GitLab

Caused a heap-use-after-free in GLES2DecoderManualInitTest.AsyncPixelTransfers
on Linux ASan+LSan Tests (1):
http://build.chromium.org/p/chromium.memory/builders/Linux%20ASan%2BLSan%20Tests%20%281%29/builds/863


> By keeping track of transfer buffer usage (both sync and async), it is
> possible to reuse an existing transfer buffer earlier than it is today.
> 
> For synchronous uploads, this is done by inserting tokens marking a
> buffer's last usage. If such a token has passed, the buffer can be
> reused.
> 
> For asynchronous uploads, this is done by adding an internal async
> upload token to the GLES2Implementation and GLES2CmdDecoderImpl that
> enumerates all asynchronous uploads. When an upload is completed, the
> token is synchronized with the client which then can free the memory
> associated with the async token.
> 
> The fenced allocator and mapped memory manager gets a callback that is
> used to allow fenced allocator to make the user, in this case
> GLES2Implementation, poll the current async upload token and free any
> associated memory.
> 
> BUG=328808
> 
> Review URL: https://codereview.chromium.org/116863003

TBR=jadahl@opera.com

Review URL: https://codereview.chromium.org/216673006

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260193 0039d316-1c4b-4281-b951-d872f2087c98

KioskTest.LaunchAppWithNetworkConfigAccelerator

TBR=zelidrag@chromium.org
NOTRY=true
NOTREECHECKS=true
BUG=350056

Review URL: https://codereview.chromium.org/217053005

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260192 0039d316-1c4b-4281-b951-d872f2087c98

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260191 0039d316-1c4b-4281-b951-d872f2087c98

This pulls in the following changes:

r111 Implementation of the woff2 format changes proposed by the W3C working group.
r112 Remove unused kShortFlagsContinue.

TBR=yusukes@chromium.org
BUG=

Review URL: https://codereview.chromium.org/216203002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260190 0039d316-1c4b-4281-b951-d872f2087c98

The cursor is invisible unless set explicitly, so set a bitmap from
CursorFactoryEvdevDri upon construction.

Configuration of the cursor plane must be delayed until we've configured
the hardware, so store the bitmap on DriSurfaceFactory and set it in
RealizeAcceleratedWidget.

TEST=content_shell --ozone-platform=dri
BUG=252315

Review URL: https://codereview.chromium.org/208553004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260189 0039d316-1c4b-4281-b951-d872f2087c98

BUG=357024

Review URL: https://codereview.chromium.org/217023002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260188 0039d316-1c4b-4281-b951-d872f2087c98

The ordering has recently changed and some alternate shelf layout got not corrected.

BUG=346678
TEST=visual testing
R=flackr@chromium.org

Review URL: https://codereview.chromium.org/215693004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260187 0039d316-1c4b-4281-b951-d872f2087c98

This code bit was missed when re-factoring in: https://codereview.chromium.org/196403008/

TBR=piman@chromium.org

Review URL: https://codereview.chromium.org/216553007

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260186 0039d316-1c4b-4281-b951-d872f2087c98

BUG=

Review URL: https://codereview.chromium.org/212623012

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260185 0039d316-1c4b-4281-b951-d872f2087c98

This change does not fix the bug, just a preparation for the actual fix.
No behavior change is intended.

Return cancel closure from DownloadOperation and FileSystem.
Change user code (SyncClient and DriveFileStreamReader) to use the returned closure.

BUG=344789
TEST=unit_tests

Review URL: https://codereview.chromium.org/214363002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260184 0039d316-1c4b-4281-b951-d872f2087c98

should_content_update may be false for new files if it's opend.

BUG=353680
TEST=unit_tests

Review URL: https://codereview.chromium.org/216333003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260183 0039d316-1c4b-4281-b951-d872f2087c98

To implement FakeDriveService without GData WAPI classes.

BUG=334544
TEST=unit_tests

Review URL: https://codereview.chromium.org/216433002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260182 0039d316-1c4b-4281-b951-d872f2087c98

sky is about to check something in that will interact poorly with this
and it'll be easier to apply the necessary changes after that.

> Fixup mojo dylibs to support mojo in content
> 
> Second try after revert of
> https://codereview.chromium.org/211503005/
> 
> The isolates failed because libmojo_system.dylib wasn't getting copied.
> 
> BUG=
> TBR=mark@chromium.org
> 
> Review URL: https://codereview.chromium.org/217003003

TBR=davemoore@chromium.org

Review URL: https://codereview.chromium.org/216793008

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260181 0039d316-1c4b-4281-b951-d872f2087c98

Adds expectations for video resolution output on all tests that use peerconnection-call.html call.

BUG= 350538

Review URL: https://codereview.chromium.org/214663004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260179 0039d316-1c4b-4281-b951-d872f2087c98

BUG=357190

Review URL: https://codereview.chromium.org/212173006

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260178 0039d316-1c4b-4281-b951-d872f2087c98

possible to reuse an existing transfer buffer earlier than it is today.

For synchronous uploads, this is done by inserting tokens marking a
buffer's last usage. If such a token has passed, the buffer can be
reused.

For asynchronous uploads, this is done by adding an internal async
upload token to the GLES2Implementation and GLES2CmdDecoderImpl that
enumerates all asynchronous uploads. When an upload is completed, the
token is synchronized with the client which then can free the memory
associated with the async token.

The fenced allocator and mapped memory manager gets a callback that is
used to allow fenced allocator to make the user, in this case
GLES2Implementation, poll the current async upload token and free any
associated memory.

BUG=328808

Review URL: https://codereview.chromium.org/116863003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260177 0039d316-1c4b-4281-b951-d872f2087c98

This is bringing Aura in line with the existing GTK behaviour. Windows
with a coloured frame still have the Windows 8 look.

App windows on Linux now always use AppWindowFrameView instead of the
blue CustomFrameView.

BUG=322256

Review URL: https://codereview.chromium.org/210363004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260176 0039d316-1c4b-4281-b951-d872f2087c98

Second try after revert of
https://codereview.chromium.org/211503005/

The isolates failed because libmojo_system.dylib wasn't getting copied.

BUG=
TBR=mark@chromium.org

Review URL: https://codereview.chromium.org/217003003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260175 0039d316-1c4b-4281-b951-d872f2087c98

BUG=353067
TBR=aboxhall@chromium.org

Review URL: https://codereview.chromium.org/217213002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260174 0039d316-1c4b-4281-b951-d872f2087c98

This change makes it so hitting the Esc while the omnibox has focus shows the URL unless search term replacement is being performed, in which case we revert to the original search terms.

This holds for both the hide-on-mouse-release and hide-on-input options.  For the latter, this means that a click in the omnibox will also hide the origin chip.  This supports the case where the user is trying to get at the URL and forgets to click on the origin chip or misses it due to existing muscle memory.  In this case, an Esc will show the URL as users are accustomed to.

In no case will hitting Esc cause the origin chip to reappear.

BUG=349497

Review URL: https://codereview.chromium.org/214613003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260173 0039d316-1c4b-4281-b951-d872f2087c98

https://codereview.chromium.org/213243002/ had an error where we log the
new metrics to the wrong bucket. This should fix that.

BUG=352266, 357162
TBR=tommi@chromium.org
NOTRY=True (ios_debug broken and tripping up CQ)

Review URL: https://codereview.chromium.org/216683002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260172 0039d316-1c4b-4281-b951-d872f2087c98

Separate pre-classification checks for client-side malware and phishing detection and enable the client-side malware
feature in all channels (not just dev and canary).

BUG=352782

Review URL: https://codereview.chromium.org/173133004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260171 0039d316-1c4b-4281-b951-d872f2087c98

NOTRY=true

BUG=342685

Review URL: https://codereview.chromium.org/213703004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260168 0039d316-1c4b-4281-b951-d872f2087c98

Needs to outlive the file thread. Fixes rare crash.

Previous attempt (https://codereview.chromium.org/174493003) caused an inconsistent leak (https://code.google.com/p/chromium/issues/detail?id=348928).

This CL makes sure pending URLFetchers are deleted before threads teardown and removes the reference to system request context in WebRtcLoggingHandlerHost.

NOTRY=true

BUG=344892

Review URL: https://codereview.chromium.org/211033006

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260167 0039d316-1c4b-4281-b951-d872f2087c98

SyncEngineContext, and they are accessed through getters.

-scoped_ptr<drive::DriveServiceInterface> drive_service_;
-scoped_ptr<drive::DriveUploaderInterface> drive_uploader_;
-scoped_ptr<MetadataDatabase> metadata_database_;
-RemoteChangeProcessor* remote_change_processor_;
-scoped_refptr<base::SequencedTaskRunner> task_runner_;



BUG=347425
NOTRY=true

Review URL: https://codereview.chromium.org/213473008

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260166 0039d316-1c4b-4281-b951-d872f2087c98

WiFiService StartConnect after CreateNetwork will try TKIP encryption if AES encryption fails to connect to newly created network.

BUG=328959

Review URL: https://codereview.chromium.org/197873012

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260165 0039d316-1c4b-4281-b951-d872f2087c98

BUG=354160
R=jonross@chromium.org, oshima@chromium.org

Review URL: https://codereview.chromium.org/217033002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260164 0039d316-1c4b-4281-b951-d872f2087c98

These APIs are //chrome-specific, and rely on no other state from
SigninManager. As these APIs are per-Profile, ChromeSigninClient is an
appropriate place for them to live.

BUG=334205
TBR=thakis

Review URL: https://codereview.chromium.org/216703002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260163 0039d316-1c4b-4281-b951-d872f2087c98

When doing recursive hit testing and the point falls within two children
of the same element, prefer the one that has an indirect descendant
(i.e. grandchild) match.

A better fix would be to send an IPC to the renderer and let Blink do
the hit testing, I filed another bug (357371) for that. This is a stopgap.

Note, though, that we still need this hit testing algorithm for
synchronous APIs (like on Windows), so it's worth improving this.

BUG=357247
R=aboxhall@chromium.org

Review URL: https://codereview.chromium.org/215633003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260161 0039d316-1c4b-4281-b951-d872f2087c98

BUG=355860

Review URL: https://codereview.chromium.org/214593006

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260160 0039d316-1c4b-4281-b951-d872f2087c98

Test that a golden version 7 file can be read and updated, in preparation for
version 8.

BUG=351448
R=tsepez@chromium.org

Review URL: https://codereview.chromium.org/211363005

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260159 0039d316-1c4b-4281-b951-d872f2087c98

Added a shortcut (Ctrl+Alt+N) for showing network config screen in during kiosk launch even if network connectivity is present. 

BUG=350065
TEST=added KioskTest.LaunchAppWithNetworkConfigAccelerator

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=259679

Review URL: https://codereview.chromium.org/209143002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260158 0039d316-1c4b-4281-b951-d872f2087c98

The code responsible for detecting similar blocks and merging
them didn't check for the next blocks if the last instruction was
not a JMP or a RET.

The patch to fix this bug (in codegen.cc) is based on a patch by
jld@panix.com, attached to the bug report.

Additional unittests are from jln@chromium.org

BUG=351103

Review URL: https://codereview.chromium.org/215173002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260157 0039d316-1c4b-4281-b951-d872f2087c98

BUG=355797
R=xiyuan@chromium.org

Review URL: https://codereview.chromium.org/203863005

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260156 0039d316-1c4b-4281-b951-d872f2087c98

This leaves the state in RenderViewHost for now, to make this an easier
first step.

BUG=304341
TEST=No behavior change.
R=nasko@chromium.org

Review URL: https://codereview.chromium.org/208243019

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260155 0039d316-1c4b-4281-b951-d872f2087c98

Same deal as issue 172853009:

Auth errors currently show up in the wrench menu as GlobalErrors. This CL
removes auth errors from the wrench menu in Chrome OS and in Windows when
using the Ash shell, showing notifications instead.

The behavior when clicking these notifications should be the same as
when clicking the toolbar menu error -- showing the sync setup dialog in
the Settings page.

The error-checking logic has been refactored into the new SyncErrorController,
which now exposes the error status to the UI classes.

The UI classes, SyncGlobalError and SyncErrorNotifier, are keyed services
instantiated by the UI.

BUG=128948
TEST=unit_tests --gtest_filter=S*n*Error*
R=stevenjb@chromium.org, atwilson@chromium.org, derat@chromium.org, ben@chromium.org

Review URL: https://codereview.chromium.org/207603002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260154 0039d316-1c4b-4281-b951-d872f2087c98

BUG=354854

Review URL: https://codereview.chromium.org/208013003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260153 0039d316-1c4b-4281-b951-d872f2087c98

See discussion at chromium issue 338885.

When USE_OPENSSL_CERTS is defined, X509::OSCertHandle is now
typedef'ed to struct X509*. 
 
When USE_OPENSSL is defined, USE_OPENSSL_CERTS will now be
defined for linux and Android, while being off for Mac and 
Windows. This allows OpenSSL to be used while leaving
certificate handling to the OS.
 
OpenSSL cert verifying code will only be used on Linux.

This patch does not change any default behavior.

Bug=none
Test=none

Review URL: https://codereview.chromium.org/206453002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260152 0039d316-1c4b-4281-b951-d872f2087c98

This uses the invoker() syntax.

R=scottmg@chromium.org, scottmg

Review URL: https://codereview.chromium.org/212843007

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260151 0039d316-1c4b-4281-b951-d872f2087c98

BUG=349158
R=mnissler@chromium.org

Review URL: https://codereview.chromium.org/210063003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@260150 0039d316-1c4b-4281-b951-d872f2087c98