Android handler: inject the application loading part from the shell .so.

Android handler: inject the application loading part from the shell .so.

To properly set thunks in the application being loaded, we need make_thunks() to
be called from an .so that has them set itself.

This patch extracts the application loading logic from the bootstrapping library
to the Mojo Shell .so. When an application is being bootstrapped, we inject the
loading function by passing a function pointer.

BUG=436508
R=qsr@chromium.org

Committed: https://chromium.googlesource.com/external/mojo/+/78ab413bd886107700215fe0acdc3e1adb11aae4

[ppi, 2014-12-03 12:24:46]

Patchset #2 (id:20001) has been deleted

[ppi, 2014-12-03 12:25:49]

Patchset #1 (id:1) has been deleted

[ppi, 2014-12-03 12:29:30]

Patchset #1 (id:40001) has been deleted

[ppi, 2014-12-03 12:30:24]

ppi@chromium.org changed reviewers:
+ qsr@chromium.org

[ppi, 2014-12-03 12:30:24]

ptal

https://codereview.chromium.org/772393003/diff/60001/examples/device_name/dev...
File examples/device_name/device_name.cc (right):

https://codereview.chromium.org/772393003/diff/60001/examples/device_name/dev...
examples/device_name/device_name.cc:63: // Call a core function to demonstrate
that the thunks were properly set.
Ben, is it possible to do it more concisely than in the three lines below?

[qsr, 2014-12-03 12:38:33]

https://codereview.chromium.org/772393003/diff/60001/examples/device_name/dev...
File examples/device_name/device_name.cc (right):

https://codereview.chromium.org/772393003/diff/60001/examples/device_name/dev...
examples/device_name/device_name.cc:63: // Call a core function to demonstrate
that the thunks were properly set.
On 2014/12/03 12:30:24, ppi wrote:
> Ben, is it possible to do it more concisely than in the three lines below?

 Yes, and no. I would use MojoGetTimeTicksNow() instead.

[ppi, 2014-12-03 13:06:05]

https://codereview.chromium.org/772393003/diff/60001/examples/device_name/dev...
File examples/device_name/device_name.cc (right):

https://codereview.chromium.org/772393003/diff/60001/examples/device_name/dev...
examples/device_name/device_name.cc:63: // Call a core function to demonstrate
that the thunks were properly set.
On 2014/12/03 12:38:33, qsr wrote:
> On 2014/12/03 12:30:24, ppi wrote:
> > Ben, is it possible to do it more concisely than in the three lines below?
> 
>  Yes, and no. I would use MojoGetTimeTicksNow() instead.

Thanks, done.

[qsr, 2014-12-03 13:27:28]

https://codereview.chromium.org/772393003/diff/100001/mojo/shell/BUILD.gn
File mojo/shell/BUILD.gn (right):

https://codereview.chromium.org/772393003/diff/100001/mojo/shell/BUILD.gn#new...
mojo/shell/BUILD.gn:161: "android/run_android_application_function.h",
The same file should never be in 2 different targets. I think you will need a
source_set for this file, and depend on it in the 2 target.

[ppi, 2014-12-03 14:27:53]

https://codereview.chromium.org/772393003/diff/100001/mojo/shell/BUILD.gn
File mojo/shell/BUILD.gn (right):

https://codereview.chromium.org/772393003/diff/100001/mojo/shell/BUILD.gn#new...
mojo/shell/BUILD.gn:161: "android/run_android_application_function.h",
On 2014/12/03 13:27:28, qsr wrote:
> The same file should never be in 2 different targets. I think you will need a
> source_set for this file, and depend on it in the 2 target.

Done.

[qsr, 2014-12-03 15:44:00]

lgtm

[ppi, 2014-12-03 15:49:41]

Committed patchset #6 (id:160001) manually as
78ab413bd886107700215fe0acdc3e1adb11aae4.