FFmpeg merge diff after patches.

FFmpeg merge diff after patches.

As titled, this is the diff between da1ba4e88b5bf723f8889798cb71f4e1cabb7a00 and our applied patches.

This is already commited, but due to rebase is difficult to see elsewhere.

[DaleCurtis, 2012-02-04 18:06:20]

https://chromiumcodereview.appspot.com/9317107/diff/1/libavformat/matroskadec.c
File libavformat/matroskadec.c (right):

https://chromiumcodereview.appspot.com/9317107/diff/1/libavformat/matroskadec...
libavformat/matroskadec.c:980: continue;
11b_mkv_buffer_overflow was partially upstreamed, and our 
local copy slightly incorrect. Upstream uses "continue", patch 
uses "return 0". Took continue since it seemed correct.

https://chromiumcodereview.appspot.com/9317107/diff/1/libavformat/oggdec.c
File libavformat/oggdec.c (left):

https://chromiumcodereview.appspot.com/9317107/diff/1/libavformat/oggdec.c#ol...
libavformat/oggdec.c:498: ogg_restore (s, 0);
Applied 35_oggdec_duration.patch, despite upstream saying the issue 
is fixed, some test cases in ffmpeg_unittests regressed.

https://chromiumcodereview.appspot.com/9317107/diff/1/libavformat/oggdec.c
File libavformat/oggdec.c (right):

https://chromiumcodereview.appspot.com/9317107/diff/1/libavformat/oggdec.c#ne...
libavformat/oggdec.c:674: pos_arg = seek_pos = avio_tell(s->pb);
Patch required swapping url_ftell w/ avio_tell.

https://chromiumcodereview.appspot.com/9317107/diff/1/libavformat/oggdec.c#ne...
libavformat/oggdec.c:675: seek_pts = ogg_read_timestamp(s, stream_index,
&pos_arg, avio_size(s->pb));
Patch required swapping url_fsize w/ avio_size.

https://chromiumcodereview.appspot.com/9317107/diff/1/libavformat/oggdec.c#ne...
libavformat/oggdec.c:687: avio_seek(s->pb, seek_pos, SEEK_SET);
Patch required swapping url_fseek w/ avio_seek.

https://chromiumcodereview.appspot.com/9317107/diff/1/libavutil/attributes.h
File libavutil/attributes.h (right):

https://chromiumcodereview.appspot.com/9317107/diff/1/libavutil/attributes.h#...
libavutil/attributes.h:92: // TODO(dalecurtis): Revert once
http://crbug.com/112673 is fixed.
See bug, I was able to remove all but one of the deprecated functions.

[scherkus (not reviewing), 2012-02-06 17:46:28]

few comments -- not sure if you wanted me to review this but I think I've
uncovered a few areas we should sanity check

https://chromiumcodereview.appspot.com/9317107/diff/1/configure
File configure (right):

https://chromiumcodereview.appspot.com/9317107/diff/1/configure#newcode2308
configure:2308: # O2 produces smaller/faster code on P4 and same performance on
Core/Atom.
this might be doable via configure script and build_ffmpeg.sh

https://chromiumcodereview.appspot.com/9317107/diff/1/configure#newcode3105
configure:3105: add_extralibs -lpthreadGC2 -lws2_32
this might be doable via configure script and build_ffmpeg.sh

https://chromiumcodereview.appspot.com/9317107/diff/1/configure#newcode3389
configure:3389: #   check_cflags -Werror=missing-prototypes
we should see whether this is still valid w/ our current windows toolchain

https://chromiumcodereview.appspot.com/9317107/diff/1/libavcodec/Makefile
File libavcodec/Makefile (right):

https://chromiumcodereview.appspot.com/9317107/diff/1/libavcodec/Makefile#new...
libavcodec/Makefile:604: OBJS-$(CONFIG_OGG_DEMUXER)             += dirac.o
mpeg12data.o vorbis_data.o
is this our patch?

https://chromiumcodereview.appspot.com/9317107/diff/1/libavformat/Makefile
File libavformat/Makefile (right):

https://chromiumcodereview.appspot.com/9317107/diff/1/libavformat/Makefile#ne...
libavformat/Makefile:189: OBJS-$(CONFIG_LIBSPEEX)                  +=
oggparsespeex.o
is this our patch?

https://chromiumcodereview.appspot.com/9317107/diff/1/libavformat/oggdec.c
File libavformat/oggdec.c (left):

https://chromiumcodereview.appspot.com/9317107/diff/1/libavformat/oggdec.c#ol...
libavformat/oggdec.c:498: ogg_restore (s, 0);
On 2012/02/04 18:06:20, DaleCurtis wrote:
> Applied 35_oggdec_duration.patch, despite upstream saying the issue 
> is fixed, some test cases in ffmpeg_unittests regressed. 

We should sanity check those tests for accuracy as well :)

https://chromiumcodereview.appspot.com/9317107/diff/1/libavformat/oggdec.c
File libavformat/oggdec.c (right):

https://chromiumcodereview.appspot.com/9317107/diff/1/libavformat/oggdec.c#ne...
libavformat/oggdec.c:674: pos_arg = seek_pos = avio_tell(s->pb);
On 2012/02/04 18:06:20, DaleCurtis wrote:
> Patch required swapping url_ftell w/ avio_tell.

For these function swaps is it because the patch was written before the ffmpeg
IO functions were updated?

[DaleCurtis, 2012-02-06 18:59:01]

https://chromiumcodereview.appspot.com/9317107/diff/1/configure
File configure (right):

https://chromiumcodereview.appspot.com/9317107/diff/1/configure#newcode2308
configure:2308: # O2 produces smaller/faster code on P4 and same performance on
Core/Atom.
On 2012/02/06 17:46:29, scherkus wrote:
> this might be doable via configure script and build_ffmpeg.sh

Sounds right, I can specify --extra-cflags=-O2 and it will override whatever
optimization level was set previously.

https://chromiumcodereview.appspot.com/9317107/diff/1/configure#newcode3105
configure:3105: add_extralibs -lpthreadGC2 -lws2_32
On 2012/02/06 17:46:29, scherkus wrote:
> this might be doable via configure script and build_ffmpeg.sh

Again, looks right. Here these would just be specified w/ win32.

https://chromiumcodereview.appspot.com/9317107/diff/1/configure#newcode3389
configure:3389: #   check_cflags -Werror=missing-prototypes
On 2012/02/06 17:46:29, scherkus wrote:
> we should see whether this is still valid w/ our current windows toolchain

Will do, we're up to 4.5.0 on the current toolchain.

https://chromiumcodereview.appspot.com/9317107/diff/1/libavcodec/Makefile
File libavcodec/Makefile (right):

https://chromiumcodereview.appspot.com/9317107/diff/1/libavcodec/Makefile#new...
libavcodec/Makefile:604: OBJS-$(CONFIG_OGG_DEMUXER)             += dirac.o
mpeg12data.o vorbis_data.o
On 2012/02/06 17:46:29, scherkus wrote:
> is this our patch?

Yup, 05_respect_flag_dirac_configure.patch

https://chromiumcodereview.appspot.com/9317107/diff/1/libavformat/Makefile
File libavformat/Makefile (right):

https://chromiumcodereview.appspot.com/9317107/diff/1/libavformat/Makefile#ne...
libavformat/Makefile:189: OBJS-$(CONFIG_LIBSPEEX)                  +=
oggparsespeex.o
On 2012/02/06 17:46:29, scherkus wrote:
> is this our patch?

Yup, 05_respect_flac_dirac_configure.patch

https://chromiumcodereview.appspot.com/9317107/diff/1/libavformat/oggdec.c
File libavformat/oggdec.c (left):

https://chromiumcodereview.appspot.com/9317107/diff/1/libavformat/oggdec.c#ol...
libavformat/oggdec.c:498: ogg_restore (s, 0);
On 2012/02/06 17:46:29, scherkus wrote:
> On 2012/02/04 18:06:20, DaleCurtis wrote:
> > Applied 35_oggdec_duration.patch, despite upstream saying the issue 
> > is fixed, some test cases in ffmpeg_unittests regressed. 
> 
> We should sanity check those tests for accuracy as well :)

Without this patch it gets the duration wrong by > 0.5 secs on sync1.ogg and
sync2.ogg. I'll recheck that these aren't fixed by the threads patch I made to
ffmpeg_unittests.

https://chromiumcodereview.appspot.com/9317107/diff/1/libavformat/oggdec.c
File libavformat/oggdec.c (right):

https://chromiumcodereview.appspot.com/9317107/diff/1/libavformat/oggdec.c#ne...
libavformat/oggdec.c:674: pos_arg = seek_pos = avio_tell(s->pb);
On 2012/02/06 17:46:29, scherkus wrote:
> On 2012/02/04 18:06:20, DaleCurtis wrote:
> > Patch required swapping url_ftell w/ avio_tell.
> 
> For these function swaps is it because the patch was written before the ffmpeg
> IO functions were updated?

Yes.

[scherkus (not reviewing), 2012-02-06 19:42:57]

LGTM w/ one more q

https://chromiumcodereview.appspot.com/9317107/diff/1/configure
File configure (right):

https://chromiumcodereview.appspot.com/9317107/diff/1/configure#newcode2308
configure:2308: # O2 produces smaller/faster code on P4 and same performance on
Core/Atom.
On 2012/02/06 18:59:01, DaleCurtis wrote:
> On 2012/02/06 17:46:29, scherkus wrote:
> > this might be doable via configure script and build_ffmpeg.sh
> 
> Sounds right, I can specify --extra-cflags=-O2 and it will override whatever
> optimization level was set previously.

another sanity check: does the -O2 we specify here match ffmpeg.gyp?

otherwise we're using different -O for win vs mac/linux and perhaps can drop
this entirely :)

[DaleCurtis, 2012-02-07 03:18:00]

https://chromiumcodereview.appspot.com/9317107/diff/1/configure
File configure (right):

https://chromiumcodereview.appspot.com/9317107/diff/1/configure#newcode2308
configure:2308: # O2 produces smaller/faster code on P4 and same performance on
Core/Atom.
On 2012/02/06 19:42:58, scherkus wrote:
> On 2012/02/06 18:59:01, DaleCurtis wrote:
> > On 2012/02/06 17:46:29, scherkus wrote:
> > > this might be doable via configure script and build_ffmpeg.sh
> > 
> > Sounds right, I can specify --extra-cflags=-O2 and it will override whatever
> > optimization level was set previously.
> 
> another sanity check: does the -O2 we specify here match ffmpeg.gyp?
> 
> otherwise we're using different -O for win vs mac/linux and perhaps can drop
> this entirely :)

Yes, we're using -O2 inside the gyp.