Support Crazy Linker in WebAPK to load Chrome's uncompressed native libraries.

Support Crazy Linker in WebAPK to load Chrome's uncompressed native libraries.

Add the host's native library directories to WebAPK's classLoader's pathlist,
so System.loadLibrary() works for WebAPK.


BUG=524670

[Xi Han, 2016-02-18 22:37:00]

Patchset #1 (id:1) has been deleted

[Xi Han, 2016-02-18 22:42:51]

Patchset #1 (id:20001) has been deleted

[Xi Han, 2016-02-19 21:45:52]

Patchset #1 (id:40001) has been deleted

[Xi Han, 2016-02-19 21:45:57]

Patchset #1 (id:60001) has been deleted

[Xi Han, 2016-02-19 21:48:11]

Description was changed from

==========
Use Crazy Linker in WebAPK to load Chrome's uncompressed native libraries.


BUG=524670
==========

to

==========
Support Crazy Linker in WebAPK to load Chrome's uncompressed native libraries.

Add the host's native library directories to WebAPK's classLoader's pathlist, so
System.loadLibrary() works for WebAPK.


BUG=524670
==========

[Xi Han, 2016-02-19 21:48:24]

Description was changed from

==========
Support Crazy Linker in WebAPK to load Chrome's uncompressed native libraries.

Add the host's native library directories to WebAPK's classLoader's pathlist, so
System.loadLibrary() works for WebAPK.


BUG=524670
==========

to

==========
Support Crazy Linker in WebAPK to load Chrome's uncompressed native libraries.

Add the host's native library directories to WebAPK's classLoader's pathlist,
so System.loadLibrary() works for WebAPK.


BUG=524670
==========

[Xi Han, 2016-02-19 21:48:39]

hanxi@chromium.org changed reviewers:
+ yfriedman@chromium.org

[Yaron, 2016-02-19 22:10:20]

https://codereview.chromium.org/1710083004/diff/80001/base/android/java/src/o...
File base/android/java/src/org/chromium/base/library_loader/LibraryLoader.java
(right):

https://codereview.chromium.org/1710083004/diff/80001/base/android/java/src/o...
base/android/java/src/org/chromium/base/library_loader/LibraryLoader.java:340:
// In minted runtime, getApplicationContext() of its host context returns null,
since
I wonder if this might bite us elsewhere but seems ok

https://codereview.chromium.org/1710083004/diff/80001/build/android/rezip/Rez...
File build/android/rezip/RezipApk.java (right):

https://codereview.chromium.org/1710083004/diff/80001/build/android/rezip/Rez...
build/android/rezip/RezipApk.java:310: if (name.endsWith(".apk") &&
!name.endsWith("MintingExample.template.apk")) {
You don't need to do this now but I suspect that when we upstream this, we could
just have this warn (rather than fail) that it won't recursively rezip an
internal apk.

https://codereview.chromium.org/1710083004/diff/80001/web_apks/minting_exampl...
File web_apks/minting_example/src/org/chromium/minting/MintingApplication.java
(right):

https://codereview.chromium.org/1710083004/diff/80001/web_apks/minting_exampl...
web_apks/minting_example/src/org/chromium/minting/MintingApplication.java:47:
concatAndRemoveEndDuplicates(mintDirsAsList, (List<Object>) hostCurrentDirs);
Are the duplicates a concern? Are they not just ignored? If they are ignored,
just use Reflect.concatArrays

https://codereview.chromium.org/1710083004/diff/80001/web_apks/minting_exampl...
web_apks/minting_example/src/org/chromium/minting/MintingApplication.java:49:
for (Object o: combined) {
nit: the previous line and this section are just for debugging, right? Please
remove. Same for the else block

https://codereview.chromium.org/1710083004/diff/80001/web_apks/minting_exampl...
File web_apks/minting_example/src/org/chromium/minting/Reflect.java (right):

https://codereview.chromium.org/1710083004/diff/80001/web_apks/minting_exampl...
web_apks/minting_example/src/org/chromium/minting/Reflect.java:13: final class
Reflect {
Is this a straight copy?

[Xi Han, 2016-02-24 19:09:39]

I just noticed that you had commented on this review. Updated accordingly.
Please take a look. Thanks!

https://codereview.chromium.org/1710083004/diff/80001/base/android/java/src/o...
File base/android/java/src/org/chromium/base/library_loader/LibraryLoader.java
(right):

https://codereview.chromium.org/1710083004/diff/80001/base/android/java/src/o...
base/android/java/src/org/chromium/base/library_loader/LibraryLoader.java:340:
// In minted runtime, getApplicationContext() of its host context returns null,
since
On 2016/02/19 22:10:19, Yaron wrote:
> I wonder if this might bite us elsewhere but seems ok

Do you know when the splitSourceDirs are used?

https://codereview.chromium.org/1710083004/diff/80001/build/android/rezip/Rez...
File build/android/rezip/RezipApk.java (right):

https://codereview.chromium.org/1710083004/diff/80001/build/android/rezip/Rez...
build/android/rezip/RezipApk.java:310: if (name.endsWith(".apk") &&
!name.endsWith("MintingExample.template.apk")) {
On 2016/02/19 22:10:20, Yaron wrote:
> You don't need to do this now but I suspect that when we upstream this, we
could
> just have this warn (rather than fail) that it won't recursively rezip an
> internal apk.

That is a option to keep in mind:)

https://codereview.chromium.org/1710083004/diff/80001/web_apks/minting_exampl...
File web_apks/minting_example/src/org/chromium/minting/MintingApplication.java
(right):

https://codereview.chromium.org/1710083004/diff/80001/web_apks/minting_exampl...
web_apks/minting_example/src/org/chromium/minting/MintingApplication.java:47:
concatAndRemoveEndDuplicates(mintDirsAsList, (List<Object>) hostCurrentDirs);
On 2016/02/19 22:10:20, Yaron wrote:
> Are the duplicates a concern? Are they not just ignored? If they are ignored,
> just use Reflect.concatArrays

If simply inserting the Chrome's array before WebAPK's array, it seems ok for
now, but he order of the paths will be weird:
Chrome's dirs,
/vendor/lib,
/system/lib,
WebAPK's dirs,
/vendor/lib,
/system/lib.
It may leave potential issue here. With my changes, the resulting array is:
WebAPK's dirs,
Chrome's dirs,
/vendor/lib,
/system/lib,

https://codereview.chromium.org/1710083004/diff/80001/web_apks/minting_exampl...
web_apks/minting_example/src/org/chromium/minting/MintingApplication.java:49:
for (Object o: combined) {
On 2016/02/19 22:10:20, Yaron wrote:
> nit: the previous line and this section are just for debugging, right? Please
> remove. Same for the else block

Done.

https://codereview.chromium.org/1710083004/diff/80001/web_apks/minting_exampl...
File web_apks/minting_example/src/org/chromium/minting/Reflect.java (right):

https://codereview.chromium.org/1710083004/diff/80001/web_apks/minting_exampl...
web_apks/minting_example/src/org/chromium/minting/Reflect.java:13: final class
Reflect {
On 2016/02/19 22:10:20, Yaron wrote:
> Is this a straight copy?

Yes:)

[Yaron, 2016-02-25 05:32:50]

lgtm

https://codereview.chromium.org/1710083004/diff/80001/base/android/java/src/o...
File base/android/java/src/org/chromium/base/library_loader/LibraryLoader.java
(right):

https://codereview.chromium.org/1710083004/diff/80001/base/android/java/src/o...
base/android/java/src/org/chromium/base/library_loader/LibraryLoader.java:340:
// In minted runtime, getApplicationContext() of its host context returns null,
since
On 2016/02/24 19:09:38, Xi Han wrote:
> On 2016/02/19 22:10:19, Yaron wrote:
> > I wonder if this might bite us elsewhere but seems ok
> 
> Do you know when the splitSourceDirs are used?

It's not used. We thought we'd be using split apks but we aren't. I was
referring to the fact that getApplicationContext is null which may be a problem
in the future

https://codereview.chromium.org/1710083004/diff/80001/web_apks/minting_exampl...
File web_apks/minting_example/src/org/chromium/minting/MintingApplication.java
(right):

https://codereview.chromium.org/1710083004/diff/80001/web_apks/minting_exampl...
web_apks/minting_example/src/org/chromium/minting/MintingApplication.java:47:
concatAndRemoveEndDuplicates(mintDirsAsList, (List<Object>) hostCurrentDirs);
On 2016/02/24 19:09:38, Xi Han wrote:
> On 2016/02/19 22:10:20, Yaron wrote:
> > Are the duplicates a concern? Are they not just ignored? If they are
ignored,
> > just use Reflect.concatArrays
> 
> If simply inserting the Chrome's array before WebAPK's array, it seems ok for
> now, but he order of the paths will be weird:
> Chrome's dirs,
> /vendor/lib,
> /system/lib,
> WebAPK's dirs,
> /vendor/lib,
> /system/lib.
> It may leave potential issue here. With my changes, the resulting array is:
> WebAPK's dirs,
> Chrome's dirs,
> /vendor/lib,
> /system/lib,

Perhaps you're reight and we shouldn't have the /vendor/lib and /system/lib in
bteween

https://codereview.chromium.org/1710083004/diff/80001/web_apks/minting_exampl...
web_apks/minting_example/src/org/chromium/minting/MintingApplication.java:49:
for (Object o: combined) {
On 2016/02/24 19:09:38, Xi Han wrote:
> On 2016/02/19 22:10:20, Yaron wrote:
> > nit: the previous line and this section are just for debugging, right?
Please
> > remove. Same for the else block
> 
> Done.

You can also get rid of Reflect.getField from both

[Xi Han, 2016-02-25 14:55:38]

https://codereview.chromium.org/1710083004/diff/80001/base/android/java/src/o...
File base/android/java/src/org/chromium/base/library_loader/LibraryLoader.java
(right):

https://codereview.chromium.org/1710083004/diff/80001/base/android/java/src/o...
base/android/java/src/org/chromium/base/library_loader/LibraryLoader.java:340:
// In minted runtime, getApplicationContext() of its host context returns null,
since
On 2016/02/25 05:32:50, Yaron wrote:
> On 2016/02/24 19:09:38, Xi Han wrote:
> > On 2016/02/19 22:10:19, Yaron wrote:
> > > I wonder if this might bite us elsewhere but seems ok
> > 
> > Do you know when the splitSourceDirs are used?
> 
> It's not used. We thought we'd be using split apks but we aren't. I was
> referring to the fact that getApplicationContext is null which may be a
problem
> in the future

Got it. We need to be more careful with dealing with remote context.

https://codereview.chromium.org/1710083004/diff/80001/web_apks/minting_exampl...
File web_apks/minting_example/src/org/chromium/minting/MintingApplication.java
(right):

https://codereview.chromium.org/1710083004/diff/80001/web_apks/minting_exampl...
web_apks/minting_example/src/org/chromium/minting/MintingApplication.java:49:
for (Object o: combined) {
On 2016/02/25 05:32:50, Yaron wrote:
> On 2016/02/24 19:09:38, Xi Han wrote:
> > On 2016/02/19 22:10:20, Yaron wrote:
> > > nit: the previous line and this section are just for debugging, right?
> Please
> > > remove. Same for the else block
> > 
> > Done.
> 
> You can also get rid of Reflect.getField from both

Good catch, removed.

[Xi Han, 2016-02-26 19:59:53]

Patchset #4 (id:140001) has been deleted

[Xi Han, 2016-02-26 20:19:23]

Patchset #4 (id:160001) has been deleted