<Regression> Firefox - TResize - MacOSX 10.8 - 51.5%

[nob...@cruncher.build.mozilla.org]

Regression: Firefox - TResize - MacOSX 10.8 - 51.5% increase
------------------------------------------------------------
Previous: avg 11.937 stddev 0.412 of 12 runs up to revision 5b6e82e7bbbf
New : avg 18.082 stddev 0.693 of 12 runs since revision 56da72c95d48
Change : +6.144 (51.5% / z=14.898)
Graph : http://mzl.la/1mfjScN

Changeset range: http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=5b6e82e7bbbf&tochange=56da72c95d48

Changesets:
* http://hg.mozilla.org/mozilla-central/rev/4b1495af7084
: Nicolas Silva <ni...@mozilla.com> - Bug 994024 - Add assertions in ShadowLayers.cpp. r=bjacob
: http://bugzilla.mozilla.org/show_bug.cgi?id=994024

* http://hg.mozilla.org/mozilla-central/rev/7b0b53c3f057
: Nicolas Silva <ni...@mozilla.com> - Bug 994024 - Add assertions in ImageBridgeChild. r=bjacob
: http://bugzilla.mozilla.org/show_bug.cgi?id=994024

* http://hg.mozilla.org/mozilla-central/rev/7e9c6b964434
: Nicolas Silva <ni...@mozilla.com> - Bug 994024 - Discard TextureClients that we failed to allocate or lock. r=bjacob
: http://bugzilla.mozilla.org/show_bug.cgi?id=994024

* http://hg.mozilla.org/mozilla-central/rev/186859975a8c
: Nicolas Silva <ni...@mozilla.com> - Bug 995143 - Add missing CanExposeDrawTarget in CairoTextureClientD3D9. r=mattwoodrow
: http://bugzilla.mozilla.org/show_bug.cgi?id=995143

* http://hg.mozilla.org/mozilla-central/rev/92dd49bd29b6
: Terrence Cole <terr...@mozilla.com> - Bug 990336 - Deduplicate the slots edge buffer normally; r=jonco
: http://bugzilla.mozilla.org/show_bug.cgi?id=990336

* http://hg.mozilla.org/mozilla-central/rev/febd5ad98adb
: L. David Baron <dba...@dbaron.org> - Bug 995661 patch 1 - Remove dead code resulting from removal of gfxPlatform::SupportsAzureContent(). r=mattwoodrow

This should have been removed in bug 985049.

The changes here can be understood as a simplification based on drawSurf
always being null.
: http://bugzilla.mozilla.org/show_bug.cgi?id=995661

* http://hg.mozilla.org/mozilla-central/rev/2e0abf6eeec3
: L. David Baron <dba...@dbaron.org> - Bug 995661 patch 2 - Use OP_OVER rather than OP_SOURCE to handle canvas drawWindow calls that don't cover the canvas or (I believe) that contain transparency. r=mattwoodrow

This is needed to allow us to return to using partial-canvas invalidates
in the reftest harness, though only for the platforms that go through
this codepath.
: http://bugzilla.mozilla.org/show_bug.cgi?id=995661

* http://hg.mozilla.org/mozilla-central/rev/33bfd3ceee1f
: L. David Baron <dba...@dbaron.org> - Bug 995661 patch 3 - Add mochitest for canvas drawWindow. r=mattwoodrow

This test covers some basic cases, such as drawing the complete window
to a canvas of the same size, drawing a part of the window to a
different offset within the canvas, drawing parts of the window against
different backgrounds and against a transparent background, and doing
multiple such drawWindow calls on the same canvas.

It runs its tests both with an iframe (whose background is transparent)
and content in a new window (which has a white backdrop).

Bug 995721 will add further tests (in mochitest-chrome) that exercise
drawWindow from a toplevel chrome window, which exercises the
DRAWWINDOW_USE_WIDGET_LAYERS codepath, which these tests do not.

The final 4 tests in each set of 7 fail without patch 2 in this bug.
: http://bugzilla.mozilla.org/show_bug.cgi?id=995661

* http://hg.mozilla.org/mozilla-central/rev/4428373539b7
: L. David Baron <dba...@dbaron.org> - Bug 995721 patch 1 - Honor the translation on mShadowTarget in ClientLayerManager::MakeSnapshotIfRequired. r=mattwoodrow

This switches from DrawTarget::CopySurface (which ignores the transform)
to DrawSurface (which honors the transform).
: http://bugzilla.mozilla.org/show_bug.cgi?id=995721

* http://hg.mozilla.org/mozilla-central/rev/7ba6e4eeb392
: L. David Baron <dba...@dbaron.org> - Bug 995721 patch 2 - Run drawWindow mochitests on a toplevel window. r=mattwoodrow

Without patch 1, the final 4 of the set of 7 tests fail on Linux with
accelerated layers (whereas all the ones they were copied from, added in
bug 995661, pass, given the patches in that bug but not this one). (I
don't need to enable skia-canvas to see the failures without patch 1. I
thought that didn't match my earlier testing, but I probably just never
tested that combination. This means the failures happen even without
CanvasRenderingContext2D::DrawWindow going through the codepath where it
creates a DrawTarget.)
: http://bugzilla.mozilla.org/show_bug.cgi?id=995721

* http://hg.mozilla.org/mozilla-central/rev/33d397b3b008
: L. David Baron <dba...@dbaron.org> - Bug 995410 - Call shouldSnapshotWholePage rather than testing for its existence. r=mattwoodrow

This is clearly what was intended when this code was introduced in bug
876626.
: http://bugzilla.mozilla.org/show_bug.cgi?id=995410

* http://hg.mozilla.org/mozilla-central/rev/c7c4c0d0f613
: Masayuki Nakano <masa...@d-toybox.com> - Bug 930893 part.1 Implement constructor of KeyboardEvent r=smaug
: http://bugzilla.mozilla.org/show_bug.cgi?id=930893

* http://hg.mozilla.org/mozilla-central/rev/220e9d8b1454
: Masayuki Nakano <masa...@d-toybox.com> - Bug 930893 part.2 Implement KeyboardEvent.initKeyboardEvent() r=smaug
: http://bugzilla.mozilla.org/show_bug.cgi?id=930893

* http://hg.mozilla.org/mozilla-central/rev/8e5f7f4427df
: Timothy Nikkel <tni...@gmail.com> - Bug 994861. Only overlay scrollbars should be positioned and have z-index. r=roc

Bug 896443 added these but didn't guard them on overlay scrollbars.

When drawing the scrollbars layout interprets a positioned scrollbar as an scrollbar that should be placed above the content (like overlay scrollbars). Classic scrollbars should be drawn in the background layer.
: http://bugzilla.mozilla.org/show_bug.cgi?id=994861

* http://hg.mozilla.org/mozilla-central/rev/a1b2bc9af796
: Terrence Cole <terr...@mozilla.com> - Bug 990336 - Deduplicate subsequent pushes to the store buffer; r=jonco
: http://bugzilla.mozilla.org/show_bug.cgi?id=990336

* and 56 more

Bugs:
* http://bugzilla.mozilla.org/show_bug.cgi?id=990336 - Improve the performance of DenseRangeWriteBarrierPost
* http://bugzilla.mozilla.org/show_bug.cgi?id=995076 - Unconditionally replace null/undefined/Magic-typed OSR slots
* http://bugzilla.mozilla.org/show_bug.cgi?id=995077 - Preserve phis whose slots are arguments object inside functions that need it
* http://bugzilla.mozilla.org/show_bug.cgi?id=995721 - ClientLayerManager::BeginTransactionWithTarget doesn't honor translation on target
* http://bugzilla.mozilla.org/show_bug.cgi?id=994444
* http://bugzilla.mozilla.org/show_bug.cgi?id=983053 - Is it really necessary to js_SuppressDeletedElements on obj if CanOptimizeForDenseStorage(obj, ...) was true?
* http://bugzilla.mozilla.org/show_bug.cgi?id=993423
* http://bugzilla.mozilla.org/show_bug.cgi?id=995345 - Improve efficiency of NS_DescribeCodeAddress for mac/linux
* http://bugzilla.mozilla.org/show_bug.cgi?id=980117 - Implement sccache for Windows builds
* http://bugzilla.mozilla.org/show_bug.cgi?id=994711 - Expose --this-chunk and --total-chunks to {reftest|crashtest}-remote mach targets
* http://bugzilla.mozilla.org/show_bug.cgi?id=901146 - Fix misleading comment in nsImageFrame.cpp
* http://bugzilla.mozilla.org/show_bug.cgi?id=994971 - Don't export nsObserverService.h
* http://bugzilla.mozilla.org/show_bug.cgi?id=990876 - Remove TextureClientSurface
* http://bugzilla.mozilla.org/show_bug.cgi?id=994335 - startup crash in nsXULPrototypeDocument::GetCompilationGlobal()
* http://bugzilla.mozilla.org/show_bug.cgi?id=471713 - Document opcodes (stack conventions, preconditions, postconditions, other implicit assumptions) in jsopcode.tbl
* and 33 more

[nob...@cruncher.build.mozilla.org]

Regression: Firefox - TResize - MacOSX 10.6 (rev4) - 25% increase
-----------------------------------------------------------------
Previous: avg 17.708 stddev 0.455 of 12 runs up to revision 5b6e82e7bbbf
New : avg 22.128 stddev 0.374 of 12 runs since revision 56da72c95d48
Change : +4.420 (25% / z=9.720)
Graph : http://mzl.la/1mfjXxc

[nob...@cruncher.build.mozilla.org]

Improvement: Firefox-Non-PGO - tscroll-ASAP - WINNT 6.2 x64 - 8.43% decrease
----------------------------------------------------------------------------
Previous: avg 7.657 stddev 0.092 of 12 runs up to revision 5b6e82e7bbbf
New : avg 7.012 stddev 0.091 of 12 runs since revision 56da72c95d48
Change : -0.646 (8.43% / z=7.013)
Graph : http://mzl.la/1mfm585

[nob...@cruncher.build.mozilla.org]

Regression: mobile - Ts, Paint - Android 4.0.4 - 6% increase
------------------------------------------------------------
Previous: avg 3436.608 stddev 31.994 of 12 runs up to revision 5b6e82e7bbbf
New : avg 3642.742 stddev 59.883 of 12 runs since revision 56da72c95d48
Change : +206.133 (6% / z=6.443)
Graph : http://mzl.la/1nd8a5h

[nob...@cruncher.build.mozilla.org]

Regression: Firefox-Non-PGO - tscroll-ASAP - Ubuntu HW 12.04 - 3.68% increase
-----------------------------------------------------------------------------
Previous: avg 9.652 stddev 0.027 of 12 runs up to revision 5b6e82e7bbbf
New : avg 10.008 stddev 0.030 of 12 runs since revision 3d34a3b6443a
Change : +0.356 (3.68% / z=13.336)
Graph : http://mzl.la/1jPi1KS

Changeset range: http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=5b6e82e7bbbf&tochange=3d34a3b6443a

* and 112 more

Bugs:
* http://bugzilla.mozilla.org/show_bug.cgi?id=990336 - Improve the performance of DenseRangeWriteBarrierPost

* http://bugzilla.mozilla.org/show_bug.cgi?id=990003 - [Tarako][Perf][Dialer] It takes a long time for the call screen shows up
* http://bugzilla.mozilla.org/show_bug.cgi?id=2
* http://bugzilla.mozilla.org/show_bug.cgi?id=994659 - Exceptions in all command line code when reloading shell.html

* http://bugzilla.mozilla.org/show_bug.cgi?id=995076 - Unconditionally replace null/undefined/Magic-typed OSR slots
* http://bugzilla.mozilla.org/show_bug.cgi?id=995077 - Preserve phis whose slots are arguments object inside functions that need it
* http://bugzilla.mozilla.org/show_bug.cgi?id=995721 - ClientLayerManager::BeginTransactionWithTarget doesn't honor translation on target

* http://bugzilla.mozilla.org/show_bug.cgi?id=970263 - B2G NFC: Enable NFC when running NFC marionette tests
* http://bugzilla.mozilla.org/show_bug.cgi?id=970007 - [tarako]monkey test crash at libxul.so!BufferUnrotate

* http://bugzilla.mozilla.org/show_bug.cgi?id=994444
* http://bugzilla.mozilla.org/show_bug.cgi?id=983053 - Is it really necessary to js_SuppressDeletedElements on obj if CanOptimizeForDenseStorage(obj, ...) was true?
* http://bugzilla.mozilla.org/show_bug.cgi?id=993423
* http://bugzilla.mozilla.org/show_bug.cgi?id=995345 - Improve efficiency of NS_DescribeCodeAddress for mac/linux

* http://bugzilla.mozilla.org/show_bug.cgi?id=993095 - Add ta locale
* http://bugzilla.mozilla.org/show_bug.cgi?id=975299 - [Sora][STK]Can not enter to 'Refresh' Menu again after execute Refresh command.
* and 65 more

[nob...@cruncher.build.mozilla.org]

Improvement: Firefox-Non-PGO - SVG-ASAP - Ubuntu HW 12.04 - 5.19% decrease
--------------------------------------------------------------------------
Previous: avg 758.682 stddev 4.740 of 12 runs up to revision 5b6e82e7bbbf
New : avg 719.284 stddev 5.439 of 12 runs since revision 3d34a3b6443a
Change : -39.398 (5.19% / z=8.312)
Graph : http://mzl.la/1jPi2P1

[nob...@cruncher.build.mozilla.org]

Regression: Firefox-Non-PGO - tscroll-ASAP - Ubuntu HW 12.04 x64 - 6.69% increase
---------------------------------------------------------------------------------
Previous: avg 8.695 stddev 0.105 of 12 runs up to revision 5b6e82e7bbbf
New : avg 9.277 stddev 0.059 of 12 runs since revision 3d34a3b6443a
Change : +0.582 (6.69% / z=5.528)
Graph : http://mzl.la/1jPkrZW

[nob...@cruncher.build.mozilla.org]

Improvement: Firefox-Non-PGO - SVG-ASAP - Ubuntu HW 12.04 x64 - 5.65% decrease
------------------------------------------------------------------------------
Previous: avg 597.481 stddev 4.714 of 12 runs up to revision 5b6e82e7bbbf
New : avg 563.704 stddev 3.459 of 12 runs since revision 3d34a3b6443a
Change : -33.777 (5.65% / z=7.165)
Graph : http://mzl.la/1jPktRN

[nob...@cruncher.build.mozilla.org]

Regression: Firefox - tscroll-ASAP - Ubuntu HW 12.04 - 3.35% increase
---------------------------------------------------------------------
Previous: avg 9.079 stddev 0.021 of 12 runs up to revision 5b6e82e7bbbf
New : avg 9.383 stddev 0.022 of 12 runs since revision 3d34a3b6443a
Change : +0.304 (3.35% / z=14.261)
Graph : http://mzl.la/1r5gtO5

[nob...@cruncher.build.mozilla.org]

Improvement: Firefox - SVG-ASAP - Ubuntu HW 12.04 - 5.81% decrease
------------------------------------------------------------------
Previous: avg 619.678 stddev 3.965 of 12 runs up to revision 5b6e82e7bbbf
New : avg 583.663 stddev 3.611 of 12 runs since revision 3d34a3b6443a
Change : -36.015 (5.81% / z=9.084)
Graph : http://mzl.la/1r5gxxd

[nob...@cruncher.build.mozilla.org]

Regression: Firefox - tscroll-ASAP - Ubuntu HW 12.04 x64 - 5.31% increase
-------------------------------------------------------------------------
Previous: avg 8.090 stddev 0.086 of 12 runs up to revision 5b6e82e7bbbf
New : avg 8.519 stddev 0.092 of 12 runs since revision 3d34a3b6443a
Change : +0.430 (5.31% / z=5.012)
Graph : http://mzl.la/1mdWZZW

[nob...@cruncher.build.mozilla.org]

Improvement: Firefox - SVG-ASAP - Ubuntu HW 12.04 x64 - 8.04% decrease
----------------------------------------------------------------------
Previous: avg 506.398 stddev 5.032 of 12 runs up to revision 5b6e82e7bbbf
New : avg 465.697 stddev 4.997 of 12 runs since revision 3d34a3b6443a
Change : -40.701 (8.04% / z=8.088)
Graph : http://mzl.la/1mdX2EY

[nob...@cruncher.build.mozilla.org]

Improvement: Firefox - tscroll-ASAP - WINNT 6.2 x64 - 9.31% decrease
--------------------------------------------------------------------
Previous: avg 7.163 stddev 0.103 of 12 runs up to revision 5b6e82e7bbbf
New : avg 6.497 stddev 0.057 of 12 runs since revision 3d34a3b6443a
Change : -0.667 (9.31% / z=6.482)
Graph : http://mzl.la/1me4LTu