Enable BUILD file compilation of skia and dm with --config=android_arm.

Enable BUILD file compilation of skia and dm with --config=android_arm.

Corresponding google3 cl is 105687529.

BUG=skia:

Committed: https://skia.googlesource.com/skia/+/86ea33e96341ec3de03bcb6aef9a2cfce371df6a

[dogben, 2015-10-18 03:32:36]

Patchset #1 (id:1) has been deleted

[dogben, 2015-10-18 03:40:10]

benjaminwagner@google.com changed reviewers:
+ mtklein@google.com

[mtklein, 2015-10-20 17:48:55]

https://codereview.chromium.org/1414643002/diff/20001/BUILD.public
File BUILD.public (right):

https://codereview.chromium.org/1414643002/diff/20001/BUILD.public#newcode25
BUILD.public:25: "src/doc/SkDocument_PDF.cpp",  # Doesn't compile on Android.
This is surprising.  SkDocument_PDF should work fine on Android.  Let's follow
this up.

https://codereview.chromium.org/1414643002/diff/20001/BUILD.public#newcode38
BUILD.public:38: "src/utils/SkThreadUtils_pthread_linux.cpp",  # Linux-only.
Move to ports?
Most of this SkThreadUtils stuff is pointless.  I'm trying to kill it off here:
https://codereview.chromium.org/1408213005/

https://codereview.chromium.org/1414643002/diff/20001/BUILD.public#newcode50
BUILD.public:50: "src/gpu/gl/GrGLCreateNativeInterface_none.cpp",
Why resort g above f?

https://codereview.chromium.org/1414643002/diff/20001/BUILD.public#newcode337
BUILD.public:337: "SK_ARM_HAS_OPTIONAL_NEON",
Probably safe to go with SK_ARM_HAS_NEON.  We intend to deprecate optional-NEON
pretty soon (once Chrome stops using it).

https://codereview.chromium.org/1414643002/diff/20001/BUILD.public#newcode393
BUILD.public:393: copts = COPTS + ["-fomit-frame-pointer"],
Try dropping this?  I think this is no longer true.

https://codereview.chromium.org/1414643002/diff/20001/tools/flags/SkCommandLi...
File tools/flags/SkCommandLineFlags.cpp (right):

https://codereview.chromium.org/1414643002/diff/20001/tools/flags/SkCommandLi...
tools/flags/SkCommandLineFlags.cpp:16: DECLARE_bool(undefok)
.... wow, this works?  That's a huge surprise.  Presumably, in this case we're
hopping onto a gflags undefok bool*.  That's sort of unexpectedly convenient.

It'd be nice to indent these DECLARE_ 4 spaces inside the #if.

[dogben, 2015-10-20 22:20:23]

Patchset #2 (id:40001) has been deleted

[dogben, 2015-10-20 22:20:43]

The CQ bit was checked by benjaminwagner@google.com to run a CQ dry run

[commit-bot: I haz the power, 2015-10-20 22:20:47]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1414643002/60001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1414643002/60001

[dogben, 2015-10-20 22:21:03]

Rebased.

https://codereview.chromium.org/1414643002/diff/20001/BUILD.public
File BUILD.public (right):

https://codereview.chromium.org/1414643002/diff/20001/BUILD.public#newcode25
BUILD.public:25: "src/doc/SkDocument_PDF.cpp",  # Doesn't compile on Android.
On 2015/10/20 17:48:55, mtklein wrote:
> This is surprising.  SkDocument_PDF should work fine on Android.  Let's follow
> this up.

There is an issue with sfntly. It seems to compile if I disable the sfntly
define.

https://codereview.chromium.org/1414643002/diff/20001/BUILD.public#newcode38
BUILD.public:38: "src/utils/SkThreadUtils_pthread_linux.cpp",  # Linux-only.
Move to ports?
On 2015/10/20 17:48:55, mtklein wrote:
> Most of this SkThreadUtils stuff is pointless.  I'm trying to kill it off
here:
> https://codereview.chromium.org/1408213005/

Removed, modulo win.

https://codereview.chromium.org/1414643002/diff/20001/BUILD.public#newcode50
BUILD.public:50: "src/gpu/gl/GrGLCreateNativeInterface_none.cpp",
On 2015/10/20 17:48:55, mtklein wrote:
> Why resort g above f?

Cuz I ain't no gud at them their alfabits.

https://codereview.chromium.org/1414643002/diff/20001/BUILD.public#newcode337
BUILD.public:337: "SK_ARM_HAS_OPTIONAL_NEON",
On 2015/10/20 17:48:55, mtklein wrote:
> Probably safe to go with SK_ARM_HAS_NEON.  We intend to deprecate
optional-NEON
> pretty soon (once Chrome stops using it).

So, SK_ARM_HAS_NEON enables this line in SkBlitter_RGB16.cpp:
https://code.google.com/p/chromium/codesearch#chromium/src/third_party/skia/s...

...which gives me this compile error:
error You must enable NEON instructions (e.g. -mfloat-abi=softfp -mfpu=neon) to
use arm_neon.h

Given that SkBlitter_RGB16.cpp seems to both depend on and be a dependency of
things currently in :skia, putting it in a separate library with different copts
seems like more trouble than it's worth.

https://codereview.chromium.org/1414643002/diff/20001/BUILD.public#newcode393
BUILD.public:393: copts = COPTS + ["-fomit-frame-pointer"],
On 2015/10/20 17:48:55, mtklein wrote:
> Try dropping this?  I think this is no longer true.

Done.

https://codereview.chromium.org/1414643002/diff/20001/tools/flags/SkCommandLi...
File tools/flags/SkCommandLineFlags.cpp (right):

https://codereview.chromium.org/1414643002/diff/20001/tools/flags/SkCommandLi...
tools/flags/SkCommandLineFlags.cpp:16: DECLARE_bool(undefok)
On 2015/10/20 17:48:55, mtklein wrote:
> .... wow, this works?  That's a huge surprise.  Presumably, in this case we're
> hopping onto a gflags undefok bool*.  That's sort of unexpectedly convenient.
> 
> It'd be nice to indent these DECLARE_ 4 spaces inside the #if.

s/convenient/inconvenient/, since whether it's defined depends on the platform.
:-/

[commit-bot: I haz the power, 2015-10-20 22:31:42]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2015-10-20 22:31:43]

Dry run: This issue passed the CQ dry run.

[mtklein, 2015-10-21 18:06:06]

https://codereview.chromium.org/1414643002/diff/60001/BUILD.public
File BUILD.public (right):

https://codereview.chromium.org/1414643002/diff/60001/BUILD.public#newcode319
BUILD.public:319: "SK_ARM_HAS_OPTIONAL_NEON",
OOOH right, SK_ARM_HAS_NEON only works if we set -mfpu=neon everywhere.
In fact, it should set itself if -mfpu=neon is set.

[dogben, 2015-10-21 20:14:00]

On 2015/10/21 at 18:06:07, mtklein wrote:
> 

Did you intend to add a comment or to LGTM?

[mtklein, 2015-10-22 21:58:32]

https://codereview.chromium.org/1414643002/diff/60001/BUILD.public
File BUILD.public (right):

https://codereview.chromium.org/1414643002/diff/60001/BUILD.public#newcode319
BUILD.public:319: "SK_ARM_HAS_OPTIONAL_NEON",
On 2015/10/21 18:06:06, mtklein wrote:
> OOOH right, SK_ARM_HAS_NEON only works if we set -mfpu=neon everywhere.
> In fact, it should set itself if -mfpu=neon is set.

Can we set -mfpu=neon everywhere?

[dogben, 2015-10-24 22:02:14]

Patchset #3 (id:80001) has been deleted

[dogben, 2015-10-24 22:05:52]

https://codereview.chromium.org/1414643002/diff/60001/BUILD.public
File BUILD.public (right):

https://codereview.chromium.org/1414643002/diff/60001/BUILD.public#newcode319
BUILD.public:319: "SK_ARM_HAS_OPTIONAL_NEON",
On 2015/10/22 21:58:32, mtklein wrote:
> On 2015/10/21 18:06:06, mtklein wrote:
> > OOOH right, SK_ARM_HAS_NEON only works if we set -mfpu=neon everywhere.
> > In fact, it should set itself if -mfpu=neon is set.
> 
> Can we set -mfpu=neon everywhere?

Done.

[mtklein, 2015-10-26 15:35:02]

lgtm

[dogben, 2015-10-26 17:36:15]

The CQ bit was checked by benjaminwagner@google.com

[commit-bot: I haz the power, 2015-10-26 17:36:20]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1414643002/100001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1414643002/100001

[commit-bot: I haz the power, 2015-10-26 17:46:28]

Committed patchset #3 (id:100001) as
https://skia.googlesource.com/skia/+/86ea33e96341ec3de03bcb6aef9a2cfce371df6a