libvpx: add configurations excluding VP9 for ARM Neon and ARM64.

libvpx: add configurations excluding VP9 for ARM Neon and ARM64.

This generates configuration files that can be used in cases when
VP9 is not desired due to binary size constraints.

BUG=612174
TESTED=Successfully executed:
cd third_party/libvpx
mkdir source/config/linux/arm-neon-no-vp9
mkdir source/config/linux/arm64-no-vp9
./generate_gypi.sh

[kjellander_chromium, 2016-05-19 08:48:15]

kjellander@chromium.org changed reviewers:
+ johannkoenig@chromium.org

[kjellander_chromium, 2016-05-19 08:48:15]

This is essentially what I meant in
https://bugs.chromium.org/p/chromium/issues/detail?id=612174

there's no need to commit this any time soon, but I wanted to check that it's
possible to get the VP9 exclusion with the GN build as well.

https://codereview.chromium.org/1997603002/diff/1/third_party/libvpx/generate...
File third_party/libvpx/generate_gypi.sh (right):

https://codereview.chromium.org/1997603002/diff/1/third_party/libvpx/generate...
third_party/libvpx/generate_gypi.sh:463: all_platforms="--enable-external-build
--enable-postproc --disable-install-srcs --enable-multi-res-encoding
--enable-temporal-denoising --disable-unit-tests --disable-install-docs
--disable-examples --enable-vp9-temporal-denoising --enable-vp9-postproc
--size-limit=16384x16384 $DISABLE_AVX --as=yasm"
Do I have to remove '--enable-vp9-temporal-denoising --enable-vp9-postproc' when
I do --disable-vp9 or are they just no-ops?
It'll make the code harder to read if I move them out into a separate variable
etc.

https://codereview.chromium.org/1997603002/diff/1/third_party/libvpx/libvpx_s...
File third_party/libvpx/libvpx_srcs_arm_neon_cpu_detect_intrinsics.gypi (left):

https://codereview.chromium.org/1997603002/diff/1/third_party/libvpx/libvpx_s...
third_party/libvpx/libvpx_srcs_arm_neon_cpu_detect_intrinsics.gypi:16:
'<(libvpx_source)/vp8/common/arm/neon/bilinearpredict_neon.c',
When I run ./generate_gypi.sh from a clean checkout these changes also show up,
so they're not related to my CL. I guess a separate update CL is needed for
these?

[Johann, 2016-05-24 20:13:38]

johannkoenig@google.com changed reviewers:
+ johannkoenig@google.com

[Johann, 2016-05-24 20:13:39]

The cpu thing is weird but otherwise it LGTM. Not sure how to modify
BUILD.gn/libvpx.gyp to make those targets available though

https://codereview.chromium.org/1997603002/diff/1/third_party/libvpx/generate...
File third_party/libvpx/generate_gypi.sh (right):

https://codereview.chromium.org/1997603002/diff/1/third_party/libvpx/generate...
third_party/libvpx/generate_gypi.sh:463: all_platforms="--enable-external-build
--enable-postproc --disable-install-srcs --enable-multi-res-encoding
--enable-temporal-denoising --disable-unit-tests --disable-install-docs
--disable-examples --enable-vp9-temporal-denoising --enable-vp9-postproc
--size-limit=16384x16384 $DISABLE_AVX --as=yasm"
On 2016/05/19 08:48:15, kjellander (chromium) wrote:
> Do I have to remove '--enable-vp9-temporal-denoising --enable-vp9-postproc'
when
> I do --disable-vp9 or are they just no-ops?
> It'll make the code harder to read if I move them out into a separate variable
> etc.

you can leave it, it is harmless.

https://codereview.chromium.org/1997603002/diff/1/third_party/libvpx/libvpx_s...
File third_party/libvpx/libvpx_srcs_arm_neon_cpu_detect_intrinsics.gypi (left):

https://codereview.chromium.org/1997603002/diff/1/third_party/libvpx/libvpx_s...
third_party/libvpx/libvpx_srcs_arm_neon_cpu_detect_intrinsics.gypi:16:
'<(libvpx_source)/vp8/common/arm/neon/bilinearpredict_neon.c',
On 2016/05/19 08:48:15, kjellander (chromium) wrote:
> When I run ./generate_gypi.sh from a clean checkout these changes also show
up,
> so they're not related to my CL. I guess a separate update CL is needed for
> these?

Hmm this looks very strange. cpu-detect is not used in Chromium so there may be
a bug in the configuration but this is odd.

https://codereview.chromium.org/1997603002/diff/1/third_party/libvpx/source/c...
File third_party/libvpx/source/config/linux/arm-neon-no-vp9/vp9_rtcd.h (right):

https://codereview.chromium.org/1997603002/diff/1/third_party/libvpx/source/c...
third_party/libvpx/source/config/linux/arm-neon-no-vp9/vp9_rtcd.h:1: #ifndef
VP9_RTCD_H_
Hmm this shouldn't get built but I think it's a weird side effect of how we
generate these files for chromium and not worth fixing.

[kjellander_chromium, 2017-09-29 07:09:10]

Description was changed from

==========
libvpx: add configurations excluding VP9 for ARM Neon and ARM64.

This generates configuration files that can be used in cases when
VP9 is not desired due to binary size constraints.

BUG=612174
TESTED=Successfully executed:
cd third_party/libvpx
mkdir source/config/linux/arm-neon-no-vp9
mkdir source/config/linux/arm64-no-vp9
./generate_gypi.sh
==========

to

==========
libvpx: add configurations excluding VP9 for ARM Neon and ARM64.

This generates configuration files that can be used in cases when
VP9 is not desired due to binary size constraints.

BUG=612174
TESTED=Successfully executed:
cd third_party/libvpx
mkdir source/config/linux/arm-neon-no-vp9
mkdir source/config/linux/arm64-no-vp9
./generate_gypi.sh
==========