2.2% ts_paint_webext (windows7-32-shippable) regression on push ad38bcbe7879cf715df2a49840b62d1235689882
Categories
(Firefox :: General, defect, P1)
Tracking
()
Tracking | Status | |
---|---|---|
firefox-esr60 | --- | unaffected |
firefox67 | --- | unaffected |
firefox67.0.1 | --- | unaffected |
firefox68 | --- | unaffected |
firefox69 | --- | fixed |
People
(Reporter: igoldan, Assigned: bdahl)
References
(Depends on 1 open bug, Regression)
Details
(Keywords: perf, regression, talos-regression)
Attachments
(1 file)
Talos has detected a Firefox performance regression from push:
As author of one of the patches included in that push, we need your help to address this regression.
Regressions:
2% ts_paint_webext windows7-32-shippable opt e10s stylo 326.42 -> 333.59
You can find links to graphs and comparison views for each of the above tests at: https://treeherder.mozilla.org/perf.html#/alerts?id=21112
On the page above you can see an 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(s), please see: https://wiki.mozilla.org/Performance_sheriffing/Talos/Tests
For information on reproducing and debugging the regression, either on try or locally, see: https://wiki.mozilla.org/Performance_sheriffing/Talos/Running
*** Please let us know your plans within 3 business days, or the offending patch(es) will be backed out! ***
Our wiki page outlines the common responses and expectations: https://wiki.mozilla.org/Performance_sheriffing/Talos/RegressionBugsHandling
Reporter | ||
Updated•6 years ago
|
Reporter | ||
Updated•6 years ago
|
Reporter | ||
Updated•6 years ago
|
Reporter | ||
Updated•6 years ago
|
Comment 1•6 years ago
|
||
Were other platforms affected as well, or is this only win7-32?
Comment 2•6 years ago
|
||
What's the difference between ts_paint_webext and ts_paint? I don't see it listed in https://wiki.mozilla.org/Buildbot/Talos/Tests#ts_paint.
Reporter | ||
Comment 3•6 years ago
|
||
(In reply to Brian Grinstead [:bgrins] from comment #1)
Were other platforms affected as well, or is this only win7-32?
Only Windows platforms got affected. This includes Windows 10.
Reporter | ||
Updated•6 years ago
|
Reporter | ||
Comment 4•6 years ago
|
||
(In reply to Brian Grinstead [:bgrins] from comment #2)
What's the difference between ts_paint_webext and ts_paint? I don't see it listed in https://wiki.mozilla.org/Buildbot/Talos/Tests#ts_paint.
:kmag can answer that.
Comment 5•6 years ago
|
||
(In reply to Brian Grinstead [:bgrins] from comment #2)
What's the difference between ts_paint_webext and ts_paint? I don't see it listed in https://wiki.mozilla.org/Buildbot/Talos/Tests#ts_paint.
ts_paint_webext has an extra extension installed that's meant to test overhead from the extension framework.
Comment 6•6 years ago
|
||
It's not clear to me what in particular about browser.xhtml + the extension framework would trigger that's slower. The document load should be the same (using prototype cache), and ts_paint itself is fine. Maybe the extension framework is special casing XULDocument somehow, though I'm not seeing anything obvious at https://searchfox.org/mozilla-central/search?q=xul&path=toolkit%2Fcomponents%2Fextensions.
Comment 7•6 years ago
|
||
Brendan, do you have any ideas on the next step for debugging this? I can do some try pushes to gather profiles with and without the change but not sure where to go besides that.
Comment 8•6 years ago
|
||
If I had to guess, I'd say the most likely candidates for being affected by browser.xhtml would be the browser action and page action code. I'm not sure exactly why it would make a difference, though.
The sheriffs should be able to get you before and after profiles.
Comment 9•6 years ago
|
||
Pushes that should get profiles (gathered via ./mach try fuzzy -q 'win talos-g5' --gecko-profile --no-artifact
):
Comment 10•6 years ago
|
||
(In reply to Brian Grinstead [:bgrins] from comment #9)
Pushes that should get profiles (gathered via
./mach try fuzzy -q 'win talos-g5' --gecko-profile --no-artifact
):
Hm, looks like these didn't work. Ionuț, could you help get profiles before/after the regression (and let me know how you do it so I can do it next time)?
Updated•6 years ago
|
Comment 11•6 years ago
|
||
A couple of pushes from today:
- There's a low confidence improvement by using a sync domcontentloaded event: https://treeherder.mozilla.org/perf.html#/compare?originalProject=try&originalRevision=5cf85a61738a894836ecf791d2721a1828b6112e&newProject=try&newRevision=6de35e9b14c741cfc693804e942fea70075ef34f&framework=1&showOnlyImportant=1
- No change when using dummy.xhtml instead of dummy.xul: https://treeherder.mozilla.org/perf.html#/compare?originalProject=try&originalRevision=5cf85a61738a894836ecf791d2721a1828b6112e&newProject=try&newRevision=c8a70836428841df0ee6fde55620e0797e5e8827&framework=1&showOnlyImportant=1
Reporter | ||
Comment 12•6 years ago
|
||
(In reply to Brian Grinstead [:bgrins] from comment #7)
Brendan, do you have any ideas on the next step for debugging this? I can do some try pushes to gather profiles with and without the change but not sure where to go besides that.
I tried to run the profiles, but I get permafails on Windows.
Assignee | ||
Comment 13•6 years ago
|
||
Some of the talos test timers seem to be affected when DOMContentLoaded
is not dispatched synchronously. This will also help when we load all XUL
as XHTML (bug 1550801), since certain tests rely on sync dispatch.
Comment 14•6 years ago
|
||
Assignee | ||
Updated•6 years ago
|
![]() |
||
Comment 15•6 years ago
|
||
bugherder |
Updated•6 years ago
|
Reporter | ||
Updated•6 years ago
|
Updated•6 years ago
|
Updated•4 years ago
|
Description
•