Reland Compile under-invalidation checking in all builds

Reland Compile under-invalidation checking in all builds

This relands https://codereview.chromium.org/2299223002 which was
reverted because of random pixel under-invalidations.

Now clear the bitmaps before drawing the pictures.

BTW changed the error indication color from solid red to translucent
magenta to distinguish from the normal red pixels in test results.

> This will let the tests having under-invalidation issues produce the
> same result on all builds, so enable us to rebaseline the tests with
> failure results. In this way we can track either regressions and
> progressions instead of ignoring the tests.
>
> BUG=619103
> CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
>
> Review-Url: https://codereview.chromium.org/2299223002
> Cr-Commit-Position: refs/heads/master@{#416060}

BUG=619103
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
R=pdr@chromium.org

[Xianzhu, 2016-09-01 23:35:11]

Description was changed from

==========
Reland Compile under-invalidation checking in all builds

This relands https://codereview.chromium.org/2299223002 which was
reverted because of random pixel under-invalidations.

Now clear the bitmaps before drawing the pictures.

BTW changed the error indication color from solid red to translucent
magenta to distinguish from the normal red pixels in test results.

> This will let the tests having under-invalidation issues produce the
> same result on all builds, so enable us to rebaseline the tests with
> failure results. In this way we can track either regressions and
> progressions instead of ignoring the tests.
>
> BUG=619103
>
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
>
> Review-Url: https://codereview.chromium.org/2299223002
> Cr-Commit-Position: refs/heads/master@{#416060}

BUG=619103
==========

to

==========
Reland Compile under-invalidation checking in all builds

This relands https://codereview.chromium.org/2299223002 which was
reverted because of random pixel under-invalidations.

Now clear the bitmaps before drawing the pictures.

BTW changed the error indication color from solid red to translucent
magenta to distinguish from the normal red pixels in test results.

> This will let the tests having under-invalidation issues produce the
> same result on all builds, so enable us to rebaseline the tests with
> failure results. In this way we can track either regressions and
> progressions instead of ignoring the tests.
>
> BUG=619103
>
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
>
> Review-Url: https://codereview.chromium.org/2299223002
> Cr-Commit-Position: refs/heads/master@{#416060}

BUG=619103
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
==========

[Xianzhu, 2016-09-01 23:41:36]

The CQ bit was checked by wangxianzhu@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-09-01 23:42:10]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[Xianzhu, 2016-09-01 23:47:45]

wangxianzhu@chromium.org changed reviewers:
+ pdr@chromium.org

[Xianzhu, 2016-09-01 23:47:46]

Please use this link for the differences from the original patch:
https://codereview.chromium.org/2301303002/diff2/1:20001/third_party/WebKit/S...

[commit-bot: I haz the power, 2016-09-01 23:51:38]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-09-01 23:51:38]

Dry run: Try jobs failed on following builders:
  linux_layout_tests_slimming_paint_v2 on master.tryserver.chromium.linux
(JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_layout_te...)

[pdr., 2016-09-02 00:25:33]

Thanks for the quick fix.

The failures that caused the revert[1] look like random garbage leftover in the
canvas. Were we using uninitialized memory? Do you know why this only started
showing up once we switched to compiling in release?

[1]
https://storage.googleapis.com/chromium-layout-test-archives/WebKit_Win10/142...

[Xianzhu, 2016-09-02 00:49:13]

On 2016/09/02 00:25:33, pdr. wrote:
> Thanks for the quick fix.
> 
> The failures that caused the revert[1] look like random garbage leftover in
the
> canvas. Were we using uninitialized memory? Do you know why this only started
> showing up once we switched to compiling in release?
> 
> [1]
>
https://storage.googleapis.com/chromium-layout-test-archives/WebKit_Win10/142...

I guess skia or the memory allocator clears newly bitmap buffer in debug mode,
but not in release mode, so I added canvas.clear(Sk_ColorTRANSPARENT).

[pdr., 2016-09-02 01:08:17]

On 2016/09/02 at 00:49:13, wangxianzhu wrote:
> On 2016/09/02 00:25:33, pdr. wrote:
> > Thanks for the quick fix.
> > 
> > The failures that caused the revert[1] look like random garbage leftover in
the
> > canvas. Were we using uninitialized memory? Do you know why this only
started
> > showing up once we switched to compiling in release?
> > 
> > [1]
> >
https://storage.googleapis.com/chromium-layout-test-archives/WebKit_Win10/142...
> 
> I guess skia or the memory allocator clears newly bitmap buffer in debug mode,
but not in release mode, so I added canvas.clear(Sk_ColorTRANSPARENT).

That makes sense. I wonder if we should clear with all red in debug to catch
unitialized bitmap issues? Anyway... for another patch.

LGTM

[Xianzhu, 2016-09-02 03:35:50]

The CQ bit was checked by wangxianzhu@chromium.org

[commit-bot: I haz the power, 2016-09-02 03:36:02]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-09-02 03:42:39]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-09-02 03:42:40]

Try jobs failed on following builders:
  linux_layout_tests_slimming_paint_v2 on master.tryserver.chromium.linux
(JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_layout_te...)

[Xianzhu, 2016-09-02 04:57:04]

The CQ bit was checked by wangxianzhu@chromium.org

[Xianzhu, 2016-09-02 04:57:05]

The patchset sent to the CQ was uploaded after l-g-t-m from pdr@chromium.org
Link to the patchset: https://codereview.chromium.org/2301303002/#ps40001
(title: "Rebase")

[commit-bot: I haz the power, 2016-09-02 04:57:19]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-09-02 05:06:16]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-09-02 05:06:18]

Try jobs failed on following builders:
  linux_layout_tests_slimming_paint_v2 on master.tryserver.chromium.linux
(JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_layout_te...)

[Xianzhu, 2016-09-02 05:12:30]

On 2016/09/02 01:08:17, pdr. wrote:
> On 2016/09/02 at 00:49:13, wangxianzhu wrote:
> > On 2016/09/02 00:25:33, pdr. wrote:
> > > Thanks for the quick fix.
> > > 
> > > The failures that caused the revert[1] look like random garbage leftover
in
> the
> > > canvas. Were we using uninitialized memory? Do you know why this only
> started
> > > showing up once we switched to compiling in release?
> > > 
> > > [1]
> > >
>
https://storage.googleapis.com/chromium-layout-test-archives/WebKit_Win10/142...
> > 
> > I guess skia or the memory allocator clears newly bitmap buffer in debug
mode,
> but not in release mode, so I added canvas.clear(Sk_ColorTRANSPARENT).
> 
> That makes sense. I wonder if we should clear with all red in debug to catch
> unitialized bitmap issues? Anyway... for another patch.

We are doing this in cc
(https://cs.chromium.org/chromium/src/cc/playback/raster_source.cc?sq=package:...)
for opaque layers. This is why we removed the duplicate feature in GraphicsLayer
(DebugRedFill).

> 
> LGTM

[Xianzhu, 2016-09-02 15:44:12]

The CQ bit was checked by wangxianzhu@chromium.org

[commit-bot: I haz the power, 2016-09-02 15:44:36]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-09-02 15:50:22]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-09-02 15:50:23]

Try jobs failed on following builders:
  linux_layout_tests_slimming_paint_v2 on master.tryserver.chromium.linux
(JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_layout_te...)

[Xianzhu, 2016-09-02 15:54:35]

The CQ bit was checked by wangxianzhu@chromium.org

[Xianzhu, 2016-09-02 15:54:36]

The patchset sent to the CQ was uploaded after l-g-t-m from pdr@chromium.org
Link to the patchset: https://codereview.chromium.org/2301303002/#ps60001
(title: "Rebase")

[commit-bot: I haz the power, 2016-09-02 15:54:56]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-09-02 16:00:59]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-09-02 16:01:00]

Try jobs failed on following builders:
  linux_layout_tests_slimming_paint_v2 on master.tryserver.chromium.linux
(JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_layout_te...)

[commit-bot: I haz the power, 2016-09-02 16:03:45]

Description was changed from

==========
Reland Compile under-invalidation checking in all builds

This relands https://codereview.chromium.org/2299223002 which was
reverted because of random pixel under-invalidations.

Now clear the bitmaps before drawing the pictures.

BTW changed the error indication color from solid red to translucent
magenta to distinguish from the normal red pixels in test results.

> This will let the tests having under-invalidation issues produce the
> same result on all builds, so enable us to rebaseline the tests with
> failure results. In this way we can track either regressions and
> progressions instead of ignoring the tests.
>
> BUG=619103
>
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
>
> Review-Url: https://codereview.chromium.org/2299223002
> Cr-Commit-Position: refs/heads/master@{#416060}

BUG=619103
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
==========

to

==========
Reland Compile under-invalidation checking in all builds

This relands https://codereview.chromium.org/2299223002 which was
reverted because of random pixel under-invalidations.

Now clear the bitmaps before drawing the pictures.

BTW changed the error indication color from solid red to translucent
magenta to distinguish from the normal red pixels in test results.

> This will let the tests having under-invalidation issues produce the
> same result on all builds, so enable us to rebaseline the tests with
> failure results. In this way we can track either regressions and
> progressions instead of ignoring the tests.
>
> BUG=619103
>
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
>
> Review-Url: https://codereview.chromium.org/2299223002
> Cr-Commit-Position: refs/heads/master@{#416060}

BUG=619103
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
R=pdr@chromium.org

Committed: https://crrev.com/730c50d75791af41558efece2aa560a0c2fc97e9
Cr-Commit-Position: refs/heads/master@{#416276}
==========

[commit-bot: I haz the power, 2016-09-02 16:03:47]

Patchset 4 (id:??) landed as
https://crrev.com/730c50d75791af41558efece2aa560a0c2fc97e9
Cr-Commit-Position: refs/heads/master@{#416276}

[Xianzhu, 2016-09-02 16:04:00]

Description was changed from

==========
Reland Compile under-invalidation checking in all builds

This relands https://codereview.chromium.org/2299223002 which was
reverted because of random pixel under-invalidations.

Now clear the bitmaps before drawing the pictures.

BTW changed the error indication color from solid red to translucent
magenta to distinguish from the normal red pixels in test results.

> This will let the tests having under-invalidation issues produce the
> same result on all builds, so enable us to rebaseline the tests with
> failure results. In this way we can track either regressions and
> progressions instead of ignoring the tests.
>
> BUG=619103
>
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
>
> Review-Url: https://codereview.chromium.org/2299223002
> Cr-Commit-Position: refs/heads/master@{#416060}

BUG=619103
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
R=pdr@chromium.org

Committed: https://crrev.com/730c50d75791af41558efece2aa560a0c2fc97e9
Cr-Commit-Position: refs/heads/master@{#416276}
==========

to

==========
Reland Compile under-invalidation checking in all builds

This relands https://codereview.chromium.org/2299223002 which was
reverted because of random pixel under-invalidations.

Now clear the bitmaps before drawing the pictures.

BTW changed the error indication color from solid red to translucent
magenta to distinguish from the normal red pixels in test results.

> This will let the tests having under-invalidation issues produce the
> same result on all builds, so enable us to rebaseline the tests with
> failure results. In this way we can track either regressions and
> progressions instead of ignoring the tests.
>
> BUG=619103
>
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
>
> Review-Url: https://codereview.chromium.org/2299223002
> Cr-Commit-Position: refs/heads/master@{#416060}

BUG=619103
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
R=pdr@chromium.org
==========

[Xianzhu, 2016-09-02 16:04:01]

Committed patchset #4 (id:60001) to pending queue manually as
c452a1ccf272f35a6fc83be009852579908abd9d (presubmit successful).

[Matt Giuca, 2016-09-05 05:43:36]

A revert of this CL (patchset #4 id:60001) has been created in
https://codereview.chromium.org/2309193002/ by mgiuca@chromium.org.

The reason for reverting is: Causing failures on WebKit Mac 10.9:
https://build.chromium.org/p/chromium.webkit/builders/WebKit%20Mac10.9

updating-scrolling-container-and-content.html is hitting the new CHECK on
PaintController.cpp:662. It's not clear why this passed on build #36527; all
other builds since this landed failed..

[Matt Giuca, 2016-09-05 05:45:57]

On 2016/09/05 05:43:36, Matt Giuca wrote:
> A revert of this CL (patchset #4 id:60001) has been created in
> https://codereview.chromium.org/2309193002/ by mailto:mgiuca@chromium.org.
> 
> The reason for reverting is: Causing failures on WebKit Mac 10.9:
> https://build.chromium.org/p/chromium.webkit/builders/WebKit%20Mac10.9
> 
> updating-scrolling-container-and-content.html is hitting the new CHECK on
> PaintController.cpp:662. It's not clear why this passed on build #36527; all
> other builds since this landed failed..

Sorry to revert your reland :(

[Matt Giuca, 2016-09-05 07:11:50]

On 2016/09/05 05:45:57, Matt Giuca wrote:
> On 2016/09/05 05:43:36, Matt Giuca wrote:
> > A revert of this CL (patchset #4 id:60001) has been created in
> > https://codereview.chromium.org/2309193002/ by mailto:mgiuca@chromium.org.
> > 
> > The reason for reverting is: Causing failures on WebKit Mac 10.9:
> > https://build.chromium.org/p/chromium.webkit/builders/WebKit%20Mac10.9
> > 
> > updating-scrolling-container-and-content.html is hitting the new CHECK on
> > PaintController.cpp:662. It's not clear why this passed on build #36527; all
> > other builds since this landed failed..
> 
> Sorry to revert your reland :(

Note: There were some merge conflicts during the revert, which I manually fixed:
https://codereview.chromium.org/2309193002/

Therefore, when re-landing this, if you reverse-apply my revert CL (instead of
just trying to reland this CL), you will not have to deal with those conflicts
yourself :)

[Xianzhu, 2016-09-06 16:05:32]

A revert of this CL (patchset #4 id:60001) has been created in
https://codereview.chromium.org/2314863002/ by wangxianzhu@chromium.org.

The reason for reverting is: This should be landed to discover
under-invalidations. If any test hit this, we should mark the test fail instead
of reverting this..

[Xianzhu, 2016-09-06 16:08:19]

On 2016/09/06 16:05:32, Xianzhu wrote:
> A revert of this CL (patchset #4 id:60001) has been created in
> https://codereview.chromium.org/2314863002/ by
mailto:wangxianzhu@chromium.org.
> 
> The reason for reverting is: This should be landed to discover
> under-invalidations. If any test hit this, we should mark the test fail
instead
> of reverting this..

Ignore the above. I should create a revert of the revert instead of this one.