Issue metadata
Sign in to add a comment
|
Regression: Content becomes blur on turning on/off emulation |
||||||||||||||||||||||
Issue descriptionVersion: 54.0.2806.0 OS: Ubuntu 14.04,windows What steps will reproduce the problem? (1) Launch chrome and go to chrome://history >> Make sure you have some history (2)Open devtools using F12 >> Click on inspect option >> Now click emulation and select Nexus5X >> Now Off emulation and then again On it[Please refer video] and observe content Expected: Content should be seen blur on on/off of emulation. Actual: Instead content is seen blur NOTE: Issue is also seen in chrome://downloads This is a regression issue broken in M53. Good Build: 53.0.2776.0 dev Bad Build: 53.0.2777.0 dev CHANGELOG URL: https://chromium.googlesource.com/chromium/src/+log/b6de7c9a7644fbcde203abda75c7a243dc4038da..8f268bb64214eab73f00216b96b919a1b2955999 Suspecting https://codereview.chromium.org/2068413002 from changelog. @vmpstr: Please confirm the issue and help in re-assigning if it is not related to your change.
,
Aug 17 2016
As per comment#1 checked on latest 54.0.2831.0 Ubuntu 14.04 with steps mentioned in comment#0. Issue is still reproducible. Hence re-opening again. Please find attached video for reference.
,
Sep 17 2016
,
Sep 19 2016
I think this bisect is correct. I can repro this if I switch back and forth from emulated mode and back very quickly. It doesn't happen every time, though.
Looking at the page css itself:
@media only screen and (-webkit-max-device-pixel-ratio: 1) {
:host {
will-change: transform;
}
This appears in both downloads and history pages (and appears to come from polymer). This is applied to the iron-list element which is why the list itself appears blurry but the bar on top does not.
Polymer probably wants will-change:transform on the list itself so that it scrolls quickly, but I think this is undesirable behavior while scaling, although the scaling is out of the page's control. Maybe we need to reraster everything after switching emulation modes? Maybe Polymer should use some other layerization trigger?
,
Oct 4 2016
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/8a4cb786c49505784365acdb1ca4037df64b341f commit 8a4cb786c49505784365acdb1ca4037df64b341f Author: enne <enne@chromium.org> Date: Tue Oct 04 20:36:45 2016 Fix blurry content after disabling dev tools emulation There's a race between page resize and devtools emulation disabling which can change the root layer transform scale. If a commit occurs after the page resize but before the scale, then layers end up picking the wrong scale, and if they have will-change on them then they will not change again after the root layer transform scale. This scale isn't really part of the page or under the author's control, and so will-change should be ignored when devtools changes the scale and content should always be rerastered and the ideal scale. This patch implements that behavior by plumbing a flag all the way from the devtools scale change to LayerTreeHost which then then resets the raster scale on all the picture layers after commit. BUG= 630979 CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel Review-Url: https://codereview.chromium.org/2390923002 Cr-Commit-Position: refs/heads/master@{#422914} [modify] https://crrev.com/8a4cb786c49505784365acdb1ca4037df64b341f/cc/blimp/layer_tree_host_remote.cc [modify] https://crrev.com/8a4cb786c49505784365acdb1ca4037df64b341f/cc/blimp/layer_tree_host_remote.h [modify] https://crrev.com/8a4cb786c49505784365acdb1ca4037df64b341f/cc/layers/picture_layer_impl.h [modify] https://crrev.com/8a4cb786c49505784365acdb1ca4037df64b341f/cc/trees/layer_tree_host.h [modify] https://crrev.com/8a4cb786c49505784365acdb1ca4037df64b341f/cc/trees/layer_tree_host_in_process.cc [modify] https://crrev.com/8a4cb786c49505784365acdb1ca4037df64b341f/cc/trees/layer_tree_host_in_process.h [modify] https://crrev.com/8a4cb786c49505784365acdb1ca4037df64b341f/cc/trees/layer_tree_host_unittest_picture.cc [modify] https://crrev.com/8a4cb786c49505784365acdb1ca4037df64b341f/cc/trees/layer_tree_impl.cc [modify] https://crrev.com/8a4cb786c49505784365acdb1ca4037df64b341f/cc/trees/layer_tree_impl.h [modify] https://crrev.com/8a4cb786c49505784365acdb1ca4037df64b341f/content/renderer/gpu/render_widget_compositor.cc [modify] https://crrev.com/8a4cb786c49505784365acdb1ca4037df64b341f/content/renderer/gpu/render_widget_compositor.h [modify] https://crrev.com/8a4cb786c49505784365acdb1ca4037df64b341f/third_party/WebKit/Source/web/DevToolsEmulator.cpp [modify] https://crrev.com/8a4cb786c49505784365acdb1ca4037df64b341f/third_party/WebKit/Source/web/WebViewImpl.cpp [modify] https://crrev.com/8a4cb786c49505784365acdb1ca4037df64b341f/third_party/WebKit/Source/web/WebViewImpl.h [modify] https://crrev.com/8a4cb786c49505784365acdb1ca4037df64b341f/third_party/WebKit/Source/web/tests/WebViewTest.cpp [modify] https://crrev.com/8a4cb786c49505784365acdb1ca4037df64b341f/third_party/WebKit/public/platform/WebLayerTreeView.h
,
Oct 4 2016
|
|||||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||||
Comment 1 by vmp...@chromium.org
, Aug 11 2016