This time I just turned off the checking for Chrome/Windows because Chrome/Windows shutdown is a clusterf***.  TODO(willchan): Give a shit about Windows, get a machine that runs Windows, debug Chrome shutdown on Windows, enable this ThreadRestrictions for Chrome/Windows.

Consider PlatformThread::Join() to be blocking IO.

Marks PlatformThread::Join() as blocking IO using ThreadRestrictions.
Whitelists existing spots where we join on the UI/IO threads.

Also noteworthy is I allow blocking IO on shutdown.

BUG=65530, 66077, 66082
TEST=existing

Review URL: http://codereview.chromium.org/5750003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69079 0039d316-1c4b-4281-b951-d872f2087c98

  mac-release-10.5/intl1/times/t
  mac-release-10.5/intl2/times/t
  mac-release-10.5/startup/times/t
  mac-release-10.5/v8_benchmark/score/score
  mac-release-10.6/v8_benchmark/score/score
  vista-release-dual-core/v8_benchmark/score/score
  xp-release-dual-core/intl1/times/t
  xp-release-dual-core/intl2/times/t
  xp-release-dual-core/morejs/times/t
  xp-release-dual-core/v8_benchmark/score/score

BUG=none
TEST=perf steps are green
TBR=nsylvain@chromium.org

Review URL: http://codereview.chromium.org/5781007

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69078 0039d316-1c4b-4281-b951-d872f2087c98

I don't think eliminating _all_ objc from base is the right answer -- the remaining objc has to move out of the static libbase. But if it's easy to have fewer objc to move, that's good.

BUG=46929
TEST=One fewer duplicate symbol is reported when loading the pdf plugin.

Review URL: http://codereview.chromium.org/5716007

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69077 0039d316-1c4b-4281-b951-d872f2087c98

Rename ResourceMessageFilter to RenderMessageFilter, since that's what it actually is.  I will add a ResourceMessageFilter just for resource IPCs in a future change.
Review URL: http://codereview.chromium.org/5701004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69076 0039d316-1c4b-4281-b951-d872f2087c98

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69075 0039d316-1c4b-4281-b951-d872f2087c98

BUG=23581
TEST=compiles

Review URL: http://codereview.chromium.org/5712003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69074 0039d316-1c4b-4281-b951-d872f2087c98

Fix extension packer command-line code for Mac. Instead of blocking in a janky way by running the message loop, I added an option to run the extension packer job synchronously.

One thing to note as it pertains to the bug report is that the path must be fully qualified. ~/ for home directories doesn't work for me.

BUG=65645
TEST=none

Review URL: http://codereview.chromium.org/5703004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69073 0039d316-1c4b-4281-b951-d872f2087c98

BUG=none
TEST=safe_browsing_service_browsertest
Review URL: http://codereview.chromium.org/5544008

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69072 0039d316-1c4b-4281-b951-d872f2087c98

BUG=none
TEST=Make sure 'dry run' jobs does not get printed.

Review URL: http://codereview.chromium.org/5736004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69071 0039d316-1c4b-4281-b951-d872f2087c98

We receive multiple NavigateComplete notifications at times.

BUG=none
TEST=none

Review URL: http://codereview.chromium.org/5703005

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69070 0039d316-1c4b-4281-b951-d872f2087c98

By using a command to fetch trybot output, we can pipe the result
through a sed script that trims down the Mac output.  This makes
the output small enough so that compilation-mode can handle it.

While I'm at it, I can use file:/// URLs to unify more of the
"real" and "test" code paths.

Review URL: http://codereview.chromium.org/5737005

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69069 0039d316-1c4b-4281-b951-d872f2087c98

Bookmark buttons would get stuck with a hover state if they were obscured while the mouse was inside the button.

The problem was that the hover state only tracked the mouse entered and exited events. Unfortunately if the button is obscured because it has been removed from the window then a mouse exited event is not sent.

Fix was to also update the hover state in the -[NSCell resetCursorRect:inView:] event.

This change also improves the mouse tracking code by checking if the mouse is already in over the button.

BUG=27073
TEST=Changed tabs with the mouse over a bookmark button. Verified that the hover effect was no longer there after closing the new tab. Did a similar test by entering full screen. Verified that the hover effect was no longer there.


git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69068 0039d316-1c4b-4281-b951-d872f2087c98

The reason of the issue is ibus intercepts all key events and does not
forward them to chrome at all, while webkit relies on key events to
reset one of its internal flag. This CL just fakes key down/up events if
the input method intercepts all key events.

BUG=45478
TEST=see bug report.

Review URL: http://codereview.chromium.org/5787004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69067 0039d316-1c4b-4281-b951-d872f2087c98

...by inspecting if the module with the crash has the topmost handler
in the SEH chain.  If that's the case then the exception will very
likely be handled. This covers the case of IsBadReadPtr adn family
without any special code. It also lets us address IsValidInterface,
an internal helper in urlmon.

BUG=64348
TEST=covered by existing unit tests

Review URL: http://codereview.chromium.org/5622006

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69066 0039d316-1c4b-4281-b951-d872f2087c98

BUG=59282
TEST=none

Review URL: http://codereview.chromium.org/5810002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69065 0039d316-1c4b-4281-b951-d872f2087c98

(This used to break chrome_frame linking, but after rebasing, there doesn't seem to be any issue.)

BUG=none
TEST=compiles

Review URL: http://codereview.chromium.org/5708004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69064 0039d316-1c4b-4281-b951-d872f2087c98

BUG=64964
TEST=1. On Windows go to Bookmarks manager.
     2. Go to a folder with multiple bookmarks.
     3. Select a bookmark
     4. Hold down Shift and Ctrl. Click on another bookmark. All the bookmarks
        between the items should also be selected.

Review URL: http://codereview.chromium.org/5780002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69063 0039d316-1c4b-4281-b951-d872f2087c98

BUG=none
TEST=OTR icon's glasses are near-black, not gray

Review URL: http://codereview.chromium.org/5808002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69062 0039d316-1c4b-4281-b951-d872f2087c98

Shoulda written a test!

Review URL: http://codereview.chromium.org/5739006

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69061 0039d316-1c4b-4281-b951-d872f2087c98

when installed from store.

BUG=65346
TEST=Install any extension from the web store. Should see icon
in install success bubble.

Review URL: http://codereview.chromium.org/5690004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69060 0039d316-1c4b-4281-b951-d872f2087c98

BUG=none
TEST=none
Review URL: http://codereview.chromium.org/5721004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69058 0039d316-1c4b-4281-b951-d872f2087c98

TBR=jorlow
Review URL: http://codereview.chromium.org/5696003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69057 0039d316-1c4b-4281-b951-d872f2087c98

  mac-release-10.5/morejs/times/t
  mac-release-10.6/dhtml/times/t
  mac-release-10.6/morejs/times/t
  mac-release-10.6/moz/times/t
  vista-release-dual-core/intl1/times/t
  vista-release-dual-core/morejs/times/t
  vista-release-dual-core/moz/times/t
  xp-release-dual-core/moz/times/t
  xp-release-dual-core/moz/total_byte_b/IO_b
  xp-release-dual-core/moz/total_byte_r/IO_r
  xp-release-dual-core/moz/total_op_b/IO_op_b
  xp-release-dual-core/moz/total_op_r/IO_op_r
  xp-release-dual-core/moz/vm_peak_b/vm_pk_b
  xp-release-dual-core/moz/vm_peak_r/vm_pk_r

BUG=none
TEST=perf waterfall steps are green
TBR=nsylvain@chromium.org

Review URL: http://codereview.chromium.org/5782002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69056 0039d316-1c4b-4281-b951-d872f2087c98

We don't want to run the (expensive) Windows path fixup regex
on the (verbose) Mac output.

Review URL: http://codereview.chromium.org/5774003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69055 0039d316-1c4b-4281-b951-d872f2087c98

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69054 0039d316-1c4b-4281-b951-d872f2087c98

TBR=jorlow
Review URL: http://codereview.chromium.org/5809002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69053 0039d316-1c4b-4281-b951-d872f2087c98

Review URL: http://codereview.chromium.org/5680007

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69050 0039d316-1c4b-4281-b951-d872f2087c98

Make DOMStorageDispatcherHost be a message filter (and rename it accordingly).  It now derives from BrowserMessageFilter and hence gets message dispatching to different threads for free.
Review URL: http://codereview.chromium.org/5722003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69049 0039d316-1c4b-4281-b951-d872f2087c98

Just a bit of code cleanup

BUG=none
TEST=none

Review URL: http://codereview.chromium.org/5241001

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69048 0039d316-1c4b-4281-b951-d872f2087c98

BUG=none
TEST=unit_tests --gtest_filter=AutoFillProfileTest.Clone

Review URL: http://codereview.chromium.org/5608004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69047 0039d316-1c4b-4281-b951-d872f2087c98

Verify that previously saved username/password shows up in Regular Window, NTP and incognito window.

Review URL: http://codereview.chromium.org/5675001

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69046 0039d316-1c4b-4281-b951-d872f2087c98

BUG=none
TEST=none

Review URL: http://codereview.chromium.org/5728005

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69045 0039d316-1c4b-4281-b951-d872f2087c98

This shouldn't be needed since the hook machinery is responsible to load our DLL when it's needed, and we only need to register hooks for threads running in the IE process, but we received a dump for a user where the Broker crashed because of a call from the hook machinery to an unmapped address, and this address is within a range where we have seen the ceee_ie dll sit in other debugging session, and in the crash dump, the ceee_ie dll isn't part of the module list.

BUG=66434
TEST=None
Review URL: http://codereview.chromium.org/5778002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69044 0039d316-1c4b-4281-b951-d872f2087c98

real bounds of the omnibox. The bug occurred because we didn't
initially send down the bounds so that when the init script ran it
sent a bounds of 0x0. When the browser then processed the response
that the page supported instant it incorrectly assumed the page was
told the right bounds.

BUG=65463
TEST=see bug (specifcally comment 5). this is now covered by an
     interactive ui test as well.

Review URL: http://codereview.chromium.org/5697003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69043 0039d316-1c4b-4281-b951-d872f2087c98

BUG=none
TEST=register Chrome Frame on a fresh machine and note that bogus key HKCR\MIME\Database\s is no longer created.
Review URL: http://codereview.chromium.org/5810001

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69042 0039d316-1c4b-4281-b951-d872f2087c98

BUG=none
TEST=compiles

Review URL: http://codereview.chromium.org/5741001

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69041 0039d316-1c4b-4281-b951-d872f2087c98

Added facility to compile Mesa EGL implementation. Fixed a few issues so that it can be compiled against standard EGL 1.4 headers.
Review URL: http://codereview.chromium.org/5750002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69040 0039d316-1c4b-4281-b951-d872f2087c98

TEST=unit tests
BUG=65705

Review URL: http://codereview.chromium.org/5676003

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69039 0039d316-1c4b-4281-b951-d872f2087c98

Add compile assertions to enforce the sizes of all structs and enums in the C API.  Adjust some structs to make their sizes consistent across architectures.  Note that some structs contain pointers, so are difficult to make consistent between 32-bit and 64-bit.  Those types are in test_struct_sizes.c.  Other types have a compile assertion immediately after their definition.

This was broken off from a bigger CL:
http://codereview.chromium.org/5340003/

BUG=61004,92983
TEST=test_struct_sizes.c, compile assertions throughout

See this CL for the code that helped generate the static assertions and find affected interfaces:
http://codereview.chromium.org/5730003
Review URL: http://codereview.chromium.org/5674004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69038 0039d316-1c4b-4281-b951-d872f2087c98

BUG=none
TEST=none
Review URL: http://codereview.chromium.org/5676004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@69037 0039d316-1c4b-4281-b951-d872f2087c98