831123 - Tp regression on Jan 11th

Status: RESOLVED FIXED

(Firefox for Android Graveyard :: General, defect)

Description

[Mark Finkle (:mfinkle) (use needinfo?)]

Looks like we have a small Tp regression on Jan 11th:
http://graphs.mozilla.org/graph.html#tests=[[84,63,20]]&sel=1357694590490.722,1357998888966.0583&displayrange=7&datatype=running

Range looks to be:
http://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=4cb44c4e4629&tochange=e523e9cd94dc

Comment 1

[Mark Finkle (:mfinkle) (use needinfo?)]

I think this might be coming from:
http://hg.mozilla.org/integration/mozilla-inbound/rev/028ce4b36b4e

Only because we send error output to the Android logcat, which has a measurable affect on performance. I made some Try runs with the CSS error reporting turned off and found Tp was faster:
https://tbpl.mozilla.org/?tree=Try&rev=3df9fa8c6692

Since there is no easy way to see the errors, I suggest we disable outputting CSS errors until we add a dev UI for toggling them on/off.

Comment 2

[Mark Finkle (:mfinkle) (use needinfo?)]

Attachment: patch

Disable CSS error reporting by default

Comment 3

[Mark Finkle (:mfinkle) (use needinfo?)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/ec5df8e5a755

Comment 4

[Marco Bonardo [:mak]]

backed out for test failures... not pasting them cause "Only displaying first 20 of 100+ failures"
https://tbpl.mozilla.org/php/getParsedLog.php?id=18896131&tree=Mozilla-Inbound

https://hg.mozilla.org/integration/mozilla-inbound/rev/f7ae1bf4479a

Comment 5

[Mark Finkle (:mfinkle) (use needinfo?)]

Attachment: patch 2

Same as previous patch, but adds fixes to the two tests that failed. I use SpecialPowers.pushPrefEnv to temporarily set the pref for the test. Try run is green now:
https://tbpl.mozilla.org/?tree=Try&rev=697b280e4c0b

Comment 6

[Joel Maher ( :jmaher ) (UTC -0800)]

Comment on attachment 703627 [details] [diff] [review]
patch 2

Review of attachment 703627 [details] [diff] [review]:
-----------------------------------------------------------------

looks great, thanks for using these specialpowers functions.

Comment 7

[Mark Finkle (:mfinkle) (use needinfo?)]

Attachment: alt patch

I was not able to figure out a way to fix test_bug513194.html:
http://mxr.mozilla.org/mozilla-central/source/content/base/test/test_bug513194.html?force=1

So, I did a quickie fix and turned on the pref for mochitests using the pre-seed file. The pref is not pre-seeded for talos, so we can measure the impact on Tp4.

Comment 8

[Mark Finkle (:mfinkle) (use needinfo?)]

Try Server run with green M1 and M8 runs. Tp4 is improved.

Comment 9

[Mark Finkle (:mfinkle) (use needinfo?)]

(In reply to Mark Finkle (:mfinkle) from comment #8)
> Try Server run with green M1 and M8 runs. Tp4 is improved.

The link would be handy:
https://tbpl.mozilla.org/?tree=Try&rev=86b551105152

Comment 10

[Joel Maher ( :jmaher ) (UTC -0800)]

Comment on attachment 704601 [details] [diff] [review]
alt patch

Review of attachment 704601 [details] [diff] [review]:
-----------------------------------------------------------------

r=me if we can get a bug on file to fix the two tests needed to not hack this setting on.

Comment 11

[Mark Finkle (:mfinkle) (use needinfo?)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/f8cb1716f5d4

Comment 12

[(no longer active)]

https://hg.mozilla.org/mozilla-central/rev/f8cb1716f5d4