seem div#login_panel causes this bug.

It seems it can only be reproduced in chrome for android.

Inspector with emulator set to on, specification set to nexus 7 can also reproduce this bug.

Seems causes by composited scrolling.

#0  cc::Layer::SetSubtreePropertyChanged (this=0xb629dc00) at ../../cc/layers/layer.cc:1407
#1  cc::Layer::SetShouldFlattenTransform (this=0xb629dc00, should_flatten=<optimized out>) at ../../cc/layers/layer.cc:1028
#2  0xc83f4750 in blink::ApplyToGraphicsLayers<blink::UpdateShouldFlattenTransformFunctor> (mode=1, f=..., mapping=0x100c0640)
    at ../../third_party/WebKit/Source/core/layout/compositing/CompositedLayerMapping.cpp:1992
#3  blink::CompositedLayerMapping::updateShouldFlattenTransform (this=this@entry=0x100c0640)
    at ../../third_party/WebKit/Source/core/layout/compositing/CompositedLayerMapping.cpp:2086
#4  0xc83f7cb2 in blink::CompositedLayerMapping::updateGraphicsLayerGeometry (this=this@entry=0x100c0640, 
    compositingContainer=compositingContainer@entry=0xbf214d20, compositingStackingContext=0xbf214d20, layersNeedingPaintInvalidation=...)
    at ../../third_party/WebKit/Source/core/layout/compositing/CompositedLayerMapping.cpp:1065

The call trace above is set should flatten transform to true, here's the call trace set should flatten transform to false:
#0  cc::Layer::SetShouldFlattenTransform (this=0xb87e7900, should_flatten=false) at ../../cc/layers/layer.cc:1024
#1  0xc83f5728 in blink::CompositedLayerMapping::updateShouldFlattenTransform (this=this@entry=0x100c0640)
    at ../../third_party/WebKit/Source/core/layout/compositing/CompositedLayerMapping.cpp:2102
#2  0xc83f8cb2 in blink::CompositedLayerMapping::updateGraphicsLayerGeometry (this=this@entry=0x100c0640, 
    compositingContainer=compositingContainer@entry=0xbfd24d20, compositingStackingContext=0xbfd24d20, layersNeedingPaintInvalidation=...)
    at ../../third_party/WebKit/Source/core/layout/compositing/CompositedLayerMapping.cpp:1065

The toggle above make the cc thinks there is a sub tree property changed happened.

diff --git a/third_party/WebKit/Source/core/layout/compositing/CompositedLayerMapping.cpp b/third_party/WebKit/Source/core/layout/compositing/CompositedLayerMapping.cpp
index 2d2167ead07f..cf0ab087762e 100644
--- a/third_party/WebKit/Source/core/layout/compositing/CompositedLayerMapping.cpp
+++ b/third_party/WebKit/Source/core/layout/compositing/CompositedLayerMapping.cpp
@@ -2181,8 +2181,8 @@ void CompositedLayerMapping::UpdateShouldFlattenTransform() {
   // causes unclipped render surfaces which cause bugs.
   //  http://crbug.com/521768 
   if (HasScrollingLayer()) {
-    graphics_layer_->SetShouldFlattenTransform(false);
-    scrolling_layer_->SetShouldFlattenTransform(false);
+    // graphics_layer_->SetShouldFlattenTransform(false);
+    // scrolling_layer_->SetShouldFlattenTransform(false);
   }
 }


Comment out these two lines stop the frame redundant generation.

Both the mapping->MainGraphicsLayer() and mapping->ScrollingContentsLayer() SetShouldFlattenTransform to true by UpdateShouldFlattenTransformFunctor

seem div#login_panel causes this bug.

No it's not. It is div.topnav_list

@manjian2006 -- Thank You for the report.
Could you please provide us with device details, OS and also a screen cast which would help us triage the issue further.
Thanks in Advance.

I am using a nexus 7 model. But you don't need such a device. Just use the inspector and toggle the device toolbar then it will reproduce.

Thank you for providing more feedback. Adding requester "msrchandra@chromium.org" to the cc list and removing "Needs-Feedback" label.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

Could some one from Dev team please look into the issue and provide an update.
Thanks in Advance.

over to trchen@ for triage. Can confirm (via fps meter) that we do appear to keep generated frames although nothing is visibly changing.

Here's a minimal repro: http://jsbin.com/rojetevoji/

This bug is a combination of unfortunate multi-fail...

1. The page author added some animation (through RAF or other mechanism?) that is not visible. This caused us to recalculate style @ 60fps, which is already bad for battery even if we don't actually redraw the frame.

2. PaintLayer::StyleDidChange / PaintLayer::AttemptDirectCompositingUpdate didn't bother to look into the style diffs, and issued compositing update anyway.

3. During compositing update, CompositedLayerMapping::UpdateShouldFlattenTransform() flips flatten flag back and forth, as the reporter discovered.

4. cc::Layer::SetShouldFlattenTransform dirtied the layer too early, so the said back and forth resulted in a redraw despite nothing changed.

I don't think this needs to be cherry-picked to a branch though. 1. It doesn't seem to be a new regression, and 2. It is more of an optimization for badly-written pages and 3. It doesn't affect normal operation.

To the reporter: Thank you for reporting! Your analysis is really helpful for us to pinpoint the root cause. I will be working on a fix now.

My pleasure. I am a browser developer. So I just fix this by now allowing the CompositedLayerMapping::UpdateShouldFlattenTransform() flips flatten flag back and forth for now.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/f498304ebf21ed9de9b459a0087d8c9817e25c52

commit f498304ebf21ed9de9b459a0087d8c9817e25c52
Author: Tien-Ren Chen <trchen@chromium.org>
Date: Tue Nov 28 00:35:47 2017

[Blink] Avoid SetShouldFlattenTransform() back and forth

In CompositedLayerMapping::UpdateShouldFlattenTransform() we apply
different values on the same layer multiple times for different layer
configuration. This resulted in unnecessary redraw in the compositor.

This CL collapsed the logic so that each layer will be programmed
exactly one time.

BUG= 783614 

Cq-Include-Trybots: master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
Change-Id: Id2551e6e49d2555359da9d954b44555cb1f41a63
Reviewed-on: https://chromium-review.googlesource.com/780803
Commit-Queue: Tien-Ren Chen <trchen@chromium.org>
Reviewed-by: Chris Harrelson <chrishtr@chromium.org>
Cr-Commit-Position: refs/heads/master@{#519477}
[modify] https://crrev.com/f498304ebf21ed9de9b459a0087d8c9817e25c52/third_party/WebKit/Source/core/paint/compositing/CompositedLayerMapping.cpp