GitLab

This is now unused on all platforms.

R=aelias
BUG=350142

Review URL: https://codereview.chromium.org/189433003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255752 0039d316-1c4b-4281-b951-d872f2087c98

This code is stable for a while now, there's no longer any reason to
be able to quickly turn the feature off.

BUG=chromium:350298

Review URL: https://codereview.chromium.org/189163006

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255751 0039d316-1c4b-4281-b951-d872f2087c98

This CL is adding a ScreenOrientationListener abstract class with one
implementation that is based on the current mechanism used for listening
to screen orientation changes. The other implementation is based on
DisplayListener but this API is only available from API Level 17.

BUG=342714, 346696

Review URL: https://codereview.chromium.org/179783003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255750 0039d316-1c4b-4281-b951-d872f2087c98

It is not used anymore. The standard API keys workflow should be
used instead.
See http://www.chromium.org/developers/how-tos/api-keys.

BUG=350333

Review URL: https://codereview.chromium.org/189603005

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255749 0039d316-1c4b-4281-b951-d872f2087c98

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255748 0039d316-1c4b-4281-b951-d872f2087c98

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255747 0039d316-1c4b-4281-b951-d872f2087c98

TBR=arv@chromium.org
BUG=None

Review URL: https://codereview.chromium.org/189723011

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255746 0039d316-1c4b-4281-b951-d872f2087c98

The docserver is timing out and it has something to do with this. Note
that as part of the rollback I need to make sure that app.yaml and
cron.yaml still increase their versions.

BUG=350640
TBR=rockot@chromium.org

Review URL: https://codereview.chromium.org/191853002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255741 0039d316-1c4b-4281-b951-d872f2087c98

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255739 0039d316-1c4b-4281-b951-d872f2087c98

There are no ModifierObservers anymore. So remove them.

BUG=none
R=derat@chromium.org

Review URL: https://codereview.chromium.org/191523002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255736 0039d316-1c4b-4281-b951-d872f2087c98

The remaining uses of context_menu_node_ are easy to convert, but I've left them to a future cl so as to not make this cl any bigger. The main part of this cl is to move the Copy edit command from RenderViewHost to RenderFrameHost. To do that, I also had to convert Cut and Paste at the same time because of BrowserView.

BUG=304341
R=nasko@chromium.org

Review URL: https://codereview.chromium.org/183923030

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255735 0039d316-1c4b-4281-b951-d872f2087c98

BUG=349578
TEST=Manual test by following repro steps
R=fgorski@chromium.org

Review URL: https://codereview.chromium.org/189803002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255729 0039d316-1c4b-4281-b951-d872f2087c98

Currenly we have UUID in a lot of places in the EME implementation on Chromium
for Android. But UUID should really just be an implementation detail of
MediaDrmBridge. This CL moves all UUID related code to MediaDrmBridge and uses
key system elsewhere.

BUG=347596
R=boliu@chromium.org, dcheng@chromium.org, ddorwin@chromium.org, joi@chromium.org
TBR=dcheng@chromium.org

Review URL: https://codereview.chromium.org/185993004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255727 0039d316-1c4b-4281-b951-d872f2087c98

Reverting to see whether this introduced flakiness in the WebGL conformance
tests on the Win GPU bots. If it did, that means this CL exposed a
preexisting race condition. See Issue 350572.

> Remove --ui-prioritize-in-gpu-process
> 
> It is already always on on Android and Chrome OS. Consensus is that we should
> turn on on all Aura platform. The flag is a noop on other platforms.
> 
> BUG=344083
> 
> Review URL: https://codereview.chromium.org/183883023

TBR=piman@chromium.org
BUG=344083,350572

Review URL: https://codereview.chromium.org/189373012

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255722 0039d316-1c4b-4281-b951-d872f2087c98

BUG=https://code.google.com/p/nativeclient/issues/detail?id=3803
TEST=browser_tests
R=jvoung@chromium.org

Review URL: https://codereview.chromium.org/190823004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255721 0039d316-1c4b-4281-b951-d872f2087c98

Blink assumes that all scrollbar layers need repaint when scrolling.
However, SolidColorScrollbarLayer updates itself, without needing
to repaint. Ignore setNeedsDisplay for it.

BUG=333026

Review URL: https://codereview.chromium.org/189523002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255720 0039d316-1c4b-4281-b951-d872f2087c98

R=xiyuan@chromium.org
BUG=350484

Review URL: https://codereview.chromium.org/189833007

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255719 0039d316-1c4b-4281-b951-d872f2087c98

BUG=None
R=tedchoc@chromium.org

Review URL: https://codereview.chromium.org/191133003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255718 0039d316-1c4b-4281-b951-d872f2087c98

This includes many new changes. The primary reason for this change is to roll
freetype so it builds with the PNaCl Clang 3.4 toolchain.

NOTE: no tryjobs because the ports are not built by the trybots. Please see the successful SDK build @1152 here: http://build.chromium.org/p/client.nacl.ports/builders/linux-sdk/builds/456

r1152 | Fix building freetype w/ Glibc.
r1151 | openssl: Fix non-glibc builds.
r1150 | Roll freetype to version 2.5.2
r1149 | openssl: fix shared library permissions.
r1148 | Use config (Debug/Release) in sentinel. This fixes the SDK bundle script.
r1147 | openssl: build as shared library (when building with glibc)
r1146 | Force build dependencies when running naclports.py -f
r1145 | Fix libtheora build after PNaCl upgrade to clang3.4
r1144 | curl: add dependency on openssl.
r1143 | Fixes for mac OSX builds.
r1142 | Fix a few builds that were broken by running each step in a subshell.
r1141 | Run each build phase in its own subshell.
r1140 | naclports.py: fix check for NACL_SDK_ROOT.
r1139 | Add EXECUTABLES line to ffmpeg build.
r1138 | Fix mac builds of various ports.
r1137 | Fix more mac build issues.
r1136 | drod: fix mac build
r1135 | Fix mac build of serveral ports.
r1134 | Remove native_client/build from DEPS except on windows.
r1133 | tar: add missing dependency on nacl-spawn
r1132 | naclports.py: remove out/repository as part of clean --all
r1131 | Fix buildbots script so that disabled ports don't count as failures.
r1130 | Update license information for several ports.
r1129 | Add naclports.py 'build' command and use it to drive the build.
r1128 | Fix nighly builders.
r1127 | Split CPPFLAGS into sperate variable.
r1126 | Fix pnacl leveldb build.
r1125 | Go back to the right directory before building each arch.
r1124 | Update system dependency list
r1123 | Fix git newlib build
r1122 | unzip: Add missing dependency on nacl-spawn
r1121 | Bringing back a bot_common.sh as the sdk bundle needs it.
r1120 | Resurrect sdk script (deleted by mistake).
r1119 | Resharding in preparation for new bots.
r1118 | Build git with openssl
r1117 | Update ruby so that it doesn't require irt to be executable.
r1116 | Add more inforation to the PortList wiki page
r1115 | Add libssh port.
r1114 | Update and cleanup imagemagick port.
r1113 | Fix newlib build of toybox.
r1112 | Fix jpeg8b glibc build.
r1110 | Automatically create sel_ldr scrips for executables.
r1109 | Fix several package builds that were broken by recent changes
r1108 | Fix openssl build issue.
r1107 | Another attempt to fix configure rule so it works with both CMake and autotools.
r1106 | Fix yajl build.
r1105 | Don't use CMake unless configure is missing.
r1104 | Move sha1 hashes in pkg_info.
r1103 | Automatically use CMake where appropriate.
r1102 | Use _NEWLIB_VERSION macro to detect newlib.
r1101 | Update hterm DEPS to use https
r1100 | Clean config.cache from gcc build tree between builds.
r1099 | naclterm: only print status information for pid 0
r1098 | Cleanup mingw build script
r1097 | Add unzip and use it from mingn
r1096 | Two small fixes for ruby_ppapi
r1095 | Do not package everything in usr/lib for mingn
r1094 | Fix sdk bundle buildbot.
r1093 | Fix pnacl build for packages which depend on cli_main
r1092 | Fix extra file downloading in scummvm and bochs
r1091 | common.sh: Fix typo in MIRROR_URL
r1090 | Start using a sperate Google Clound Storage bucket for naclports.
r1089 | Fix typo in git's nacl.patch
r1088 | Fix zlib build on OSX
r1087 | Put bashrc in HTTP FS instead of bash.tar
r1086 | Update cairo to 1.12.14
r1085 | Add mingn package
r1084 | Link some CLI apps against libcli_main.a
r1083 | Use getwd in nacl_spawn for glibc
r1082 | Update ffmpeg to 2.1.3
r1081 | Create out directory before generating dependencies
r1080 | Do not show the same load progress twice
r1079 | Downgrade coreutils to avoid openat call
r1078 | Run executables in HTML5 filesystem by nacl-spawn
r1077 | Move dependency information into pkg_info files.
r1076 | Update pixman to 0.32.4
r1075 | Update libpng to 1.6.8
r1074 | Update SDL_image to v1.2.12
r1073 | Add zlib to the list of known licenses
r1072 | Add GNU tar-1.15
r1071 | Mount httpfs in cli_main.c
r1070 | Update google storage url [commondatastorage -> storage]
r1069 | Fix typo in common.sh:RunGitCloneStep which was causing infinite loop.
r1068 | Convert port directories to lowercase.
r1067 | Add recipe to common.sh for building using the SDK build system.
r1066 | Update ImageMagic to 6.8.8-2
r1065 | Update zlib to 1.2.8.
r1064 | Minor cleanup of naclports build system.
r1063 | Adding port of HDF5 to naclports.
r1062 | Do not specify names of nexes explicitly in coreutils
r1061 | Add gcc
r1060 | Add binutils
r1059 | Parameterize the ar command in jpeg6b port
r1058 | Add coreutils
r1057 | Send pid properly in nacl_waitpid of nacl-spawn
r1056 | Fix toybox build for getcwd, nacl_spawn and configure.
r1055 | Add pnacl + newlib support to apr and apr-util.
r1054 | Disable all ASM code for all archs in zeromq.
r1053 | Change toybox HOSTCC settings to cc, which should be safe everywhere.
r1052 | Add top level build script for building packages.
r1051 | Do not build Ada code in ncurses
r1050 | Hooking git back into the build, adding newlib+pnacl, use nacl-spawn.
r1049 | Enable newlib for nacl-spawn
r1048 | Initial port of Toybox and toysh.
r1047 | Enabling nano under newlib.
r1046 | Add nacl-spawn
r1045 | Adding netcat 0.7.1.
r1044 | Remove bashism from /bin/sh script.
r1043 | Use naclports expat in apr-util to make build stable.
r1042 | Revert "Remove SKIP_SEL_LDR_TESTS from common.sh."
r1041 | Remove SKIP_SEL_LDR_TESTS from common.sh.
r1040 | Add leveldb port
r1039 | Adding subversion and its dependencies apr and apr-util.
r1038 | Adding busybox 1.22.0.
r1037 | Add clapack to naclports
r1036 | Adding gnumake 3.82.
r1035 | Remove stubs glibc-compat that are now part of libnacl.
r1034 | Add libpcre to naclports
r1033 | Add gmp, mpfr, and mpc
r1032 | Fix libsodium build by removing direct reference to libstdc++.
r1031 | Remove nacl-mounts
r1030 | Pull in newer gdb version and fix step and continue.
r1029 | Fix libsodium build.
r1028 | Add glibc-comapt dependency to zeromq.
r1027 | Cleanup tommath/tomcrypt build.
r1026 | Adding ZeroMQ 4.0.3 port.
r1025 | Fix libsodium build on the bots by removing realpath dependency.
r1024 | Fix linking of bochs under newlib.
r1023 | Fix scummvm after r1021.
r1022 | Add libsodium to naclports
r1021 | Convert SDL and SDLmain to use ppapi instead of ppapi_cpp.
r1020 | Fix linking issues when building SDL_image/net/mixer.
r1019 | Enabling newlib for vim.
r1018 | Fix glibc-compat linkage issue
r1017 | Fix bochs build
r1016 | Add third_party/ to .gitignore
r1015 | Convert bochs to use SDLmain and remove use of nacl-mounts.
r1014 | Use SDLmain in dosbox.
r1013 | Fix nacl-mounts compiler warnings.
r1012 | Fix sqlite PNaCl build.
r1011 | Fix openal-soft to use the new PPAPI_Audio_Callback structure.
r1010 | Add out and DEPS locations to svn:ignore
r1009 | OpenCV in Nacl, sort out some minor issues
r1008 | Reinstate libtiff pnacl fix until pepper_32 goes stable.
r1007 | naclports: only use libc++ as the default C++ standard library on PNaCl for SDk version 33 and later
r1006 | Update and improve boost port.
r1005 | Add common build rules for CMake builds.
r1004 | Turn on naclmounts tests for pnacl -- seems fixed by r999
r1003 | Build and install civetweb as library
r1002 | Unbreak naclports build.
r1000 | Prepare naclports for the PNaCl change to libc++ as the default.
r999  | Fix race crash in nacl-mounts tests.
r998  | Add python3 port
r997  | Remove use of libnosys.
r996  | snes9x: remove dependency on nacl-mounts
r995  | Fix opencv build by setting NACL_TOOLCHAIN_ROOT on the cmake commandline.
r994  | OpenCV 2.4.7 porting to PNaCl - missing executable flag in build.sh
r993  | OpenCV 2.4.7 porting to PNaCl
r992  | Add glibc-compat stubs for sigaction and sigsuspend.
r991  | Update terminal-based webstore apps.
r990  | TeX Live 2013 build in examples
r988  | Move all ports from libraries/* or examples/*/* to ports/*.
r987  | Add script to generate wiki page listing all ports.
r977  | lua_ppapi: ignore failure to mount html5fs.
r976  | Helper script for generating terminal apps
r975  | Update lua example to include benchmarks.
r974  | Add lua tests to lua_ppapi example.
r973  | Add package for jpeg6b.
r972  | Some fixes needed to build bullet on Mac.
r971  | Add openssh port.
r970  | Roll hterm.
r969  | Factor out naclterm installation into a shared function.
r968  | Revert newlib build of vim.
r967  | Enable newlib build of vim.
r966  | Fix bullet test under glibc.
r965  | Fix ruby_ppapi so it runs 'irb' rather than just ruby.
r964  | Fix readline: Don't change directory in test phase.
r963  | Add a common TestStep to naclports build system.
r962  | Upgrade fftw3 from 3.2.2 -> 3.3.3.
r961  | Don't publish nethack sources by default.
r960  | Don't publish binaries from the periodic builders.
r959  | Fixes for upload to commondatastorage.
r958  | Add presubmit check for archives missing from mirror.
r957  | Fix openssl "make install".  Actually popd...
r956  | snes9x port now uses nacl_io and ppapi_simple.
r955  | Enable threads for fftw.

BUG=none
R=sbc@chromium.org

Review URL: https://codereview.chromium.org/189623019

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255717 0039d316-1c4b-4281-b951-d872f2087c98

The js files ended up in mojo/public/bindings/js
The C++ code in mojo/bindings/js

Adds a new target with the c++, mojo_js_bindings_lib

I'm leaving the unit tests where they are for now, which is not
ideal. Will move them in a followup patch.

BUG=none
TEST=none
R=vtl@chromium.org
TBR=jochen@chromium.org

Review URL: https://codereview.chromium.org/189453003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255716 0039d316-1c4b-4281-b951-d872f2087c98

> Telemetry support for iframes.
> 
> * Add methods EvaluateJavaScriptInContext,
> ExecuteJavaScriptInContext.
> * Backend classes take an additional context_id argument, which refers to the iframe.
> * WebContents has an EnableAllContexts method, to enable access to iframes.
> * Add tests testIFrame, and unit test data files host.html, iframe*.html.
> 
> BUG=237032
> TEST=unittest
> 
> Review URL: https://codereview.chromium.org/183863007

BUG=350568
TBR=achuith@chromium.org

Review URL: https://codereview.chromium.org/191643002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255715 0039d316-1c4b-4281-b951-d872f2087c98

Interesting enough this was happening because of a focus bug. Turns
out a FocusController associated with one DesktopNativeWidgetAura
could attempt to focus an aura::Window in a different
DesktopNativeWidgetAura. When this happened it would move the
content_window on top of the status bar.

We should only activate windows within the same root.

BUG=345698
TEST=covered by unit test, see bug as well.
R=ben@chromium.org

Review URL: https://codereview.chromium.org/186733011

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255714 0039d316-1c4b-4281-b951-d872f2087c98

BUG=276131

Review URL: https://codereview.chromium.org/189543003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255713 0039d316-1c4b-4281-b951-d872f2087c98

R=noelallen@chromium.org

Review URL: https://codereview.chromium.org/131853012

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255712 0039d316-1c4b-4281-b951-d872f2087c98

BUG=349200,350129
TBR=xiyuan@chromium.org

Review URL: https://codereview.chromium.org/191183002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255711 0039d316-1c4b-4281-b951-d872f2087c98

BUG=338759
NOTRY=true

Review URL: https://codereview.chromium.org/184783005

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255710 0039d316-1c4b-4281-b951-d872f2087c98

Add support for providing incremental updates for multipage
distillation. Updates are now broadcasted to the viewers, whenever a new
page in the article finishes distillation. This allows viewer to show
data to the user, before the distillation of the entire article is
completed.
This change required changing management of pages in distiller to be
refcounted.

BUG=288015
NOTRY=true
R=cjhopman@chromium.org

Review URL: https://codereview.chromium.org/178303004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255708 0039d316-1c4b-4281-b951-d872f2087c98

Prevent one risk of a /proc descriptor leaking by using O_CLOEXEC.

BUG=350145
R=jorgelo@chromium.org

Review URL: https://codereview.chromium.org/189113009

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255707 0039d316-1c4b-4281-b951-d872f2087c98

The JavaScript API for getting the set of devices known to the adapter
and dealing with discovery was difficult to use, and had a messy
implementation. By streamlining the API not only does it become much
easier to work with, but the implementation becomes much cleaner as
well.

chrome.bluetooth.getDevices() now simply returns an array of devices in
its callback, it's up to the application to filter these since profile
filtering is unreliable during discovery and on Low Energy.

chrome.bluetooth.startDiscovery() now only has a callback to indicate
a success to the command, the hidden event and listener has been
removed in favor of the new device events.

Add chrome.bluetooth.onDeviceAdded, chrome.bluetooth.onDeviceChanged
and chrome.bluetooth.onDeviceRemoved events. These return updated
device objects when they are first known, changed and removed
respectively.

Take the opportunity of writing a proper documentation page for these
new methods and events showing their usage.

BUG=345050
TEST=BluetoothApiTest updated and included
R=armansito@chromium.org, miket@chromium.org, rpaquay@chromium.org, sky@chromium.org

Review URL: https://codereview.chromium.org/177113013

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255706 0039d316-1c4b-4281-b951-d872f2087c98

BUG=None
NOTRY=true

Review URL: https://codereview.chromium.org/185563010

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255705 0039d316-1c4b-4281-b951-d872f2087c98

When setting a texture mailbox on TextureLayer, reusing the same
TextureMailbox is very fishy since the gpu::Mailbox would end up
being released back to the client multiple times.

However since pepper does magical bad things by using a single
texture that somehow is double buffered (this is wrong and
to be fixed), it needs to give the texture to cc multiple times.
It is safer in this case since its ReleaseCallback is empty. We
can encode this with a SetTextureMailboxWithoutReleaseCallback
method on TextureLayer that pepper can use, and it avoids this
DCHECK.

The method comes with a big warning. Pepper should be fixed to
avoid visual tearing, and this method should go away.

R=piman
BUG=350204

Review URL: https://codereview.chromium.org/189333003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255704 0039d316-1c4b-4281-b951-d872f2087c98

(Crashy on today's canary)
> Make chrome_elf use thunks instead of function pointers.
> 
> 1. Add functionality to ServiceResolverThunk to copy a thunk without patching.
> 2. Move chrome_elf thunk-handling code to a common location.
> 3. Use a thunk instead of a f'n ptr for redirects.
> 
> BUG=334379
> 
> Review URL: https://codereview.chromium.org/183833004

TBR=caitkp@chromium.org

Review URL: https://codereview.chromium.org/189803007

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255703 0039d316-1c4b-4281-b951-d872f2087c98

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255702 0039d316-1c4b-4281-b951-d872f2087c98

Fails to compile on "Android Webview AOSP Builder"

http://build.chromium.org/p/chromium.linux/builders/Android%20Webview%20AOSP%20Builder/builds/9360

> Blink roll 168699:168744
> 
> http://build.chromium.org/f/chromium/perf/dashboard/ui/changelog_blink.html?url=/trunk&range=168700:168744&mode=html
> TBR=
> BUG=
> 
> Review URL: https://codereview.chromium.org/190633004

TBR=arv@chromium.org

Review URL: https://codereview.chromium.org/189373011

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255701 0039d316-1c4b-4281-b951-d872f2087c98

If a seed is received that fails verification, don't store it.
Similarly, on startup discard the seed if it fails verification.

Refactors some of the code to make tests pass (since they're
not testing signed data). Additionally, deprecates the previous
seed hashing code.

BUG=336536

Review URL: https://codereview.chromium.org/183003008

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255697 0039d316-1c4b-4281-b951-d872f2087c98

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255695 0039d316-1c4b-4281-b951-d872f2087c98

http://build.chromium.org/f/chromium/perf/dashboard/ui/changelog_blink.html?url=/trunk&range=168700:168744&mode=html
TBR=
BUG=

Review URL: https://codereview.chromium.org/190633004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255694 0039d316-1c4b-4281-b951-d872f2087c98

Quickoffice will not be included in webview. But there are
some issues on the android license check bot.

BUG=350472
NOTRY=true
NOTREECHECKS=true
R=dpolukhin@chromium.org

Review URL: https://codereview.chromium.org/191323002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255693 0039d316-1c4b-4281-b951-d872f2087c98

OWNER reviewer: arv

BUG=341256

Review URL: https://codereview.chromium.org/176823009

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255692 0039d316-1c4b-4281-b951-d872f2087c98

This is a follow up of r254755 that updated the pointer to the chunk with the
highest address in the region when the highest chunk in the region was being
split (during an allocation reusing a free chunk).

While this was enough to fix the bug specified below, it was only partly
addressing the problem since this pointer also needs to be updated when free
chunks are merged. This CL fixes this issue and adds an extra DCHECK() exposing
the problem with the existing unit tests.

BUG=347919

Review URL: https://codereview.chromium.org/183763037

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@255691 0039d316-1c4b-4281-b951-d872f2087c98