1355051 - 0.48 - 0.97% compiler_metrics num_static_constructors (linux32, linux64, linux64-stylo) regression on push 6066c5ca8f705fe06004a1123678dd427905fb2b (Tue Mar 14 2017)

Status: RESOLVED WONTFIX

(Core :: XPCOM, defect, P4)

Description

[Ionuț Goldan [:igoldan]]

We have detected a build metrics regression from push 6066c5ca8f705fe06004a1123678dd427905fb2b. As author of one of the patches included in that push, we need your help to address this regression.

Regressions:

  1%  compiler_metrics num_static_constructors linux64-stylo opt      103 -> 104
  1%  compiler_metrics num_static_constructors linux32 pgo            103.08 -> 104
  1%  compiler_metrics num_static_constructors linux64 pgo            103.08 -> 104
  1%  compiler_metrics num_static_constructors linux32 opt            103.08 -> 104
  1%  compiler_metrics num_static_constructors linux64-stylo debug    190 -> 191
  0.5%  compiler_metrics num_static_constructors linux32 debug          190.08 -> 191


You can find links to graphs and comparison views for each of the above tests at: https://treeherder.mozilla.org/perf.html#/alerts?id=5382

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 jobs in a pushlog format.

To learn more about the regressing test(s), please see: https://developer.mozilla.org/en-US/docs/Mozilla/Performance/Automated_Performance_Testing_and_Sheriffing/Build_Metrics

Comment 1

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

as a note, we recently shifted our threshold to be last least 3 num_static_constructors to generate an alert, please use this as good information, but maybe not a requirement to do anything.

Comment 2

[(Away)]

Does the automation have the ability to give me the name of the new constructor that appeared, or at least a list before and after the patch?

I'm very surprised that there was an _increase_ in constructors on Linux. On Windows at least, bug 1344629 actually _removed_ several dozen constructors, which was one of the main motivations for making that change.

Comment 3

[(Away)]

For comment 2.

Comment 4

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

I don't believe there is a way to get the actual list of constructors, lets see if :froydnj knows of a trick or two.  As a note, this is an increase of 1 which is not too worrisome.

Comment 5

[Nathan Froyd [:froydnj]]

You can do something like:

readelf -sW libxul.so |grep GLOBAL__sub | awk '{print $8;}'

to get a (rough) list static constructors.  Like Joel said, an increase of one, particularly for a change like this, is not too worrisome.

Comment 6

[(Away)]

$ diff before.txt after.txt 
94a95
> _GLOBAL__sub_I_Unified_cpp_xpfe_appshell0.cpp

So pretty surely from https://hg.mozilla.org/integration/autoland/rev/a06239255301.

Anyway, I was more concerned with the fact that my expectations weren't being met than with the new constructor itself. I agree that this is not too worrisome.