Remove GetPlatformDefaultEGLNativeDisplay() function.

Remove GetPlatformDefaultEGLNativeDisplay() function.

During EGL initialization GetPlatformDefaultEGLNativeDisplay() gets
called. This function is conditionally compiled for each platform. Under
Ozone the function calls into OzonePlatform to get the EGL
native_display to use. This ordering requires that //ui/gl depends on
//ui/ozone which we want to break.

This CL removes GetPlatformDefaultEGLNativeDisplay() and instead passes
in the EGL native_display to GLSurfaceEGL::InitializeOneOff() or
GLSurfaceEGL::InitializeDisplay(). For tests EGL_DEFAULT_DISPLAY is
always passed in.

The GL initialization and autogen binding code needs to be rearranged to
accomplish this. For EGL generated code split the client extension
bindings and extension bindings into two separate methods. The client
extension bindings need to be initialized before
GLSurfaceEGL::InitializeDisplay() is called while the extension bindings
need to be initialized after. Also fix a problem preventing
generate_bindings.py from running with missing extension information due
to a recent change.

BUG=611142
Committed: https://crrev.com/7dd36fe1e9bab00078cecd98616364ccf5e8b957
Cr-Commit-Position: refs/heads/master@{#406643}

[kylechar, 2016-07-08 15:42:27]

The CQ bit was checked by kylechar@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-07-08 15:42:51]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[kylechar, 2016-07-08 16:47:24]

kylechar@chromium.org changed reviewers:
+ piman@chromium.org

[piman, 2016-07-08 16:56:20]

LGTM, thanks for the refactor!

[commit-bot: I haz the power, 2016-07-08 19:00:52]

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

[commit-bot: I haz the power, 2016-07-08 19:00:53]

Dry run: Try jobs failed on following builders:
  win_chromium_rel_ng on master.tryserver.chromium.win (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.win/builders/win_chromium_rel_...)

[kylechar, 2016-07-13 14:46:46]

Patchset #4 (id:50001) has been deleted

[kylechar, 2016-07-14 17:11:17]

The CQ bit was checked by kylechar@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-07-14 17:11:27]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-07-14 17:55:34]

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

[commit-bot: I haz the power, 2016-07-14 17:55:35]

Dry run: Try jobs failed on following builders:
  linux_chromium_chromeos_ozone_rel_ng on master.tryserver.chromium.linux
(JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)

[kylechar, 2016-07-14 18:18:15]

Patchset #5 (id:90001) has been deleted

[kylechar, 2016-07-14 18:18:23]

The CQ bit was checked by kylechar@chromium.org

[kylechar, 2016-07-14 18:18:23]

The patchset sent to the CQ was uploaded after l-g-t-m from piman@chromium.org
Link to the patchset: https://codereview.chromium.org/2136553002/#ps130001
(title: "Fix test.")

[commit-bot: I haz the power, 2016-07-14 18:18:54]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-07-14 19:09:42]

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

[commit-bot: I haz the power, 2016-07-14 19:09:43]

Try jobs failed on following builders:
  cast_shell_android on master.tryserver.chromium.android (JOB_FAILED,
https://build.chromium.org/p/tryserver.chromium.android/builders/cast_shell_a...)

[kylechar, 2016-07-14 19:48:36]

The CQ bit was checked by kylechar@chromium.org

[commit-bot: I haz the power, 2016-07-14 19:48:57]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-07-14 21:07:23]

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

[commit-bot: I haz the power, 2016-07-14 21:07:24]

Try jobs failed on following builders:
  linux_android_rel_ng on master.tryserver.chromium.android (JOB_FAILED,
https://build.chromium.org/p/tryserver.chromium.android/builders/linux_androi...)

[kylechar, 2016-07-15 16:18:57]

The CQ bit was checked by kylechar@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-07-15 16:19:28]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[kylechar, 2016-07-15 16:42:19]

It seems like changing the initialization order to avoid adding a
GLSurfaceEGL::SetNativeDisplay() is caused crashes on Windows and Android. I
debugged the failure on Windows and was able to move the call to
g_driver_egl.InitializeExtensionBindings() a bit earlier to fix it.
Unfortunately there is still a crash on some Android where variables initialized
by g_driver_egl.InitializeExtensionBindings() are needed before
GLSurfaceEGL::InitializeDisplay() is called. This is the check that fails:

07-15 13:28:27.979 23523 23594 F chromium: [FATAL:texture_definition.cc(167)]
Check failed: gl::g_driver_egl.ext.b_EGL_KHR_image_base &&
gl::g_driver_egl.ext.b_EGL_KHR_gl_texture_2D_image &&
gl::g_driver_gl.ext.b_GL_OES_EGL_image. 

I think the safest bet at this point is to not change the initialization order
and explicitly set the EGL native display before it's used. Basically, going
back to what I had at https://codereview.chromium.org/2102443003/#ps40001,
except calling GLSurfaceEGL::SetNativeDisplay() earlier in
InitializeStaticGLBindings() to address the concerns you raised on the original
CL. 

Patch 7 has these changes. I'll wait for you to take another look and lgtm again
before committing in case there are anymore concerns.

[commit-bot: I haz the power, 2016-07-15 18:12:47]

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

[commit-bot: I haz the power, 2016-07-15 18:12:48]

Dry run: This issue passed the CQ dry run.

[kylechar, 2016-07-18 13:43:58]

Description was changed from

==========
Remove GetPlatformDefaultEGLNativeDisplay() function.

The function calls into OzonePlatform in order to get the EGL
native_display. This needs to be modified so that //ui/gl can not depend
on //ui/ozone. Remove the function call and instead pass in the EGL
native_display to GLSurfaceEGL::InitializeOneOff() and
GLSurfaceEGL::InitializeDisplay().

For tests EGL_DEFAULT_DISPLAY is passed in now. There is also a bit of
GL initialization code that is rearranged to accommodate where
g_native_display is being set.

BUG=611142
==========

to

==========
Remove GetPlatformDefaultEGLNativeDisplay() function.

During EGL initialization GetPlatformDefaultEGLNativeDisplay() gets
called.  This function is conditionally compiled for each platform and
under Ozone it gets the OzonePlatform object in order to ask for the EGL
native_display. This needs to be modified so that //ui/gl can not depend
on //ui/ozone.

The function is removed and GLSurfaceEGL::SetNativeDisplay() is added.
Instead of GLSurfaceEGL asking for the EGL native display, the EGL
native display is set before initialization. For EGLApiTests
EGL_DEFAULT_DISPLAY is passed in on all platforms.

BUG=611142
==========

[piman, 2016-07-18 22:20:43]

On 2016/07/15 16:42:19, kylechar wrote:
> It seems like changing the initialization order to avoid adding a
> GLSurfaceEGL::SetNativeDisplay() is caused crashes on Windows and Android. I
> debugged the failure on Windows and was able to move the call to
> g_driver_egl.InitializeExtensionBindings() a bit earlier to fix it.
> Unfortunately there is still a crash on some Android where variables
initialized
> by g_driver_egl.InitializeExtensionBindings() are needed before
> GLSurfaceEGL::InitializeDisplay() is called. This is the check that fails:
> 
> 07-15 13:28:27.979 23523 23594 F chromium: [FATAL:texture_definition.cc(167)]
> Check failed: gl::g_driver_egl.ext.b_EGL_KHR_image_base &&
> gl::g_driver_egl.ext.b_EGL_KHR_gl_texture_2D_image &&
> gl::g_driver_gl.ext.b_GL_OES_EGL_image. 


Mmh, if that code is called before InitializeOneOff, we'll have tons of
problems. Do you have a stack trace?

> 
> I think the safest bet at this point is to not change the initialization order
> and explicitly set the EGL native display before it's used. Basically, going
> back to what I had at https://codereview.chromium.org/2102443003/#ps40001,
> except calling GLSurfaceEGL::SetNativeDisplay() earlier in
> InitializeStaticGLBindings() to address the concerns you raised on the
original
> CL. 
> 
> Patch 7 has these changes. I'll wait for you to take another look and lgtm
again
> before committing in case there are anymore concerns.

[chromium-reviews, 2016-07-18 23:55:31]

No useful stack traces (no symbols) but here is the link to the logs from
linux_android_rel_ng where the failed DCHECKs were happening. It's with the
code from patch 6.

https://isolateserver.appspot.com/browse?namespace=default-gzip&digest=905e87...

Kyle

On Mon, Jul 18, 2016 at 6:20 PM, <piman@chromium.org> wrote:

> On 2016/07/15 16:42:19, kylechar wrote:
> > It seems like changing the initialization order to avoid adding a
> > GLSurfaceEGL::SetNativeDisplay() is caused crashes on Windows and
> Android. I
> > debugged the failure on Windows and was able to move the call to
> > g_driver_egl.InitializeExtensionBindings() a bit earlier to fix it.
> > Unfortunately there is still a crash on some Android where variables
> initialized
> > by g_driver_egl.InitializeExtensionBindings() are needed before
> > GLSurfaceEGL::InitializeDisplay() is called. This is the check that
> fails:
> >
> > 07-15 13:28:27.979 23523 23594 F chromium:
> [FATAL:texture_definition.cc(167)]
> > Check failed: gl::g_driver_egl.ext.b_EGL_KHR_image_base &&
> > gl::g_driver_egl.ext.b_EGL_KHR_gl_texture_2D_image &&
> > gl::g_driver_gl.ext.b_GL_OES_EGL_image.
>
>
> Mmh, if that code is called before InitializeOneOff, we'll have tons of
> problems. Do you have a stack trace?
>
> >
> > I think the safest bet at this point is to not change the initialization
> order
> > and explicitly set the EGL native display before it's used. Basically,
> going
> > back to what I had at
> https://codereview.chromium.org/2102443003/#ps40001,
> > except calling GLSurfaceEGL::SetNativeDisplay() earlier in
> > InitializeStaticGLBindings() to address the concerns you raised on the
> original
> > CL.
> >
> > Patch 7 has these changes. I'll wait for you to take another look and
> lgtm
> again
> > before committing in case there are anymore concerns.
>
>
>
> https://codereview.chromium.org/2136553002/
>

-- 
You received this message because you are subscribed to the Google Groups
"Chromium-reviews" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[piman, 2016-07-19 00:00:59]

On Mon, Jul 18, 2016 at 4:54 PM, Kyle Charbonneau <kylechar@google.com>
wrote:

> No useful stack traces (no symbols) but here is the link to the logs from
> linux_android_rel_ng where the failed DCHECKs were happening. It's with the
> code from patch 6.
>
>
>
https://isolateserver.appspot.com/browse?namespace=default-gzip&digest=905e87...
>
> Kyle
>

+sievers, any idea?

Antoine


>
> On Mon, Jul 18, 2016 at 6:20 PM, <piman@chromium.org> wrote:
>
>> On 2016/07/15 16:42:19, kylechar wrote:
>> > It seems like changing the initialization order to avoid adding a
>> > GLSurfaceEGL::SetNativeDisplay() is caused crashes on Windows and
>> Android. I
>> > debugged the failure on Windows and was able to move the call to
>> > g_driver_egl.InitializeExtensionBindings() a bit earlier to fix it.
>> > Unfortunately there is still a crash on some Android where variables
>> initialized
>> > by g_driver_egl.InitializeExtensionBindings() are needed before
>> > GLSurfaceEGL::InitializeDisplay() is called. This is the check that
>> fails:
>> >
>> > 07-15 13:28:27.979 23523 23594 F chromium:
>> [FATAL:texture_definition.cc(167)]
>> > Check failed: gl::g_driver_egl.ext.b_EGL_KHR_image_base &&
>> > gl::g_driver_egl.ext.b_EGL_KHR_gl_texture_2D_image &&
>> > gl::g_driver_gl.ext.b_GL_OES_EGL_image.
>>
>>
>> Mmh, if that code is called before InitializeOneOff, we'll have tons of
>> problems. Do you have a stack trace?
>>
>> >
>> > I think the safest bet at this point is to not change the
>> initialization order
>> > and explicitly set the EGL native display before it's used. Basically,
>> going
>> > back to what I had at
>> https://codereview.chromium.org/2102443003/#ps40001,
>> > except calling GLSurfaceEGL::SetNativeDisplay() earlier in
>> > InitializeStaticGLBindings() to address the concerns you raised on the
>> original
>> > CL.
>> >
>> > Patch 7 has these changes. I'll wait for you to take another look and
>> lgtm
>> again
>> > before committing in case there are anymore concerns.
>>
>>
>>
>> https://codereview.chromium.org/2136553002/
>>
>
>

-- 
You received this message because you are subscribed to the Google Groups
"Chromium-reviews" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[chromium-reviews, 2016-07-19 19:05:26]

Isn't the bug with ps6 just that it calls:

void DriverEGL::InitializeExtensionBindings() {
  std::string extensions(GetPlatformExtensions());

    and then -->

 std::string DriverEGL::GetPlatformExtensions() {
   EGLDisplay display = GLSurfaceEGL::GetHardwareDisplay();
   if (display == EGL_NO_DISPLAY)
     return ""; // no extensions

But setting |g_display| happens after InitializeExtensionBindings():

bool InitializeGLOneOffImplementation(GLImplementation impl,
                                      bool fallback_to_osmesa,
                                      bool gpu_service_logging,
                                      bool disable_gl_drawing) {
  bool initialized =
      InitializeStaticGLBindings(impl) && InitializeGLOneOffPlatform();


On Mon, Jul 18, 2016 at 5:00 PM, Antoine Labour <piman@chromium.org> wrote:

>
>
> On Mon, Jul 18, 2016 at 4:54 PM, Kyle Charbonneau <kylechar@google.com>
> wrote:
>
>> No useful stack traces (no symbols) but here is the link to the logs from
>> linux_android_rel_ng where the failed DCHECKs were happening. It's with the
>> code from patch 6.
>>
>>
>>
https://isolateserver.appspot.com/browse?namespace=default-gzip&digest=905e87...
>>
>> Kyle
>>
>
> +sievers, any idea?
>
> Antoine
>
>
>>
>> On Mon, Jul 18, 2016 at 6:20 PM, <piman@chromium.org> wrote:
>>
>>> On 2016/07/15 16:42:19, kylechar wrote:
>>> > It seems like changing the initialization order to avoid adding a
>>> > GLSurfaceEGL::SetNativeDisplay() is caused crashes on Windows and
>>> Android. I
>>> > debugged the failure on Windows and was able to move the call to
>>> > g_driver_egl.InitializeExtensionBindings() a bit earlier to fix it.
>>> > Unfortunately there is still a crash on some Android where variables
>>> initialized
>>> > by g_driver_egl.InitializeExtensionBindings() are needed before
>>> > GLSurfaceEGL::InitializeDisplay() is called. This is the check that
>>> fails:
>>> >
>>> > 07-15 13:28:27.979 23523 23594 F chromium:
>>> [FATAL:texture_definition.cc(167)]
>>> > Check failed: gl::g_driver_egl.ext.b_EGL_KHR_image_base &&
>>> > gl::g_driver_egl.ext.b_EGL_KHR_gl_texture_2D_image &&
>>> > gl::g_driver_gl.ext.b_GL_OES_EGL_image.
>>>
>>>
>>> Mmh, if that code is called before InitializeOneOff, we'll have tons of
>>> problems. Do you have a stack trace?
>>>
>>> >
>>> > I think the safest bet at this point is to not change the
>>> initialization order
>>> > and explicitly set the EGL native display before it's used. Basically,
>>> going
>>> > back to what I had at
>>> https://codereview.chromium.org/2102443003/#ps40001,
>>> > except calling GLSurfaceEGL::SetNativeDisplay() earlier in
>>> > InitializeStaticGLBindings() to address the concerns you raised on the
>>> original
>>> > CL.
>>> >
>>> > Patch 7 has these changes. I'll wait for you to take another look and
>>> lgtm
>>> again
>>> > before committing in case there are anymore concerns.
>>>
>>>
>>>
>>> https://codereview.chromium.org/2136553002/
>>>
>>
>>
>

-- 
You received this message because you are subscribed to the Google Groups
"Chromium-reviews" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[kylechar, 2016-07-19 19:25:13]

Yep, that appears to be exactly why. It's not that it was called to early.
Thanks for taking a look sievers. 

I'd still go with the solution I have in patch 8, if you're okay with that
piman?

On 2016/07/19 19:05:26, chromium-reviews wrote:
> Isn't the bug with ps6 just that it calls:
> 
> void DriverEGL::InitializeExtensionBindings() {
>   std::string extensions(GetPlatformExtensions());
> 
>     and then -->
> 
>  std::string DriverEGL::GetPlatformExtensions() {
>    EGLDisplay display = GLSurfaceEGL::GetHardwareDisplay();
>    if (display == EGL_NO_DISPLAY)
>      return ""; // no extensions
> 
> But setting |g_display| happens after InitializeExtensionBindings():
> 
> bool InitializeGLOneOffImplementation(GLImplementation impl,
>                                       bool fallback_to_osmesa,
>                                       bool gpu_service_logging,
>                                       bool disable_gl_drawing) {
>   bool initialized =
>       InitializeStaticGLBindings(impl) && InitializeGLOneOffPlatform();

[piman, 2016-07-19 20:47:56]

On Tue, Jul 19, 2016 at 12:25 PM, <kylechar@chromium.org> wrote:

> Yep, that appears to be exactly why. It's not that it was called to early.
> Thanks for taking a look sievers.
>
> I'd still go with the solution I have in patch 8, if you're okay with that
> piman?'
>

I would prefer PS6, with moving
 g_driver_egl.InitializeExtensionBindings(); to the bottom
of GLSurfaceEGL::InitializeDisplay, to keep clear the order of things and
reduce our dependency on statics - which, as you found out, make things
very hard to follow.

Antoine


>
> On 2016/07/19 19:05:26, chromium-reviews wrote:
> > Isn't the bug with ps6 just that it calls:
> >
> > void DriverEGL::InitializeExtensionBindings() {
> > std::string extensions(GetPlatformExtensions());
> >
> > and then -->
> >
> > std::string DriverEGL::GetPlatformExtensions() {
> > EGLDisplay display = GLSurfaceEGL::GetHardwareDisplay();
> > if (display == EGL_NO_DISPLAY)
> > return ""; // no extensions
> >
> > But setting |g_display| happens after InitializeExtensionBindings():
> >
> > bool InitializeGLOneOffImplementation(GLImplementation impl,
> > bool fallback_to_osmesa,
> > bool gpu_service_logging,
> > bool disable_gl_drawing) {
> > bool initialized =
> > InitializeStaticGLBindings(impl) && InitializeGLOneOffPlatform();
>
>
> https://codereview.chromium.org/2136553002/
>

-- 
You received this message because you are subscribed to the Google Groups
"Chromium-reviews" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[kylechar, 2016-07-20 18:33:38]

Description was changed from

==========
Remove GetPlatformDefaultEGLNativeDisplay() function.

During EGL initialization GetPlatformDefaultEGLNativeDisplay() gets
called.  This function is conditionally compiled for each platform and
under Ozone it gets the OzonePlatform object in order to ask for the EGL
native_display. This needs to be modified so that //ui/gl can not depend
on //ui/ozone.

The function is removed and GLSurfaceEGL::SetNativeDisplay() is added.
Instead of GLSurfaceEGL asking for the EGL native display, the EGL
native display is set before initialization. For EGLApiTests
EGL_DEFAULT_DISPLAY is passed in on all platforms.

BUG=611142
==========

to

==========
Remove GetPlatformDefaultEGLNativeDisplay() function.

During EGL initialization GetPlatformDefaultEGLNativeDisplay() gets
called. This function is conditionally compiled for each platform. Under
Ozone the function calls into OzonePlatform to get the EGL
native_display to use. This ordering requires that //ui/gl depends on
//ui/ozone which we want to break.

This CL removes GetPlatformDefaultEGLNativeDisplay() and instead passes
in the EGL native_display to GLSurfaceEGL::InitializeOneOff() or
GLSurfaceEGL::InitializeDisplay(). For tests EGL_DEFAULT_DISPLAY is
always passed in.

The GL initialization and autogen binding code needs to be rearranged to
accomplish this. For EGL generated code split the client extension
bindings and extension bindings into two separate methods. The client
extension bindings need to be initialized before
GLSurfaceEGL::InitializeDisplay() is called while the extension bindings
need to be initialized after. Also fix a problem preventing
generate_bindings.py from running with missing extension information due
to a recent change.

BUG=611142
==========

[kylechar, 2016-07-20 18:37:49]

The CQ bit was checked by kylechar@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-07-20 18:38:27]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[kylechar, 2016-07-20 18:40:34]

Alright, back to the original scheme. I split the client extension
initialization and extension initialization into two functions to allow the
ordering to be consistent. I also fixed a problem with the generate_bindings.py
script because it crashes at ToT.

[commit-bot: I haz the power, 2016-07-20 19:37:03]

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

[commit-bot: I haz the power, 2016-07-20 19:37:04]

Dry run: This issue passed the CQ dry run.

[kylechar, 2016-07-20 19:37:59]

The CQ bit was checked by kylechar@chromium.org

[kylechar, 2016-07-20 19:38:00]

The patchset sent to the CQ was uploaded after l-g-t-m from piman@chromium.org
Link to the patchset: https://codereview.chromium.org/2136553002/#ps190001
(title: "Back to original.")

[commit-bot: I haz the power, 2016-07-20 19:38:31]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-07-20 19:44:20]

Committed patchset #9 (id:190001)

[commit-bot: I haz the power, 2016-07-20 19:44:30]

CQ bit was unchecked.

[commit-bot: I haz the power, 2016-07-20 19:46:57]

Description was changed from

==========
Remove GetPlatformDefaultEGLNativeDisplay() function.

During EGL initialization GetPlatformDefaultEGLNativeDisplay() gets
called. This function is conditionally compiled for each platform. Under
Ozone the function calls into OzonePlatform to get the EGL
native_display to use. This ordering requires that //ui/gl depends on
//ui/ozone which we want to break.

This CL removes GetPlatformDefaultEGLNativeDisplay() and instead passes
in the EGL native_display to GLSurfaceEGL::InitializeOneOff() or
GLSurfaceEGL::InitializeDisplay(). For tests EGL_DEFAULT_DISPLAY is
always passed in.

The GL initialization and autogen binding code needs to be rearranged to
accomplish this. For EGL generated code split the client extension
bindings and extension bindings into two separate methods. The client
extension bindings need to be initialized before
GLSurfaceEGL::InitializeDisplay() is called while the extension bindings
need to be initialized after. Also fix a problem preventing
generate_bindings.py from running with missing extension information due
to a recent change.

BUG=611142
==========

to

==========
Remove GetPlatformDefaultEGLNativeDisplay() function.

During EGL initialization GetPlatformDefaultEGLNativeDisplay() gets
called. This function is conditionally compiled for each platform. Under
Ozone the function calls into OzonePlatform to get the EGL
native_display to use. This ordering requires that //ui/gl depends on
//ui/ozone which we want to break.

This CL removes GetPlatformDefaultEGLNativeDisplay() and instead passes
in the EGL native_display to GLSurfaceEGL::InitializeOneOff() or
GLSurfaceEGL::InitializeDisplay(). For tests EGL_DEFAULT_DISPLAY is
always passed in.

The GL initialization and autogen binding code needs to be rearranged to
accomplish this. For EGL generated code split the client extension
bindings and extension bindings into two separate methods. The client
extension bindings need to be initialized before
GLSurfaceEGL::InitializeDisplay() is called while the extension bindings
need to be initialized after. Also fix a problem preventing
generate_bindings.py from running with missing extension information due
to a recent change.

BUG=611142

Committed: https://crrev.com/7dd36fe1e9bab00078cecd98616364ccf5e8b957
Cr-Commit-Position: refs/heads/master@{#406643}
==========

[commit-bot: I haz the power, 2016-07-20 19:46:58]

Patchset 9 (id:??) landed as
https://crrev.com/7dd36fe1e9bab00078cecd98616364ccf5e8b957
Cr-Commit-Position: refs/heads/master@{#406643}

[piman, 2016-07-20 21:45:24]

LGTM, thanks!