1240180 - 2-8% a11y/tart/tp5o/tresize windows* regression on Inbound (v.46) on Jan 13, from push a9f9b36c1a2e

Status: RESOLVED FIXED

(Core :: Widget: Win32, defect)

Description

[Joel Maher ( :jmaher ) (UTC -8)]

Talos has detected a Firefox performance regression from your commit a9f9b36c1a2eec7626e6b749e46ab0a8bf3323e2 in bug 1239007.  We need you to address this regression.

This is a list of all known regressions and improvements related to your bug:
http://alertmanager.allizom.org:8080/alerts.html?rev=a9f9b36c1a2eec7626e6b749e46ab0a8bf3323e2&showAll=1

On the page above you can see Talos alert for each affected platform as well as a link to a graph showing the history of scores for this test. There is also a link to a treeherder page showing the Talos jobs in a pushlog format.

To learn more about the regressing test, please see:
https://wiki.mozilla.org/Buildbot/Talos/Tests#a11y

Reproducing and debugging the regression:
If you would like to re-run this Talos test on a potential fix, use try with the following syntax:
try: -b o -p win32 -u none -t other,svgr,tp5o  # add "mozharness: --spsProfile" to generate profile data

To run the test locally and do a more in-depth investigation, first set up a local Talos environment:
https://wiki.mozilla.org/Buildbot/Talos/Running#Running_locally_-_Source_Code

Then run the following command from the directory where you set up Talos:
talos --develop -e <path>/firefox -a a11yr,tart,tp5o,tresize

Making a decision:
As the patch author we need your feedback to help us handle this regression.
*** Please let us know your plans by Tuesday, or the offending patch will be backed out! ***

Our wiki page outlines the common responses and expectations:
https://wiki.mozilla.org/Buildbot/Talos/RegressionBugsHandling

Comment 1

[Joel Maher ( :jmaher ) (UTC -8)]

a11y regressions:
https://treeherder.allizom.org/perf.html#/compare?originalProject=mozilla-inbound&originalRevision=98756a36223c&newProject=mozilla-inbound&newRevision=a9f9b36c1a2e&framework=1&filter=a11y

tart regressions:
https://treeherder.allizom.org/perf.html#/compare?originalProject=mozilla-inbound&originalRevision=98756a36223c&newProject=mozilla-inbound&newRevision=a9f9b36c1a2e&framework=1&filter=tart

tp5o (not scroll or bytes);
https://treeherder.allizom.org/perf.html#/compare?originalProject=mozilla-inbound&originalRevision=98756a36223c&newProject=mozilla-inbound&newRevision=a9f9b36c1a2e&framework=1&filter=tp5o%20opt

tresize (appears to be win7 only):
https://treeherder.allizom.org/perf.html#/compare?originalProject=mozilla-inbound&originalRevision=98756a36223c&newProject=mozilla-inbound&newRevision=a9f9b36c1a2e&framework=1&filter=tresize%20e10s

I did a lot of retriggers on the original push and the previous push to give me confidence in this assessment:
https://treeherder.mozilla.org/#/jobs?repo=mozilla-inbound&fromchange=6cc42896216e&tochange=418796e4d8de&filter-searchStr=Windows%207%2032-bit%20mozilla-inbound%20talos%20other

I would be happy to bisect on try (I have a script for that) and find which part is causing the problems, or maybe help try out some patches.

:jfkthame, can you take ownership of this and confirm you are working on this or this is an expected regression and we need to live with it because <fill in the blank> by Tuesday?

Comment 2

[Jonathan Kew [:jfkthame]]

Yes, I'll be looking into this ASAP.

If you can bisect this down to a specific one of the changesets that landed in bug 890156, that would be really great -- thanks!

Comment 3

[Joel Maher ( :jmaher ) (UTC -8)]

ok, I have a bunch of pushes:
https://treeherder.mozilla.org/#/jobs?repo=try&author=[e-mail]&fromchange=2dc5bfb75ce0&tochange=8fc9b399c0e6

Comment 4

[Joel Maher ( :jmaher ) (UTC -8)]

a11y and tp5o made a big change here:
http://hg.mozilla.org/integration/mozilla-inbound/rev/c35bafe04d6f

I don't see one for tart, maybe that was noise that i mistook for a regression.  I didn't run tresize, but I would be happy to add that in if you would like.  It is pretty easy to spot the difference in a11y and tp5o.

Comment 5

[Jonathan Kew [:jfkthame]]

Attachment: Optimize native theme scaling for the single-monitor case

AFAICS this should fix the talos regressions here: for the common single-display case, the need to scale theme metrics for a secondary display will not arise so we can skip querying the displays and computing the scale factor. Try results, compared to pre-bug 890156: https://treeherder.allizom.org/perf.html#/compare?originalProject=mozilla-inbound&originalRevision=98756a36223c&newProject=try&newRevision=7fa4ada72e6c&framework=1.

Comment 6

[Joel Maher ( :jmaher ) (UTC -8)]

thanks for getting a patch up so quickly!

Comment 7

[Jonathan Kew [:jfkthame]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/ee1922a13d8fe00ead0a87c0c95cf6a064043d25
Bug 1240180 - Optimize native theme scaling for the single-monitor case. r=emk

Comment 8

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/ee1922a13d8f

Comment 9

[Joel Maher ( :jmaher ) (UTC -8)]

looking at the graphs, I see a lot of improvements.  Regarding a11y, these are improved, but about 75%, same for tp5o.  tart/tresize looks to be 100% recovered.

Overall, I think this is a win.  Thanks for fixing this!

Comment 10

[Jonathan Kew [:jfkthame]]

Backed out of aurora-46, see bug 890156 comment 168. Bumping target milestone to 47.