first cut at making iOS work

first cut at making iOS work

Replace __arm__ with SK_CPU_ARM
add support for iOS simulator and device
fix const warning in iOSSampleApp
update gyp files

https://code.google.com/p/skia/issues/detail?id=900
tracks fixing missing arm assembly
Committed: https://code.google.com/p/skia/source/detail?r=5606

[reed1, 2012-09-20 14:03:11]

SkPreConfig.h has evolved a bit in the SSE case for this style of #define. Do
you think that might apply to ARM as well?

SK_CPU_ARM

or (if we want to encode variants, as we do for SSE)

SK_CPU_ARM_LEVEL ...

Just struck by SK_ARM as being a little brief :)

[EricB, 2012-09-20 14:06:17]

The GYP changes LGTM.

https://codereview.appspot.com/6552045/diff/1/gyp/SampleApp.gyp
File gyp/SampleApp.gyp (right):

https://codereview.appspot.com/6552045/diff/1/gyp/SampleApp.gyp#newcode281
gyp/SampleApp.gyp:281: '../../include/gpu',
Should this be '../include/gpu'?

https://codereview.appspot.com/6552045/diff/1/gyp/opts.gyp
File gyp/opts.gyp (right):

https://codereview.appspot.com/6552045/diff/1/gyp/opts.gyp#newcode77
gyp/opts.gyp:77: # this needs to be rewritten to be xcode-friendly
Have we (or should we) filed a bug?

https://codereview.appspot.com/6552045/diff/1/gyp/opts.gyp#newcode83
gyp/opts.gyp:83: [ '(skia_arch_type == "arm" and armv7 != 1) or (skia_arch_type
== "x86" and skia_os == "ios")', {
I think you can do "armv7 == 0" or even "not armv7".

https://codereview.appspot.com/6552045/diff/1/src/opts/opts_check_arm.cpp
File src/opts/opts_check_arm.cpp (right):

https://codereview.appspot.com/6552045/diff/1/src/opts/opts_check_arm.cpp#new...
src/opts/opts_check_arm.cpp:32: // FIXME: memset.arm.S is using syntax
incompatible with XCode
Again, we should probably file and link to a bug.

[caryclark1, 2012-09-20 15:06:24]

https://codereview.appspot.com/6552045/diff/1/gyp/SampleApp.gyp
File gyp/SampleApp.gyp (right):

https://codereview.appspot.com/6552045/diff/1/gyp/SampleApp.gyp#newcode281
gyp/SampleApp.gyp:281: '../../include/gpu',
done (removed)
On 2012/09/20 14:06:17, EricB wrote:
> Should this be '../include/gpu'?

https://codereview.appspot.com/6552045/diff/1/gyp/opts.gyp
File gyp/opts.gyp (right):

https://codereview.appspot.com/6552045/diff/1/gyp/opts.gyp#newcode77
gyp/opts.gyp:77: # this needs to be rewritten to be xcode-friendly
On 2012/09/20 14:06:17, EricB wrote:
> Have we (or should we) filed a bug?

Done.

https://codereview.appspot.com/6552045/diff/1/gyp/opts.gyp#newcode83
gyp/opts.gyp:83: [ '(skia_arch_type == "arm" and armv7 != 1) or (skia_arch_type
== "x86" and skia_os == "ios")', {
On 2012/09/20 14:06:17, EricB wrote:
> I think you can do "armv7 == 0" or even "not armv7".

Done.

https://codereview.appspot.com/6552045/diff/1/src/opts/opts_check_arm.cpp
File src/opts/opts_check_arm.cpp (right):

https://codereview.appspot.com/6552045/diff/1/src/opts/opts_check_arm.cpp#new...
src/opts/opts_check_arm.cpp:32: // FIXME: memset.arm.S is using syntax
incompatible with XCode
On 2012/09/20 14:06:17, EricB wrote:
> Again, we should probably file and link to a bug.

Done.

[caryclark1, 2012-09-20 15:07:20]

Replaced SK_ARM with SK_CPU_ARM, and moved definition from SkPostConfig.h to
SkPreConfig.h
On 2012/09/20 14:03:11, reed1 wrote:
> SkPreConfig.h has evolved a bit in the SSE case for this style of #define. Do
> you think that might apply to ARM as well?
> 
> SK_CPU_ARM
> 
> or (if we want to encode variants, as we do for SSE)
> 
> SK_CPU_ARM_LEVEL ...
> 
> Just struck by SK_ARM as being a little brief :)

[EricB, 2012-09-20 15:11:22]

https://codereview.appspot.com/6552045/diff/1/src/opts/opts_check_arm.cpp
File src/opts/opts_check_arm.cpp (right):

https://codereview.appspot.com/6552045/diff/1/src/opts/opts_check_arm.cpp#new...
src/opts/opts_check_arm.cpp:32: // FIXME: memset.arm.S is using syntax
incompatible with XCode
On 2012/09/20 15:06:24, caryclark1 wrote:
> On 2012/09/20 14:06:17, EricB wrote:
> > Again, we should probably file and link to a bug.
> 
> Done.

Can we link to the issue in this comment?

[reed1, 2012-09-20 15:17:27]

lgtm for the .h changes. I'll defer to eric for the gyp stuff

[EricB, 2012-09-20 15:19:05]

On 2012/09/20 15:17:27, reed1 wrote:
> lgtm for the .h changes. I'll defer to eric for the gyp stuff

Based on live discussion, ignore my comment about linking the issue.  GYP
changes LGTM.

[DerekS, 2012-09-20 15:38:09]

lgtm