' function pointer from tables
  build: Remove deleted 'check' target from .PHONY list.
  oggdec: Abort Ogg header parsing when encountering a data packet.
  Add LGPL license boilerplate to files lacking it.
  mxfenc: small typo fix
  doxygen: Fix documentation for some VP8 functions.
  sha: use AV_RB32() instead of assuming buffer can be cast to uint32_t*
  des: allow unaligned input and output buffers
  aes: allow unaligned input and output buffers
  ...

Conflicts:
	libavcodec/dct-test.c
	libavcodec/libvpxenc.c
	libavcodec/x86/dsputil_mmx.c
	libavcodec/x86/h264_qpel_mmx.c
	libavfilter/x86/gradfun.c
	libavformat/oggdec.c

Merged-by: Michael Niedermayer <michaelni@gmx.at>

This function is called with only 8-byte alignment from
imdct for size 16.  The fft4 function is not called for
the larger FFT or MDCT sizes, so this has no impact on
typical uses.

Signed-off-by: Mans Rullgard <mans@mansr.com>

This tests sizes from 16 to 4096, not only the default 512.

Signed-off-by: Mans Rullgard <mans@mansr.com>

Set the frame size when decoding DTS audio.

This has the side effect of fixing the computation of timestamps for DTS-HD in compute_pkt_fields.  Since frame_size is
not currently set, the duration of a frame is being guessed based on the streams bitrate.  But for DTS-HD, the bitrate
currently used is the rate of the DTS core which is much different than the whole DTS-HD stream and leads to a wildly
inaccurate frame duration estimate.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>

Signed-off-by: Diego Biurrun <diego@biurrun.de>

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>

Change the test program to output text instead of binary files so
it can be compared easily in FATE.

Signed-off-by: Mans Rullgard <mans@mansr.com>

Signed-off-by: Diego Biurrun <diego@biurrun.de>

filter_mb_fast assumed that qscale_table was padded like many of the other tables.

This prevents random values from the stack being used as
"variables" in expressions.

Signed-off-by: Mans Rullgard <mans@mansr.com>

The reference function can be inferred from the direction of the
transform, no need for storing it with each entry.

Signed-off-by: Mans Rullgard <mans@mansr.com>