GitLab

This is safe change for L release, but not the right way. we had right
implementation in upstream and will be full tested.

BUG:17485595
Change-Id: Ib6365511bad68f5c4a5bd3d71e6a2c57987dfb81

Cherry-pick of https://codereview.chromium.org/575293003/ PS2

BindToLoop need to set |loop_| synchronously to prevent delete from
destroying the object on the wrong thread.

BUG: 17354106
Change-Id: I3168bb9c27fa135e0c98eb018610218e65455f0a

Clean cherry-pick of chromium
crrev.com/bfb71fc03179ac0d88c6b6d188915cf0dd363071

BUG: 17354106

Review URL: https://codereview.chromium.org/581703002

Change-Id: I62059284e4ab126dbb3bfde32b248e7fb12bd842
Cr-Commit-Position: refs/heads/master@{#295295}

Compile fixes for crrev.com/7a2b1cb83546418638417fc4d3c9aa6172fe02e9

BUG: 17354106

Change-Id: I3e07ab4bfd36c17ce006a73f44402251d5c83138

Cherry-pick with conflict of chromium
crrev.com/7a2b1cb83546418638417fc4d3c9aa6172fe02e9

Conflicts:
	content/browser/android/in_process/synchronous_compositor_factory_impl.h
	content/renderer/media/android/stream_texture_factory_synchronous_impl.cc
	content/renderer/media/android/webmediaplayer_android.cc

Note this by itself does not compile.

BUG: 17354106

Original description:

Add plumbing for a ResetStreamTextureProxy which should be
called on the event of a context loss. Only hooked up the
call for the in-process implementation in this CL when the
contex is restored.

ResetStreamTextureProxy currently handles "onContextLoss" as
well as "onContextRestored" concepts. It deletes the old
stream texture id, and recreates and binds a new
StreamTextureProxy.

Review URL: https://codereview.chromium.org/532993002

Change-Id: I143678de1ddf6cd66e3c49c1f1f8615d904dfcf5
Cr-Commit-Position: refs/heads/master@{#295219}

Clean cherry-pick of chromium:

https://crrev.com/947fb1c405943173d09e721efcde628dee896e36

BUG: 17369933
Original description:
Cleanup comments, unit tests and unnecessary early out.

Review URL: https://codereview.chromium.org/545663002
Change-Id: Ia3050dfe65b406b2677f9e10de43ac6665cf953d
Cr-Commit-Position: refs/heads/master@{#294206}

Clean cherry-pick of chromium
crrev.com/1a908118bc141a4fd3793438b79cd5b9422dd22b

BUG: 17406256

Review URL: https://codereview.chromium.org/556323004

Change-Id: If519344c3b95816785f10e7a0d3a3c6ab2cbb931
Cr-Commit-Position: refs/heads/master@{#294308}

Clean cherry-pick of chromium
crrev.com/f06145bfa6fb8bb4bbff86d8bd38e6e7c4a5078e

BUG: 17415765

Original description:

This is a new requirement in L during functor teardown.

Review URL: https://codereview.chromium.org/561553002

Change-Id: Ic37699f5d73bb9451458a4301cbf6bd34880c537
Cr-Commit-Position: refs/heads/master@{#294201}

This reverts commit 81be6ff2.

We don't get any reports of crash related to this. So revert this.

BUG: 17401299
BUG: 17369933

Change-Id: I051d82f0aebda4bd148109d658597e0756c9befb

Clean cherry-pick of crrev.com/693c6a04c69d69a1240e4b83e2ec7fea459acda4

BUG: 17403173

Original description:

Framework does not clamp onDraw to only the visible area. However must
still request functor in this case to support rt-side animation, and to
obtain the correct draw matrix. This means HardwareRenderer must be able
to handle not having a parent frame available.

Review URL: https://codereview.chromium.org/551023002

Change-Id: I83b2b9c8bf553a5b11f0a87af8fe5b4329a9fabc
Cr-Commit-Position: refs/heads/master@{#294093}

Change-Id: I11a8f58d20970d334a9312ff83861b0e1f122af9

This commit was generated by merge_to_master.py.

Change-Id: Iaa180d861168905354619636c88383a34aec5e62

This commit was generated by merge_to_master.py.

Change-Id: I5d2076c73bc5d4a3117e39b68250f52c554233bd

> The previous fix in r291050 fixed source compatibility but broke ABI
> compatibility with older versions of bionic instead. Since older
> versions of bionic only provide the POSIX version of strerror_r we
> should instead make sure we always use that version.
>
> BUG=
>
> Review URL: https://codereview.chromium.org/552753002
>
> Cr-Commit-Position: refs/heads/master@{#293894}

Bug: 17384482
Change-Id: I91b55784ec64145882d12119a70acde163ea97e3

The previous CL to re-generate makefiles didn't contain the files
for the two new build targets.

BUG: 16870075
Change-Id: I941c0f12aec053b632fdbfe5851cd42a7a134a8d

BUG: 16870075
Change-Id: I7d1971f2862fa27ae4469519a81029f85eab38bb

> Rename the "linker_script_copy" target to "android_exports" and move the
> link_settings clause to that target. This avoids the linker flag being
> duplicated once for every target which includes jni_generator.gypi,
> which causes problems on some linker versions.
>
> BUG=402003
>
> Review URL: https://codereview.chromium.org/473173004
>
> Cr-Commit-Position: refs/heads/master@{#289941}
> git-svn-id: svn://svn.chromium.org/chrome/trunk/src@289941 0039d316-1c4b-4281-b951-d872f2087c98

BUG: 16870075
Change-Id: Ic8a0906563edc777c3fc9a831643b8b535ede184

> This removes the eager class registration from RegisterNatives when possible.
>
> BUG=402003
> TBR=sievers@chromium.org, brettw@chromium.org, torne@chromium.org
>
> Review URL: https://codereview.chromium.org/491043002

BUG: 16870075
Change-Id: Ia2cdbddad9a12f9101290af4c06684706eac63c6

> Rather than registering all jni bindings at startup, only get references
> to the class object for those files which require bindings. All others
> are satisfied by exporting symbols which can be found automatically by
> the VM.
>
> BUG=402003
>
> Review URL: https://codereview.chromium.org/454923002

BUG: 16870075
Change-Id: I8d7617c87ac10cb9833cdb370bc6718bd58fd587

Re-generate .mk files after "Cherry-pick: Revert "Merge 281715 "[Android WebView] Terminate execution of stuck JS ..."""

As the patch was changing android_webview.gyp file, we have to update .mk files

Bug: 13509205, 17175122, 17173843
Change-Id: I0b736bb8ced4e63040b6c203cc29dbebb78aaaa0

Merge "Cherry pick   Android WebView: clean up the AwContentsClientBridge webcontents userdata." into lmp-dev

Bug: 17396873

Original description:

When we destroy the AwContentsClientBridge, clear the pointer to it
held by webcontents user data.

Also ensure that if we don't post the callback to the java side
to run the js dialog callback in the embedding app that we run
the callback.

BUG=411399
Committed: https://chromium.googlesource.com/chromium/src/+/bafb7ef3239ed1db29393ca1d528af08a2f19dfb

Change-Id: Ib9fbb4fc6b4e4dc45688db814e3ec886d469e222

The approach implemented in the patch is prone to crashing Blink bindings code.
See more comments in the bugs.

Bug: 13509205, 17175122, 17173843

Original description:

    Revert "Merge 281715 "[Android WebView] Terminate execution of stuck JS ...""

    This reverts commit 8508a2f18909b0768d273bb9161aee749f47beca.

    As described in the bug, the proposed approach is invalid, as Blink's bindings
    code isn't ready to deal with terminated scripts properly, resulting in random
    crashes. Hence reverting the patch.

    BUG=390906

    Review URL: https://codereview.chromium.org/536593004

    Cr-Commit-Position: refs/branch-heads/2062@{#587}
    Cr-Branched-From: 2e531f7c26d0d9e2aa0cced17a35eea6687dc58c-refs/heads/master@{#278856}

Change-Id: I6805a85cefef55492d5b3934b4fbd9263c1d004e

This should not happen and BrowserViewRenderer/GlobalTileManager would
be in an invalid state when it happens.
Remove this CHECK before we ship L.

BUG: 17369933

Change-Id: I288b67ac9a8bd018aed207decabb37de41889a73

BUG: 17403571
Change-Id: Ia2fab595694aa62578f3f771a322c2938662ba94

> Move the include of the Mac TargetConditionals.h file to after we've
> checked if we're building for Android. Apparently this file doesn't
> exist on all macs used to build android and the rest of android builds
> fine without it, so including it there causes the build to fail. It
> appears to only be used to test TARGET_OS_IPHONE which will never be
> true on android.
>
> BUG=
>
> Review URL: https://codereview.chromium.org/538563002
>
> Cr-Commit-Position: refs/heads/master@{#293495}

Bug: 17401514

> Android's bionic C library is intending to adopt the same semantics as
> glibc for strerror_r: define the version that returns char* if the
> source is compiled with _GNU_SOURCE instead of the POSIX version which
> returns int. Add __BIONIC__ to the condition for
> USE_HISTORICAL_STRERRO_R so that Chromium will still work.
>
> BUG=
>
> Review URL: https://codereview.chromium.org/491893002
>
> Cr-Commit-Position: refs/heads/master@{#291050}
> git-svn-id: svn://svn.chromium.org/chrome/trunk/src@291050 0039d316-1c4b-4281-b951-d872f2087c98

Bug: 17384482
Change-Id: Ie80073df1327c8e415d83c1c9321cc3ca1beaf01

When using a prebuilt libwebviewchromium, instead of skipping the entire
directory just include the makefiles needed to build V8 for
external/chromium-libpac to use.

Bug: 17228462
Change-Id: I0f543120265707d47674a85e059787a2c77611bd

Merge "Cherry pick "Cache pending JS bridge sync IPC replies, and send in case of RenderFrame deletion."" into lmp-dev

conflicts in:
  content/common/android/gin_java_bridge_errors.cc
  content/common/android/gin_java_bridge_errors.h

Bug: 16840290

Original description:
Cache pending JS bridge sync IPC replies, and send in case of
RenderFrame deletion.

When the WebView app makes a call to java over the JavaScriptBridge,
we leave the renderer hanging on a synchronous IPC. Once control
is passed into Java, it's possible that the WebView may get destroyed
(and thus the IPC channel back to the renderer closed) which means we
can't unblock the renderer waiting on the IPC response.

Instead we cache the IPC reply message and while waiting on Java to come
back to us, if we detect that the RenderFrame has been deleted, send a
reponse back before the IPC channel is closed.

BUG=408188
Committed:
https://chromium.googlesource.com/chromium/src/+/5d3e001f79c137b2ba0f5b0e9489abea616f3431

Change-Id: I7f1b28e297059eb69c5686316b47d926ea8a3d4f

These changes have landed to the system copy of libwebP in
Idf2756b8881d10001c0663bca454aac86ab30a39

and this brings the WebView version in line with it.

Bug: 17028120
Change-Id: I9e150225d7fb8e280cb1e632dc184191ed94fae8