Closed Bug 1473008 Opened 6 years ago Closed 6 years ago

2.01 - 2.82% sessionrestore (linux64) regression on push 34f927a06700696cb78789e9e77770054556212c (Sun Jul 1 2018)

Categories

(Core :: Internationalization, defect, P1)

x86_64
Linux
defect

Tracking

()

RESOLVED WONTFIX

People

(Reporter: igoldan, Unassigned)

References

Details

(Keywords: perf, regression, talos-regression)

Talos has detected a Firefox performance regression from push:

https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?changeset=34f927a06700696cb78789e9e77770054556212c

As author of one of the patches included in that push, we need your help to address this regression.

Regressions:

  3%  sessionrestore linux64 pgo e10s stylo     255.47 -> 262.67
  2%  sessionrestore linux64 opt e10s stylo     269.75 -> 275.17


You can find links to graphs and comparison views for each of the above tests at: https://treeherder.mozilla.org/perf.html#/alerts?id=14139

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/Buildbot/Talos/Tests

For information on reproducing and debugging the regression, either on try or locally, see: https://wiki.mozilla.org/Buildbot/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/Buildbot/Talos/RegressionBugsHandling
Product: Testing → Core
:kmag please state which of the bugs bellow is more related to this perf regression. Also, can we fix it or should we accept it?

Bugs from patch:
bug 1470771
bug 1470716
bug 1470608
bug 1470365
bug 1463587
Flags: needinfo?(kmaglione+bmo)
It's probably bug 1470365. That's somewhat surprising, since the extra overhead that adds is supposed to be done in idle slices. But we're not very good at executing idle tasks in appropriately sized slices for native code, so that might be the issue.

It's interesting that this only affects Linux. My guess would be that it has something to do with the way we allocate shared memory there.

I'll look into this and see if there's anything that can be done to improve it. But, if not, given that this should actually improve content process startup responsiveness, seems to only affect Linux, and dramatically decreases content process memory usage, I think it would be an acceptable regression if we can't fix it.
Flags: needinfo?(kmaglione+bmo)
Moving to the same component as bug 1470365.
Component: General → Internationalization
Priority: -- → P1
Hm. So, that profile shows about 1ms of extra overhead from the new code. Given that we're only talking about a 6ms regression, there's probably not all that much I can do to improve it, aside from improving our scheduling of idle runnables.

So I'd lean towards closing this as WONTFIX and filing a new bug for that.
(In reply to Kris Maglione [:kmag] from comment #5)
> Hm. So, that profile shows about 1ms of extra overhead from the new code.
> Given that we're only talking about a 6ms regression, there's probably not
> all that much I can do to improve it, aside from improving our scheduling of
> idle runnables.
> 
> So I'd lean towards closing this as WONTFIX and filing a new bug for that.

:kmag did you file a bug for improving the scheduling of idle runnables?
Flags: needinfo?(kmaglione+bmo)
:m_kato are you aware of any updates here?
Flags: needinfo?(m_kato)
(In reply to Ionuț Goldan [:igoldan], Performance Sheriffing from comment #7)
> :m_kato are you aware of any updates here?

Since we use shared memory instead of per process memory, I think that we should be acceptable regression at this time.  I recommend that we should be wontfix at this bug as regression and filing a new bug for the feature.
Flags: needinfo?(m_kato)
Status: NEW → RESOLVED
Closed: 6 years ago
Resolution: --- → WONTFIX
Flags: needinfo?(kmaglione+bmo)
You need to log in before you can comment on or make changes to this bug.