libvpx: Add VP9 decoder.

libvpx: Add VP9 decoder.

This CL also makes changes to generated files to make the process
more automatic. You should not need to build on the specific
platform anymore to make changes to vpx_config.h, vpx_config.c,
and vpx_config.asm as generate_gypi.sh will generate these files
for all platforms. See README.chromium for the steps to update
libvpx.

The changes that are most interesting are in generate_gypi.sh and
libvpx.gyp. Most of the other changes deal with the libvpx roll to
vp9-preview.

- Add VP9 decoder (VP9 encoder is turned off).
- Turned off CPU detect for arm and arm-neon.
- Have StdInt is turned off for windows because target is vs9. But
  we are not currently using this flag so we should be OK.
- Generate vpx_config.h, vpx_config.c, and vpx_config.asm for all
  platforms.
- Generate vpx_version.h.
- Removed vpx_rtcd.h from all platforms.
- Added vp8_rtcd.h, vp9_rtcd.h, and vpx_scale_rtcd.h for all
  platforms.
- http://crbug/162675 shoudl be fixed.

BUG=165504
TEST=
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=172738

[Alpha Left Google, 2012-12-12 21:52:56]

https://codereview.chromium.org/11555023/diff/1/generate_gypi.sh
File generate_gypi.sh (right):

https://codereview.chromium.org/11555023/diff/1/generate_gypi.sh#newcode32
generate_gypi.sh:32: grep -v '_offsets\.c' | \
This is hard to read, can you break it into multiple lines? e.g.

source_list=`echo $source_list | grep -v '_offsets\.c'`
...

https://codereview.chromium.org/11555023/diff/1/generate_gypi.sh#newcode155
generate_gypi.sh:155: all_platforms="--enable-external-build --enable-postproc
--disable-install-srcs --enable-multi-res-encoding --enable-temporal-denoising
--disable-vp9-encoder --disable-unit-tests --disable-install-docs
--disable-examples"
Move this close to echo "Generate Config Files".

https://codereview.chromium.org/11555023/diff/1/generate_gypi.sh#newcode163
generate_gypi.sh:163: # TODO(fgalligan): Is "--disable-fast-unaligned" needed on
mipsel?
Move this down to the bottom.

https://codereview.chromium.org/11555023/diff/1/generate_gypi.sh#newcode170
generate_gypi.sh:170: gen_config_files linux/mipsel "--target=mips32-linux-gcc
--disable-fast-unaligned ${all_platforms}"
Move mipsel and comments down to the bottom. MIPS is out least concerned
platform.

https://codereview.chromium.org/11555023/diff/1/libvpx.gyp
File libvpx.gyp (right):

https://codereview.chromium.org/11555023/diff/1/libvpx.gyp#newcode411
libvpx.gyp:411: {
Merge problems? This appears twice.

https://codereview.chromium.org/11555023/diff/1/libvpx.gyp#newcode473
libvpx.gyp:473: 'target_name': 'gen_asm_offsets_vp9',
Can you combine this with gen_asm_offsets?

[fgalligan1, 2012-12-13 00:07:54]

https://codereview.chromium.org/11555023/diff/1/generate_gypi.sh
File generate_gypi.sh (right):

https://codereview.chromium.org/11555023/diff/1/generate_gypi.sh#newcode32
generate_gypi.sh:32: grep -v '_offsets\.c' | \
On 2012/12/12 21:52:56, Alpha wrote:
> This is hard to read, can you break it into multiple lines? e.g.
> 
> source_list=`echo $source_list | grep -v '_offsets\.c'`
> ...
Done.

https://codereview.chromium.org/11555023/diff/1/generate_gypi.sh#newcode155
generate_gypi.sh:155: all_platforms="--enable-external-build --enable-postproc
--disable-install-srcs --enable-multi-res-encoding --enable-temporal-denoising
--disable-vp9-encoder --disable-unit-tests --disable-install-docs
--disable-examples"
On 2012/12/12 21:52:56, Alpha wrote:
> Move this close to echo "Generate Config Files".

Done.

https://codereview.chromium.org/11555023/diff/1/generate_gypi.sh#newcode163
generate_gypi.sh:163: # TODO(fgalligan): Is "--disable-fast-unaligned" needed on
mipsel?
On 2012/12/12 21:52:56, Alpha wrote:
> Move this down to the bottom.

Done.

https://codereview.chromium.org/11555023/diff/1/generate_gypi.sh#newcode170
generate_gypi.sh:170: gen_config_files linux/mipsel "--target=mips32-linux-gcc
--disable-fast-unaligned ${all_platforms}"
On 2012/12/12 21:52:56, Alpha wrote:
> Move mipsel and comments down to the bottom. MIPS is out least concerned
> platform.
I didn't move because then I would need to create a temp directory again.

https://codereview.chromium.org/11555023/diff/1/libvpx.gyp
File libvpx.gyp (right):

https://codereview.chromium.org/11555023/diff/1/libvpx.gyp#newcode411
libvpx.gyp:411: {
On 2012/12/12 21:52:56, Alpha wrote:
> Merge problems? This appears twice.
No. I'm new wrt GYP. So can I define this action and call it from
gen_asm_offsets_vp8 & gen_asm_offsets_vp9?

https://codereview.chromium.org/11555023/diff/1/libvpx.gyp#newcode473
libvpx.gyp:473: 'target_name': 'gen_asm_offsets_vp9',
On 2012/12/12 21:52:56, Alpha wrote:
> Can you combine this with gen_asm_offsets?
I probably could if I knew gyp more, but there are some subtle differences with
naming and files between vp8 and vp9.

[Alpha Left Google, 2012-12-13 00:15:22]

https://codereview.chromium.org/11555023/diff/1/libvpx.gyp
File libvpx.gyp (right):

https://codereview.chromium.org/11555023/diff/1/libvpx.gyp#newcode411
libvpx.gyp:411: {
On 2012/12/13 00:07:54, fgalligan1 wrote:
> On 2012/12/12 21:52:56, Alpha wrote:
> > Merge problems? This appears twice.
> No. I'm new wrt GYP. So can I define this action and call it from
> gen_asm_offsets_vp8 & gen_asm_offsets_vp9?

I'm not sure what you're trying to do here, these two actions look the same to
me, so they will both be executed.

https://codereview.chromium.org/11555023/diff/1/libvpx.gyp#newcode473
libvpx.gyp:473: 'target_name': 'gen_asm_offsets_vp9',
On 2012/12/13 00:07:54, fgalligan1 wrote:
> On 2012/12/12 21:52:56, Alpha wrote:
> > Can you combine this with gen_asm_offsets?
> I probably could if I knew gyp more, but there are some subtle differences
with
> naming and files between vp8 and vp9.

I think you should try to merge them. In fact libvpx_asm_offsets_vp9 should
merge with libvpx_asm_offsets_vp8. That way vp9_asm_enc_offsets.o just replaces
asm_dec_offsets.o, no?

[fgalligan1, 2012-12-13 00:38:08]

https://codereview.chromium.org/11555023/diff/1/libvpx.gyp
File libvpx.gyp (right):

https://codereview.chromium.org/11555023/diff/1/libvpx.gyp#newcode411
libvpx.gyp:411: {
On 2012/12/13 00:15:22, Alpha wrote:
> On 2012/12/13 00:07:54, fgalligan1 wrote:
> > On 2012/12/12 21:52:56, Alpha wrote:
> > > Merge problems? This appears twice.
> > No. I'm new wrt GYP. So can I define this action and call it from
> > gen_asm_offsets_vp8 & gen_asm_offsets_vp9?
> 
> I'm not sure what you're trying to do here, these two actions look the same to
> me, so they will both be executed.
Oh I see I was looking at the actual file. Something is messed up with the side
by side diff. In the real file there is only one action. If you look at the
unified diff it is correct.

https://codereview.chromium.org/11555023/diff/1/libvpx.gyp#newcode473
libvpx.gyp:473: 'target_name': 'gen_asm_offsets_vp9',
On 2012/12/13 00:15:22, Alpha wrote:
> On 2012/12/13 00:07:54, fgalligan1 wrote:
> > On 2012/12/12 21:52:56, Alpha wrote:
> > > Can you combine this with gen_asm_offsets?
> > I probably could if I knew gyp more, but there are some subtle differences
> with
> > naming and files between vp8 and vp9.
> 
> I think you should try to merge them. In fact libvpx_asm_offsets_vp9 should
> merge with libvpx_asm_offsets_vp8. That way vp9_asm_enc_offsets.o just
replaces
> asm_dec_offsets.o, no?
OK, how about I add a TODO.

[Alpha Left Google, 2012-12-13 00:40:20]

alright, lgtm. Please make sure those are fixed, the gyp file is getting
complicated already.

[holmer, 2012-12-13 07:59:50]

On 2012/12/13 00:40:20, Alpha wrote:
> alright, lgtm. Please make sure those are fixed, the gyp file is getting
> complicated already.

The following files look like backups which were accidentally added:

source/libvpx/build/x86-msvs/obj_int_extract.bat.orig
source/libvpx/vpx_ports/arm_cpudetect.c.orig

[holmer, 2012-12-13 08:08:08]

I know this has been committed already, but please have a look at my two
comments anyway.

https://codereview.chromium.org/11555023/diff/14126/libvpx.gyp
File libvpx.gyp (right):

https://codereview.chromium.org/11555023/diff/14126/libvpx.gyp#newcode415
libvpx.gyp:415: 'action_name': 'unpack_lib_posix',
It's not clear to me what is different with this action compared to the one at
line 391. Can you clarify this with a comment?

[fgalligan1, 2012-12-13 17:09:23]

I will make a CL to remove the .orig files.

https://codereview.chromium.org/11555023/diff/14126/libvpx.gyp
File libvpx.gyp (right):

https://codereview.chromium.org/11555023/diff/14126/libvpx.gyp#newcode415
libvpx.gyp:415: 'action_name': 'unpack_lib_posix',
On 2012/12/13 08:08:08, holmer wrote:
> It's not clear to me what is different with this action compared to the one at
> line 391. Can you clarify this with a comment?
There is no difference. Actually the side by side diff is broken for this file.
If you look at the unified diff there will be only one unpack_lib_posix action.

[holmer, 2012-12-14 08:51:23]

lgtm

https://codereview.chromium.org/11555023/diff/14126/libvpx.gyp
File libvpx.gyp (right):

https://codereview.chromium.org/11555023/diff/14126/libvpx.gyp#newcode415
libvpx.gyp:415: 'action_name': 'unpack_lib_posix',
Weird. It does look okay in the unified diff.

On 2012/12/13 17:09:23, fgalligan1 wrote:
> On 2012/12/13 08:08:08, holmer wrote:
> > It's not clear to me what is different with this action compared to the one
at
> > line 391. Can you clarify this with a comment?
> There is no difference. Actually the side by side diff is broken for this
file.
> If you look at the unified diff there will be only one unpack_lib_posix
action.