1554139 - 1.98 - 2.06% Base Content Heap Unclassified (linux64-shippable, linux64-shippable-qr) regression on push 61ae24d322cb05bec32cf7620778f15a2e6908e5 (Thu May 23 2019)

Status: RESOLVED WONTFIX

(Core :: General, defect, P1)

Description

[Ionuț Goldan [:igoldan]]

We have detected an awsy regression from push:

https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=ce4a146da49d57afadd9205edd251c32d5d0bd9a&tochange=61ae24d322cb05bec32cf7620778f15a2e6908e5

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

Regressions:

2% Base Content Heap Unclassified linux64-shippable opt 5,496,701.33 -> 5,609,914.00
2% Base Content Heap Unclassified linux64-shippable-qr opt 5,493,604.33 -> 5,602,432.33

You can find links to graphs and comparison views for each of the above tests at: https://treeherder.mozilla.org/perf.html#/alerts?id=21071

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://wiki.mozilla.org/AWSY/Tests

Comment 1

[Dzmitry Malyshau [:kvark]]

I don't think there is enough reason for us here to block the dependency updates (namely, rand and parking_lot), given that only Linux is negatively affected. I'll work with library authors to figure out if that can be fixed upstream, starting with https://github.com/Amanieu/parking_lot/issues/143

Comment 2

[Nhi Nguyen (:nhi)]

P1 due to it being a regression on Fx69.
:kvark, since the bug is assigned to you, feel free to adjust the priority.

Comment 3

[Dzmitry Malyshau [:kvark]]

Here are some of the tries I pushed to narrow down the issue:

• roll back u2fhid, mozurl, style, geckolib: 5480264 of unclassified memory (fixed)
• roll back u2fhid, style, geckolib: 5581808 unclassified mem (unfixed)
• roll back u2fhid: 5586992 unclassified mem (unfixed)
• roll back mozurl: pending

Comment 4

[Dzmitry Malyshau [:kvark]]

roll back mozurl: 5641224 unclassified mem (unfixed)

original: pending

Comment 5

[Dzmitry Malyshau [:kvark]]

original: 5584776 unclassified mem (unfixed), which leaves me puzzled a bit as to why the previous try (rolling back mozurl) shows more...

Comment 6

[Eric Rahm [:erahm]]

(In reply to Dzmitry Malyshau [:kvark] from comment #5)

original: 5584776 unclassified mem (unfixed), which leaves me puzzled a bit as to why the previous try (rolling back mozurl) shows more...

Are you comparing with the same base revision? As in starting w/ base revision A you do a try run w/ a few retriggers, then you apply the mozurl change (B) to test A + B, then you try just the u2fhid change (C) to test A + C, etc. Our fission memory wiki give a quick description of this workflow.

Comment 7

[Dzmitry Malyshau [:kvark]]

Yes, Eric, I'm working off the same revision in these tests

Comment 8

[Eric Rahm [:erahm]]

(In reply to Dzmitry Malyshau [:kvark] from comment #7)

Yes, Eric, I'm working off the same revision in these tests

Okay, but that's on top of your local equivalent of https://hg.mozilla.org/mozilla-central/rev/61ae24d322cb which is what this bug says regressed things (which might be what you intended). It might be worth starting totally clean.

Comment 9

[Dzmitry Malyshau [:kvark]]

Oh, good point! I'll start off a cleaner base.

Comment 10

[Dzmitry Malyshau [:kvark]]

Clean state patches based off rev 3f656afb0c08aae2ed9b2ebe4128bac71146c679 :

• roll back mozurl
• roll back parking_lot to 0.6
• roll back parking_lot to 0.7

(the parking lot rolls include mozurl changes)
Also filed https://github.com/rust-random/rand/issues/820 for Rand maintainers.

Comment 11

[Dzmitry Malyshau [:kvark]]

Wow, all of these tries show about the same result of 159xxxx on "Base Content Heap Unclassified opt", which means they aren't addressing the regression, hmm

Comment 12

[Patricia Lawless]

Hi Dzmitry, is this still on track for 69?

Comment 13

[Dzmitry Malyshau [:kvark]]

Patricia, we should move it to 70, it's not critical.

Comment 14

[BugBot [:suhaib / :marco/ :calixte]]

The bug assignee didn't login in Bugzilla in the last months and this bug has priority 'P1'.
:overholt, could you have a look please?
For more information, please visit auto_nag documentation.