New issue
Advanced search Search tips
Note: Color blocks (like █ or █) mean that a user may not be available. Tooltip shows the reason.

Issue 742339 link

Starred by 1 user
Status: Fixed
Owner:
bokan@chromium.org
Closed: Nov 2017
Cc:
█ pmeenan@chromium.org
thestig@chromium.org
jamiewa...@chromium.org
sullivan@chromium.org
skobes@chromium.org
bokan@chromium.org
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug-Regression



Sign in to add a comment

5.1%-165.2% regression in loading.desktop at 477411:485736

Project Member Reported by pmeenan@chromium.org, Jul 13 2017 
See the link to graphs below.
Project Member

Comment 1 by 42576172...@developer.gserviceaccount.com, Jul 13 2017 

All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=742339

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=fdc5e22c8c205be8c8e8f57a07c662e521590025b9e6e4fb6ac1a5756777b0f8


Bot(s) for this bug's original alert(s):

chromium-rel-mac11
chromium-rel-mac11-pro
chromium-rel-win7-gpu-ati
chromium-rel-win7-gpu-nvidia
chromium-rel-win8-dual
linux-release
Project Member

Comment 3 by 42576172...@developer.gserviceaccount.com, Jul 14 2017 


=== BISECT JOB RESULTS ===
Bisect failed for unknown reasons

Please contact the team (see below) and report the error.


Bisect Details
  Configuration: linux_perf_bisect
  Benchmark    : loading.desktop
  Metric       : timeToFirstMeaningfulPaint_avg/pcv1-cold/FarsNews


To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=FarsNews loading.desktop

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8974178402848364496


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 6 by 42576172...@developer.gserviceaccount.com, Jul 27 2017

Cc: bokan@chromium.org
Owner: bokan@chromium.org

=== Auto-CCing suspected CL author bokan@chromium.org ===

Hi bokan@chromium.org, the bisect results pointed to your CL, please take a look at the
results.


=== BISECT JOB RESULTS ===
Perf regression found with culprit

Suspected Commit
  Author : David Bokan
  Commit : 55c943bf50024b941adb671a026f021810119732
  Date   : Tue Jul 11 18:56:45 2017
  Subject: Make document inherit bg style from rootScroller

Bisect Details
  Configuration: winx64nvidia_perf_bisect
  Benchmark    : loading.desktop
  Metric       : timeToFirstContentfulPaint_avg/pcv1-warm/FarsNews
  Change       : 6.53% | 101.3975 -> 108.020833333

Revision             Result                  N
chromium@485679      101.397 +- 4.48201      6      good
chromium@485683      99.4763 +- 5.37877      6      good
chromium@485685      102.385 +- 8.7278       9      good
chromium@485686      111.352 +- 5.92291      6      bad       <--
chromium@485687      109.255 +- 3.60741      6      bad
chromium@485694      107.765 +- 3.82644      6      bad
chromium@485708      109.405 +- 6.30732      6      bad
chromium@485736      108.021 +- 6.02243      6      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release_x64 --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=FarsNews loading.desktop

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8972881607876805424


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 7 by 42576172...@developer.gserviceaccount.com, Jul 27 2017 


=== BISECT JOB RESULTS ===
Perf regression found with culprit

Suspected Commit
  Author : David Bokan
  Commit : 55c943bf50024b941adb671a026f021810119732
  Date   : Tue Jul 11 18:56:45 2017
  Subject: Make document inherit bg style from rootScroller

Bisect Details
  Configuration: mac_pro_perf_bisect
  Benchmark    : loading.desktop
  Metric       : timeToFirstContentfulPaint_avg/pcv1-warm/FarsNews
  Change       : 9.22% | 113.156111106 -> 123.354555556

Revision             Result                  N
chromium@485638      113.156 +- 9.17786      9       good
chromium@485667      116.027 +- 18.0536      14      good
chromium@485681      112.921 +- 9.95798      9       good
chromium@485685      115.397 +- 14.8663      9       good
chromium@485686      121.474 +- 9.76306      14      bad       <--
chromium@485687      121.39 +- 12.7729       14      bad
chromium@485688      123.525 +- 15.8827      9       bad
chromium@485695      123.355 +- 13.3164      9       bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=FarsNews loading.desktop

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8972881602936994016


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 9 by 42576172...@developer.gserviceaccount.com, Jul 29 2017 


=== BISECT JOB RESULTS ===
Bisect failed for unknown reasons

Please contact the team (see below) and report the error.


Bisect Details
  Configuration: linux_perf_bisect
  Benchmark    : loading.desktop
  Metric       : timeToFirstMeaningfulPaint_avg/pcv1-warm/AllRecipes


To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=AllRecipes loading.desktop

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8972836371352134528


For feedback, file a bug with component Speed>Bisection

Comment 10 by sullivan@chromium.org, Sep 18 2017 

bokan: any update on this regression?

Comment 11 by bokan@chromium.org, Sep 19 2017

Status: Assigned (was: Untriaged)
Ah, thanks for pinging, this fell off my radar. I'll take a look today.

Comment 15 by bokan@chromium.org, Sep 19 2017 

I tried some of the regressed metrics locally at ToT and with my patch reverted. I couldn't reproduce a regression. That makes sense as my patch only has any effect if the page is setting a document.rootScroller which is still unshipped and behind a flag so it should have no effect on pages. I've kicked off a few bisects on some of the other, clear, regressed metrics...will see what it says.
Project Member

Comment 16 by 42576172...@developer.gserviceaccount.com, Sep 20 2017 


=== BISECT JOB RESULTS ===
Bisect failed for unknown reasons

Please contact the team (see below) and report the error.


Bisect Details
  Configuration: linux_perf_bisect
  Benchmark    : loading.desktop
  Metric       : cpuTimeToFirstMeaningfulPaint_avg/pcv1-cold/AirBnB


To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=AirBnB loading.desktop

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8968025437933089008


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 17 by 42576172...@developer.gserviceaccount.com, Sep 20 2017 


=== BISECT JOB RESULTS ===
Bisect failed for unknown reasons

Please contact the team (see below) and report the error.


Bisect Details
  Configuration: linux_perf_bisect
  Benchmark    : loading.desktop
  Metric       : cpuTimeToFirstMeaningfulPaint_avg/pcv1-warm/Daum


To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=Daum loading.desktop

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8968025483584257264


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 18 by 42576172...@developer.gserviceaccount.com, Sep 20 2017 


=== BISECT JOB RESULTS ===
Bisect failed for unknown reasons

Please contact the team (see below) and report the error.


Bisect Details
  Configuration: linux_perf_bisect
  Benchmark    : loading.desktop
  Metric       : cpuTimeToFirstMeaningfulPaint_avg/pcv1-warm/Chosun


To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=Chosun loading.desktop

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8968025278314850960


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 22 by 42576172...@developer.gserviceaccount.com, Sep 27 2017 


=== BISECT JOB RESULTS ===
Bisect failed for unknown reasons

Please contact the team (see below) and report the error.


Bisect Details
  Configuration: linux_perf_bisect
  Benchmark    : loading.desktop
  Metric       : cpuTimeToFirstMeaningfulPaint_avg/pcv1-cold/Pantip


To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=Pantip loading.desktop

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8967383368793043792


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 23 by 42576172...@developer.gserviceaccount.com, Sep 27 2017 


=== BISECT JOB RESULTS ===
Bisect failed for unknown reasons

Please contact the team (see below) and report the error.


Bisect Details
  Configuration: linux_perf_bisect
  Benchmark    : loading.desktop
  Metric       : cpuTimeToFirstMeaningfulPaint_avg/pcv1-cold/yahoo.co.jp


To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=yahoo.co.jp loading.desktop

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8967383304557983776


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 24 by 42576172...@developer.gserviceaccount.com, Sep 27 2017 


=== BISECT JOB RESULTS ===
Bisect failed for unknown reasons

Please contact the team (see below) and report the error.


Bisect Details
  Configuration: linux_perf_bisect
  Benchmark    : loading.desktop
  Metric       : timeToFirstMeaningfulPaint_avg/pcv1-warm/Sina


To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=Sina loading.desktop

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8967383335355991152


For feedback, file a bug with component Speed>Bisection

Comment 25 by bokan@chromium.org, Sep 27 2017

Cc: sullivan@chromium.org
sullivan@, see #15. I don't think my patch was responsible and some of the metrics have since recovered. The rest do still have a regression but I've run multiple bisects that are all failing but the build bot logs don't show any clear reason. Could you get someone who knows what they're doing here to take a look?

Comment 26 by bokan@chromium.org, Oct 6 2017 

ping
Project Member

Comment 32 by 42576172...@developer.gserviceaccount.com, Nov 15 2017 

😿 Pinpoint job stopped with an error.
https://pinpoint-dot-chromeperf.appspot.com/job/16f81d59f80000
Project Member

Comment 33 by 42576172...@developer.gserviceaccount.com, Nov 15 2017 

📍 Couldn't reproduce a difference.
https://pinpoint-dot-chromeperf.appspot.com/job/1010e479f80000
Project Member

Comment 34 by 42576172...@developer.gserviceaccount.com, Nov 15 2017

Cc: thestig@chromium.org skobes@chromium.org jamiewa...@chromium.org
Owner: skobes@chromium.org
📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/11c72845f80000

Support $CHROME_CONFIG_HOME in Linux default user data dir selection.
By skobes@chromium.org · Tue Jun 27 21:39:02 2017
chromium @ c9ae6a1a1214f69cb9ff0180468c669e82a380e6

Understanding performance regressions:
  http://g.co/ChromePerformanceRegressions
Project Member

Comment 35 by 42576172...@developer.gserviceaccount.com, Nov 15 2017

Owner: bokan@chromium.org

=== Auto-CCing suspected CL author bokan@chromium.org ===

Hi bokan@chromium.org, the bisect results pointed to your CL, please take a look at the
results.


=== BISECT JOB RESULTS ===
Perf regression found with culprit

Suspected Commit
  Author : David Bokan
  Commit : 55c943bf50024b941adb671a026f021810119732
  Date   : Tue Jul 11 18:56:45 2017
  Subject: Make document inherit bg style from rootScroller

Bisect Details
  Configuration: win_8_perf_bisect
  Benchmark    : loading.desktop
  Metric       : timeToFirstMeaningfulPaint_avg/pcv1-warm/FarsNews
  Change       : 5.09% | 250.132500005 -> 262.859499998

Revision             Result                  N
chromium@485661      250.133 +- 5.20698      6      good
chromium@485685      249.68 +- 6.21679       6      good
chromium@485686      263.794 +- 8.85031      6      bad       <--
chromium@485687      262.959 +- 4.69963      6      bad
chromium@485688      264.706 +- 6.74108      6      bad
chromium@485691      263.704 +- 6.23402      6      bad
chromium@485697      262.316 +- 9.97555      6      bad
chromium@485709      262.859 +- 9.02721      6      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=FarsNews loading.desktop

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8962845073134663680


For feedback, file a bug with component Speed>Bisection

Comment 36 by skobes@chromium.org, Nov 16 2017 

FYI - I ran a perf try job on a revert of r482753 and it seems to show no significant change in timeToFirstMeaningfulPaint:

https://build.chromium.org/p/tryserver.chromium.perf/builders/linux_perf_bisect/builds/7762 ("HTML Results" link)

So, I don't know what's going on here but I don't really trust the bisects.
Project Member

Comment 38 by 42576172...@developer.gserviceaccount.com, Nov 17 2017 

📍 Couldn't reproduce a difference.
https://pinpoint-dot-chromeperf.appspot.com/job/1488855df80000
Project Member

Comment 40 by 42576172...@developer.gserviceaccount.com, Nov 17 2017 

📍 Couldn't reproduce a difference.
https://pinpoint-dot-chromeperf.appspot.com/job/168a9e8df80000
Project Member

Comment 41 by bugdroid1@chromium.org, Nov 18 2017 

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/25cf0b97c844641e778fd1834b19489347eb1ea3

commit 25cf0b97c844641e778fd1834b19489347eb1ea3
Author: David Bokan <bokan@chromium.org>
Date: Sat Nov 18 01:36:36 2017

Revert "Make document inherit bg style from rootScroller"

This reverts commit 55c943bf50024b941adb671a026f021810119732. The bots
are convinced it's repsonsible for the linked perf regression, though it
doesn't make much sense to me and couldn't reproduce locally.

This patch has also caused lifecycle issues and isn't urgent so I'll
back it out for now and consider better ways to do this.

Bug:  742339 
Change-Id: I271c5fa31512beefc74dcb1936f570edf1cc1175
Reviewed-on: https://chromium-review.googlesource.com/777040
Reviewed-by: Steve Kobes <skobes@chromium.org>
Commit-Queue: David Bokan <bokan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#517669}
[modify] https://crrev.com/25cf0b97c844641e778fd1834b19489347eb1ea3/third_party/WebKit/LayoutTests/FlagExpectations/root-layer-scrolls
[delete] https://crrev.com/720d417bf1858ac64ff6f3ea8bbe2791aa065744/third_party/WebKit/LayoutTests/virtual/android/fast/rootscroller/browser-controls-background-expected.html
[delete] https://crrev.com/720d417bf1858ac64ff6f3ea8bbe2791aa065744/third_party/WebKit/LayoutTests/virtual/android/fast/rootscroller/browser-controls-background-iframe-expected.html
[delete] https://crrev.com/720d417bf1858ac64ff6f3ea8bbe2791aa065744/third_party/WebKit/LayoutTests/virtual/android/fast/rootscroller/browser-controls-background-iframe-scroller-expected.html
[delete] https://crrev.com/720d417bf1858ac64ff6f3ea8bbe2791aa065744/third_party/WebKit/LayoutTests/virtual/android/fast/rootscroller/browser-controls-background-iframe-scroller.html
[delete] https://crrev.com/720d417bf1858ac64ff6f3ea8bbe2791aa065744/third_party/WebKit/LayoutTests/virtual/android/fast/rootscroller/browser-controls-background-iframe.html
[delete] https://crrev.com/720d417bf1858ac64ff6f3ea8bbe2791aa065744/third_party/WebKit/LayoutTests/virtual/android/fast/rootscroller/browser-controls-background-modified-expected.html
[delete] https://crrev.com/720d417bf1858ac64ff6f3ea8bbe2791aa065744/third_party/WebKit/LayoutTests/virtual/android/fast/rootscroller/browser-controls-background-modified.html
[delete] https://crrev.com/720d417bf1858ac64ff6f3ea8bbe2791aa065744/third_party/WebKit/LayoutTests/virtual/android/fast/rootscroller/browser-controls-background.html
[delete] https://crrev.com/720d417bf1858ac64ff6f3ea8bbe2791aa065744/third_party/WebKit/LayoutTests/virtual/android/fast/rootscroller/browser-controls-gradient-background-iframe-expected.html
[delete] https://crrev.com/720d417bf1858ac64ff6f3ea8bbe2791aa065744/third_party/WebKit/LayoutTests/virtual/android/fast/rootscroller/browser-controls-gradient-background-iframe-scroller.html
[delete] https://crrev.com/720d417bf1858ac64ff6f3ea8bbe2791aa065744/third_party/WebKit/LayoutTests/virtual/android/fast/rootscroller/browser-controls-gradient-background-iframe.html
[delete] https://crrev.com/720d417bf1858ac64ff6f3ea8bbe2791aa065744/third_party/WebKit/LayoutTests/virtual/android/fast/rootscroller/browser-controls-gradient-background.html
[modify] https://crrev.com/25cf0b97c844641e778fd1834b19489347eb1ea3/third_party/WebKit/LayoutTests/virtual/android/fast/rootscroller/nested-rootscroller-browser-controls-bounds-shown-expected.html
[modify] https://crrev.com/25cf0b97c844641e778fd1834b19489347eb1ea3/third_party/WebKit/Source/core/dom/Document.cpp
[modify] https://crrev.com/25cf0b97c844641e778fd1834b19489347eb1ea3/third_party/WebKit/Source/core/frame/LocalFrameView.cpp
[modify] https://crrev.com/25cf0b97c844641e778fd1834b19489347eb1ea3/third_party/WebKit/Source/core/page/scrolling/RootScrollerController.cpp
[modify] https://crrev.com/25cf0b97c844641e778fd1834b19489347eb1ea3/third_party/WebKit/Source/core/page/scrolling/RootScrollerController.h
[modify] https://crrev.com/25cf0b97c844641e778fd1834b19489347eb1ea3/third_party/WebKit/Source/core/page/scrolling/RootScrollerTest.cpp

Comment 42 by bokan@chromium.org, Nov 20 2017

Status: Fixed (was: Assigned)
I reverted the blamed patch. Can't tell if that actually helped though since the graphs don't have any data since the end of September.
►

Sign in to add a comment