GitLab

one you reviewed in that I changed a DCHECK in ProcessPendingTabs to
an if. See the comments there for details (this was triggering a
failure on chromeos).

Description from last attempt:

With the current code, during closing a window with unload handlers if
a tab is disconnected then we delay cleanup by way of PostTask with
the tab being disconnected. During the time between when the
disconnect happens and the task is run the unloader handler sets still
maintain a reference to the tab. If ProcessPendingTabs is invoked
during this window and the tab is deleted, then we can crash.

I'm fixing it by making disconnect remove from the sets immediately,
but delay the call to ProcessPendingTabs. I'm also doing a similar
thing in TabDetachedAt.

BUG=15620
TEST=none

Review URL: http://codereview.chromium.org/6321001

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71388 0039d316-1c4b-4281-b951-d872f2087c98

TBR=tim
BUG=62103
TEST=none

Review URL: http://codereview.chromium.org/6266004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71387 0039d316-1c4b-4281-b951-d872f2087c98

BUG=68782
TEST=manual

Review URL: http://codereview.chromium.org/6334001

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71385 0039d316-1c4b-4281-b951-d872f2087c98

Add a unique ID for each URL in the malware details protocol buffer. This allows us to save some space in the protocol buffer by not repeating the urls of parents and children. Also, change some string fields to bytes to avoid the implicit utf-8 conversion when not applicable.
BUG=60831
TEST=relevant unit_tests, browser_tests
Review URL: http://codereview.chromium.org/6208003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71384 0039d316-1c4b-4281-b951-d872f2087c98

This change fixes an issue where the password completion, when the tab key is used to cycle focus, was causing focus to get lost.

BUG=67888
TEST=PasswordAutocompleteManagerTest.*, and manual tests according to bug.

Review URL: http://codereview.chromium.org/6280003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71383 0039d316-1c4b-4281-b951-d872f2087c98

This patch changes the security model for "chrome" URLs to make implementing
some features in DOMUI easier. Instead of registering "chrome" as a NoAccess
and a Local scheme, we register it as a DisplayIsolated scheme. That should
have the effects outlined in this email:

http://groups.google.com/a/chromium.org/group/chromium-dev/browse_thread/thread/863700bf99b3f3ed

See also https://bugs.webkit.org/show_bug.cgi?id=50182 which introduces the
necessary API into WebKit.

BUG=69140

Review URL: http://codereview.chromium.org/5268006

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71382 0039d316-1c4b-4281-b951-d872f2087c98

BUG=62103
TEST=none

Review URL: http://codereview.chromium.org/6309002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71381 0039d316-1c4b-4281-b951-d872f2087c98

BUG=60822
TEST=base_file_unittest.cc

Review URL: http://codereview.chromium.org/6023006

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71379 0039d316-1c4b-4281-b951-d872f2087c98

These are new hosts, but Brocade should be issuing a fix soon
(hopefully). Therefore I'm ok adding them to the blacklist.

BUG=69379
TEST=none

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71378 0039d316-1c4b-4281-b951-d872f2087c98

BUG=none
TEST=none

Review URL: http://codereview.chromium.org/6106004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71377 0039d316-1c4b-4281-b951-d872f2087c98

Review URL: http://codereview.chromium.org/6240003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71374 0039d316-1c4b-4281-b951-d872f2087c98

the chrome frame BHO dynamically and verifies whether a subsequent navigation attempt to
a chrome frame page works correctly. 

This started failing since the change to set the user agent in the registry. We need to refresh
the IE UA by invoking the UrlMkSetSessionOption API to ensure that the subsequent switching works.

BUG=none
TEST=Covered by existing ChromeFrame InstallFlowTest.

Review URL: http://codereview.chromium.org/6251005

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71373 0039d316-1c4b-4281-b951-d872f2087c98

Google home page instant works correctly.

BUG=none
TEST=none

Review URL: http://codereview.chromium.org/6331001

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71372 0039d316-1c4b-4281-b951-d872f2087c98

menu code. It seems like we end up getting mouse events from a widget
that is no longer visible. Additionally it seems possible that the
selection is cleared before we expect it. Hopefully both of these
CHECKs will help figure out what is going on.

BUG=62872
TEST=none

Review URL: http://codereview.chromium.org/6288002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71371 0039d316-1c4b-4281-b951-d872f2087c98

BUG=None
TEST=None

Review URL: http://codereview.chromium.org/6301002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71370 0039d316-1c4b-4281-b951-d872f2087c98

Review URL: http://codereview.chromium.org/6258002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71369 0039d316-1c4b-4281-b951-d872f2087c98

BUG=none
TEST=none
Review URL: http://codereview.chromium.org/6260001

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71368 0039d316-1c4b-4281-b951-d872f2087c98

I found these changes useful while debugging issues when modifying the plugin.

BUG=none
TEST=none

Review URL: http://codereview.chromium.org/6214007

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71366 0039d316-1c4b-4281-b951-d872f2087c98

Fixed bug 64847: Graphics2D paints outside plugin boundaries. Also removed unused |origin| variables.

BUG=64847
TEST=Modify the first condition in example.cc::MyInstance::DidChangeView() to use "<=" instead of "==". On the example page, use the Toggle Size button to make the plugin larger then smaller. The plugin should not paint outside its boundary.

Review URL: http://codereview.chromium.org/6207002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71365 0039d316-1c4b-4281-b951-d872f2087c98

When a panel content window has a beforeunload handler, we'd
previously close its titlebar (which would cause the window
manager to also hide the content window in anticipation of
it being closed too), but then leave the content window
mapped and open a modal dialog (which would be kept
invisible by the window manager).  This would effectively
freeze the UI unless the user happened to press the Escape
key, which would still leave us in a state where we'd end up
restoring the panel again the next time the session was
restored.

This change makes us instead call BrowserView::CanClose()
first and avoid closing the panel if we're not supposed to.

BUG=chromium-os:10910
TEST=went to air1.com/listen and checked that i get an alert when closing the panel with "leave this page" and "stay on this page" buttons that function as expected

Review URL: http://codereview.chromium.org/6247002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71363 0039d316-1c4b-4281-b951-d872f2087c98

Useful for exercising DidChangeView() and needed (along with .cc changes) to test bug 64847.
BUG=none
TEST=none

Review URL: http://codereview.chromium.org/6156004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71362 0039d316-1c4b-4281-b951-d872f2087c98

Bug 19757 discusses how we block the UI thread on the core thread when shutting
down sync. Until that is fixed, we have to fess up to the Thread police.

BUG=69570,19757
TEST=Start syncing in debug build.  Stop sync.  Don't crash.

Review URL: http://codereview.chromium.org/6328001

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71361 0039d316-1c4b-4281-b951-d872f2087c98

This adds a "killed tab" page to be displayed in the case where a tab is killed instead of crashing.

It also adds "about:kill" to simulate kills.

On ChromeOS killed tabs will be reloaded as soon as the user clicks on them unless they are the only tab, or the currently focused tab is killed.

BUG=http://crosbug.com/10043, http://crosbug.com/10711
TEST=Built linux and ChromeOS Chrome and killed a tab, and then did about:crash and made sure that the "Aw, Snap!" page was displayed. Did about:kill and made sure kill page was displayed.

Review URL: http://codereview.chromium.org/6053012

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71360 0039d316-1c4b-4281-b951-d872f2087c98

TEST=try
BUG=none

Review URL: http://codereview.chromium.org/6189008

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71359 0039d316-1c4b-4281-b951-d872f2087c98

BUG=touchbot fails
(http://build.chromium.org/p/chromium.fyi/waterfall?branch=&builder=Chromium+Linux+Touch&builder=Chromium+Linux+Touch+%28dbg%29)
TEST=touchbot works

Review URL: http://codereview.chromium.org/6317002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71358 0039d316-1c4b-4281-b951-d872f2087c98

The sync server supports a new "progress token" scheme that allows data types to be requested with a different timestamp per datatype.  The benefits of this approach are:
 - timestamps become opaque to the client, which makes
 - the client no longer has to rely on timestamps converging for all datatypes.
 - the server will send fewer redundant updates after enabling a new datatype.
 - the flexibility of the opaque field allows the server to implement other optimizations which will increase scalability & performance

BUG=58217
TEST=manual test of sync (old profile+new profile), unit tests, integ tests

Review URL: http://codereview.chromium.org/6104003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71357 0039d316-1c4b-4281-b951-d872f2087c98

BUG=69440
TEST=none

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71354 0039d316-1c4b-4281-b951-d872f2087c98

TBR=jschuh
Review URL: http://codereview.chromium.org/6325001

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71353 0039d316-1c4b-4281-b951-d872f2087c98

I'm going to be updating the underlying class (in WebKit) and don't want these
functions to be siliently disconnected in the event that the WebKit merge
doesn't work exactly as planned.

BUG=none
TEST=compile

http://codereview.chromium.org/6320001

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71352 0039d316-1c4b-4281-b951-d872f2087c98

We can't alter the envionment once we have gone multithreaded because
glibc's getenv doesn't take the environment lock. Thus we add an
EarlyNSSInit() function which can do so, early in startup.

BUG=66835
TEST=ThreadSanitizer

http://codereview.chromium.org/6185005

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71351 0039d316-1c4b-4281-b951-d872f2087c98

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71350 0039d316-1c4b-4281-b951-d872f2087c98

(Note that this code was developed by try-server so if something appears
to be terribly wrong, it probably is.)

BUG=none
TEST=net_unittests

http://codereview.chromium.org/6180001

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71349 0039d316-1c4b-4281-b951-d872f2087c98

BUG=touchy clicks don't work
TEST=touchy clicks work

Review URL: http://codereview.chromium.org/6175006

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71348 0039d316-1c4b-4281-b951-d872f2087c98

BUG=61012
TEST=Covered by existing media tests.

Review URL: http://codereview.chromium.org/6179006

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71347 0039d316-1c4b-4281-b951-d872f2087c98

BUG=none
TEST=open cert manager, try to import non-existent file / etc

Review URL: http://codereview.chromium.org/6277001

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71346 0039d316-1c4b-4281-b951-d872f2087c98

http://codereview.chromium.org/5967003/
and is no longer flaky as of Friday, Jan 7.

http://test-results.appspot.com/dashboards/flakiness_dashboard.html#tests=NamedInterfaceTest&testType=ui_tests

BUG=66414
TEST=none

Review URL: http://codereview.chromium.org/6224004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71345 0039d316-1c4b-4281-b951-d872f2087c98

BUG=none
TEST=trybots

Review URL: http://codereview.chromium.org/6254001

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71341 0039d316-1c4b-4281-b951-d872f2087c98

BUG=none
TEST=run chrome, install app with background page, see new art on wrench menu

Review URL: http://codereview.chromium.org/6252002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71339 0039d316-1c4b-4281-b951-d872f2087c98

Now that the sync integration tests are designed to be self-contained
and not rely on external network resources like GAIA, we need to ensure
that background network access for things like auto update is disabled.

This patch adds the --disable-background-networking command line switch
to the sync integration test framework.

BUG=69469, 53931
TEST=sync_integration_tests

Review URL: http://codereview.chromium.org/6282001

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71338 0039d316-1c4b-4281-b951-d872f2087c98

We typically add all known workers when Initializing the SyncBackendHost,
but this one is causing crashes and isn't used by default.

BUG=53916,69561
TEST=none

Review URL: http://codereview.chromium.org/6111012

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71337 0039d316-1c4b-4281-b951-d872f2087c98