Merge commit '6bf4c1d71199b92894f24db6386ed5070e590a16'

* commit '6bf4c1d71199b92894f24db6386ed5070e590a16':
  r3d: do not create the audio stream until we know the sample rate

Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>

Merge commit '9f1eccb97bf8894cb18b14f642500686505ef186'

* commit '9f1eccb97bf8894cb18b14f642500686505ef186':
  ff_parse_specific_params: do not use AVCodecContext.frame_size

Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>

Merge commit '1f008f34d5b2b5f6217521747e7acfe3efc0e666'

* commit '1f008f34d5b2b5f6217521747e7acfe3efc0e666':
  rmenc: do not use AVCodecContext.frame_size

Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>

doc/filters: add showwavespic colorize example

Signed-off-by: Lou Logan <lou@lrcd.com>

avcodec/ffv1enc: Support AV_PIX_FMT_YA8

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/ffv1dec: Support AV_PIX_FMT_YA8

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

mjpegdec: extend check for incompatible values of s->rgb and s->ls

This can happen if s->ls changes from 0 to 1, but picture allocation is
skipped due to s->interlaced.

In that case ff_jpegls_decode_picture could be called even though the
s->picture_ptr frame has the wrong pixel format and thus a wrong
linesize, which results in a too small zero buffer being allocated.

This fixes an out-of-bounds read in ls_decode_line.

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>

avformat/mov: support cenc (common encryption)

support reading encrypted mp4 using aes-ctr, conforming to ISO/IEC
23001-7.

a new parameter was added:
- decryption_key - 128 bit decryption key (hex)

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avfilter/avf_showspectrum: add window overlap support

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/avf_showspectrum: add fire color map

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/trim: support all channel numbers in atrim filter

Signed-off-by: Paul B Mahol <onemda@gmail.com>

lavc/dsd_tablegen: always generate tables at runtime

Commit b272c3a5aa has sped up dsd_tablegen, and now table generation takes
~ 40k cycles. Thus, these tables can always be generated at runtime.

Tested with/without --enable-hardcoded-tables.

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

avfilter/vf_scale: set proper out frame color range

Prevents that following scalers in the filter chain will do unintentional color range conversions.
Fixes Ticket #5096

Signed-off-by: Thomas Mundt <loudmax@yahoo.de>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

dirac_dwt: remove unnecessary undefs

They're all undefined within the template file.

Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>

configure: support --cpu=host and Clang

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avfilter/avf_showspectrum: add nebulae color map

Signed-off-by: Paul B Mahol <onemda@gmail.com>

mips: put "disable mipsfpu" in a better place for loongson

Let's disable the ISAs first, and then the core capabilities, as we do
for the rest of the cores. This way the code is better organized.

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

mips: add mips32r6 architecture variant

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avformat/movenc: Use strict_std_compliance from the muxer layer instead of the encoder layer

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

avfilter/avf_showspectrum: add moreland color map

Signed-off-by: Paul B Mahol <onemda@gmail.com>

lavc/cook: get rid of wasteful pow in init_pow2table

The table is highly structured, so pow (or exp2 for that matter) can entirely
be avoided, yielding a ~ 40x speedup with no loss of accuracy.

sample benchmark (Haswell, GNU/Linux):
new:
4449 decicycles in init_pow2table(loop 1000),     254 runs,      2 skips
4411 decicycles in init_pow2table(loop 1000),     510 runs,      2 skips
4391 decicycles in init_pow2table(loop 1000),    1022 runs,      2 skips

old:
183673 decicycles in init_pow2table(loop 1000),     256 runs,      0 skips
182142 decicycles in init_pow2table(loop 1000),     512 runs,      0 skips
182104 decicycles in init_pow2table(loop 1000),    1024 runs,      0 skips

Reviewed-by: Clément Bœsch <u@pkh.me>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavc/dsd_tablegen: speed up table generation

Tables are bit identical.
Sample benchmark (Haswell, GNU/Linux+gcc):
old:
 814485 decicycles in dsd_ctables_tableinit,     512 runs,      0 skips

new:
 356808 decicycles in dsd_ctable_tableinit,     512 runs,      0 skips

Binary size should essentially be identical, and is in fact identical on
the configuration I tested on.

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

configure: make gcc 2 detection more robust

Solves an issue that will get triggered when gcc 20 rolls in.

Found-by: Hendrik Leppkes <h.leppkes@gmail.com>
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

avfilter/avf_showspectrum: add rainbow colormap

Signed-off-by: Paul B Mahol <onemda@gmail.com>

oggparsedaala: check number of planes in pixel format map

This fixes crashes caused by out-of-bounds writes.

Reviewed-by: Rostislav Pehlivanov <atomnuker@gmail.com>
Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>

ffprobe: avoid unnecessary pow and exp2 calls

These are just for prefixes and may be hardcoded easily; see lavu/eval
for this approach.

Reviewed-by: Stefano Sabatini <stefasab@gmail.com>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

ffserver: Cast time_t value when using it in a format string.

Fixes ticket #5103.

lavf/qtpalette: Treat 1-bit video as palettized

This commit for qtpalette.c and qtpalette.h adds 1-bit video to the
"palettized video" category, since if the video sample description
contains a palette, the two colors in the palette can be any color, not
necessarily black & white.

Unfortunately, I've noticed that the qtrle (QuickTime Animation) decoder
blindly assumes that 1-bit video is black & white. I don't have enough
knowledge about the decoder to fix this, though.

Below is a link to a sample 1-bit QuickTime Animation clip of a rotating
earth that uses blueish colors, and they will be correctly rendered in
QuickTime, but not in FFmpeg (which will use black & white).

https://drive.google.com/open?id=0B3_pEBoLs0faUlItWm9KaGJSTEE

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec: Use get_ue_golomb_long() when needed

get_ue_golomb() cannot decode values larger than 8190 (the maximum
value that can be golomb encoded in 25 bits) and produces the error
"Invalid UE golomb code" if a larger value is encountered.  Use
get_ue_golomb_long() instead (which supports 63 bits, up to 4294967294)
when valid h264/hevc values can exceed 8190.

This updates decoding of the following values:   (maximum)
  first_mb_in_slice                                36863* for level 5.2
  abs_diff_pic_num_minus1                         131071
  difference_of_pic_nums_minus1                   131071
  idr_pic_id                                       65535
  recovery_frame_cnt                               65535
  frame_packing_arrangement_id                4294967294
  frame_packing_arrangement_repetition_period      16384
  display_orientation_repetition_period            16384

An alternative would be to modify get_ue_golomb() to handle encoded
values of up to 49 bits as was done for get_se_golomb() in a92816c.
In that case get_ue_golomb() could continue to be used for all of
these except frame_packing_arrangement_id.

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avfilter/avf_showspectrum: properly calculate w factor

It is used in calculating loudness of each frequency bin.

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/avf_showspectrum: add horizontal orientation support

Signed-off-by: Paul B Mahol <onemda@gmail.com>

lavfi/af_anequalizer: replace pow(x,-2) by 1/(x*x)

Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavfi/af_anequalizer: replace pow(10,x) by ff_exp10(x)

Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavfi/af_anequalizer: remove cabs, cexp dependencies

Replaces by real arithmetic. Tested the validity of these transformations separately.
Numerical differences are ~1e-15, and should not matter: it is not even
clear which is more precise mathematically.

Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

avfilter/avf_showspectrum: use FF_ARRAY_ELEMS()

Signed-off-by: Paul B Mahol <onemda@gmail.com>

x86/vf_stereo3d: make ff_anaglyph_sse4 work on x86_32

Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>

x86/vf_stereo3d: optimize register usage

Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>

avfilter/avf_showspectrum: make colors for log scale more user friendly

Previosly output was almost useless because background noise, due to
windowing function picked and which is not actually present in audio,
had too much brightness.
Now output of sine wave matches more with SoX.

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/avf_showspectrum: use ff_generate_window_func

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter: move window function generation into separate file

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/avf_showspectrum: add rscroll sliding mode

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avformat/http: Documented http_proxy option

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

lavf/mpegtsenc: add automatic bitstream filtering

lavf/matroskaenc: add automatic bitstream filtering

lavf: add internal API to append a bsf to a stream's list

lavf: add automatic bitstream filtering; bump version

This solves the problem discussed in https://ffmpeg.org/pipermail/ffmpeg-devel/2015-September/179238.html
by allowing AVCodec::write_header to be delayed until after packets have been
run through required bitstream filters in order to generate global extradata.

It also provides a mechanism by which a muxer can add a bitstream filter to a
stream automatically, rather than prompting the user to do so.

lavf/tee: use lavf API for applying bitstream filters

ffmpeg: use lavf API for applying bitstream filters

lavf: add API to apply a list of bsfs to a packet

doc/filters: add one more silenceremove example

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/af_silenceremove: make size of window user configurable

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/af_silenceremove: lower number of operations in for loop

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avcodec: properly check pkt_timebase for validity

Unset/invalid timebases have a zero numerator.
This makes the checks consistent with other timebase checks and fixes an
integer division by 0.

Revert "ffplay: Fix auto insertion point of rotation filter"

Revert requested by Balint Marton <cus@passwd.hu>

See: [FFmpeg-devel] ffplay: insertion point of the auto rotation filter - Github ticket #141

This reverts commit 9cc1e644f39ed6a32e3a9bf5d0d6d663ca5791ab.

avformat/qtpalette: Move ff_get_qtpalette() doxy to header

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avformat/matroskadec: palettized QuickTime video in Matroska

Palettized QuickTime video in Matroska has hitherto not been recognized
whatsoever, and the "palette" used has been completely random.

The patch for matroskadec.c fixes this issue by adding a palette side
data packet in matroska_deliver_packet(), much in the same way as it's
done in mov.c.

Video samples for testing are available at
https://drive.google.com/open?id=0B3_pEBoLs0faWElmM2FnLTZYNlk.

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avformat: factor ff_get_qtpalette() out of mov.c

This consists mainly of moving the palette handling from
the mov_parse_stsd_video() function to a new ff_get_qtpalette() function
in the new file qtpalette.c, which will be shared by both matroskadec.c and
mov.c.

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avformat/hls: Added http_proxy support

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avformat/hls: Remember to free HLSContext::headers

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avformat/http: Added http_proxy option

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avfilter/af_silenceremove: add peak detector

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avcodec/on2avc: Fix stability issues with scale_tab generation

This also simplifies the code
the resulting values are binary identical to what pow(10, i/10.0) produces

avio: add detail to avio_printf() size warning

Previous "currently size is limited" didn't give away
much in terms of useful info.

Signed-off-by: Reynaldo H. Verdejo Pinochet <reynaldo@osg.samsung.com>

ffserver: add a doctype heading to our HTML pages

Signed-off-by: Reynaldo H. Verdejo Pinochet <reynaldo@osg.samsung.com>

ffserver: HTML encode msgs instead of blindly stripping chars out

Fixes weirdness like our "??filename? not found" 404.

None of the chars being used from the previously blacklisted
list needs to be scaped on an UTF-8 document context

Signed-off-by: Reynaldo H. Verdejo Pinochet <reynaldo@osg.samsung.com>

ffserver: make our 404 explicitly HTML5/UTF-8

Signed-off-by: Reynaldo H. Verdejo Pinochet <reynaldo@osg.samsung.com>

lavc/snowenc: replace rint by lrint

avoids float to int cast.

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavc/dds: replace rint by lrint

avoids float to int cast.

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavc/texturedsp: replace rint by lrint

avoids float to int cast.

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavfi/avf_showspectrum: replace rint by lrint

avoids float to int cast.

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavfi/vf_hue: replace rint by lrint

avoids float to int cast.

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

doc/filters: note NetCDF dependency for sofalizer

Also add @url for link.

Signed-off-by: Lou Logan <lou@lrcd.com>

avcodec/on2avc: fix regression on icc since 5495c7f

Should fix the regression, and also speeds up table generation.
Tables tested on GNU/Linux+clang: they are identical to the ones prior
to 5495c7f. ff_exp10 caused one slight change in one entry, 50000 became
50001 due to somewhat incorrect rounding.

Untested on ICC; passes FATE on GNU/Linux+gcc.

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

avformat/img2dec: Reuse main IO context instead of reopening a single file

Fixes part of Ticket4849

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avformat/img2dec: Skip checking the input files existence if it has already been opened

Avoids a unneeded open
Fixes part of Ticket4849

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

ffplay: replace rint by lrint

avoids the float to integer cast, and is slightly superior in terms of
rounding ("Dutch/Gauss rounding").

Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavc/acelp_pitch_delay: replace exp2f(M_LOG2_10 *x) by ff_exp10f(x)

Suggested-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavc/wmaprodec: replace pow(10,x) by ff_exp10(x)

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavc/wmaenc: replace pow(10,x) by ff_exp10(x)

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavc/wmadec: replace pow(10,x) by ff_exp10(x)

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavc/opus: replace pow(10,x) by ff_exp10(x)

Reviewed-by: Rostislav Pehlivanov <atomnuker@gmail.com>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavc/on2avc: replace pow(10,x) by ff_exp10(x)

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavc/imc: replace pow(10,x) by ff_exp10(x)

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavc/dcaenc: replace pow(10,x) by ff_exp10(x)

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavc/cngdec: replace pow(10,x) by ff_exp10(x)

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavc/aacpsy: replace pow(10,x) by ff_exp10(x)

Reviewed-by: Rostislav Pehlivanov <atomnuker@gmail.com>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavfi/af_compand: replace pow(10,x) by ff_exp10(x)

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavfi/af_volume: replace pow(10,x) by ff_exp10(x)

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavfi/f_ebur128: replace pow(10,x) by ff_exp10(x)

Reviewed-by: Clément Bœsch <u@pkh.me>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavfi/vsrc_testsrc: replace pow(10,x) by ff_exp10(x)

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavu/eval: replace pow(10,x) by ff_exp10(x)

Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavc/libopusdec: replace pow(10,x) by ff_exp10(x)

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavu/internal: add ff_exp10

Fast, reasonably accurate 10^x. Alternative of detection of libm exp10 at configure
time is not worth the trouble, since it is anyway not POSIX or ISO C,
and currently only the GNU libm has it. Furthermore, GNU libm's variant
is ~ 2x slower, and is ironically not correctly rounded (2 ulp off) to justify all
that slowdown.

Reviewed-by: James Almer <jamrial@gmail.com>
Reviewed-by: Hendrik Leppkes <h.leppkes@gmail.com>
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavu/libm,configure: remove exp10, exp10f detection

Subsequent commit introduces ff_exp10 instead.

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavu/libm: misc, minor changes

Addition of comments marking the end of ifdef blocks, correction of an
incorrect (at double precision) M_LN2, removal of an unnecessary undef.

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

tools/bookmarklets: HTML code fixes

Minor HTML5 code changes.

Closes #167

avfilter/af_anequalizer: make cliping filter type actually useful

Previously result was ignored.

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/af_anequalizer: use pow instead of exp10

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter: add high-order parametric multiband equalizer filter

Signed-off-by: Paul B Mahol <onemda@gmail.com>

doc/muxers/mpegts: update doc after adding flag to embed an AC-3/E-AC-3 ES the DVB way

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>