1142183 - ~8MiB AWSY regression due to enabling desktop reader on nightly

Status: RESOLVED FIXED

(Toolkit :: Reader Mode, defect)

Description

[Eric Rahm [:erahm]]

+++ This bug was initially created as a clone of Bug #1140045 +++

areweslimyet.com indicates an 8MiB regression on 2015/03/10 corresponding with desktop reader being enabled by default on nightly. A diff of memory reports clearly indicates the reader worker is involved:

>7.65 MB (100.0%) -- explicit
>├──3.92 MB (51.31%) ++ workers/workers(chrome)/worker(resource://gre/modules/reader/ReaderWorker.js, 0xNNN)
>├──3.37 MB (44.01%) ── heap-unclassified
>├──1.39 MB (18.20%) -- js-non-window
>│  ├──1.11 MB (14.56%) -- runtime
>│  │  ├──0.99 MB (12.92%) -- gc
>│  │  │  ├──1.00 MB (13.08%) ── nursery-committed

We also see a rather large increase in heap-unclassified which seems to indicate there is a significant portion desktop reader that is not being measured by a memory reporter.

Comment 1

[Eric Rahm [:erahm]]

Attachment: memory report after regression

Comment 2

[Eric Rahm [:erahm]]

Attachment: memory before regression

Comment 3

[:Margaret Leibovic]

I am working on a fix in bug 1139678 to remove the background readability parsing we're currently doing. That will likely fix this regression.

Comment 4

[:Margaret Leibovic]

How can I check to see if my patch from bug 1139678 fixed this?

Comment 5

[Eric Rahm [:erahm]]

(In reply to :Margaret Leibovic from comment #4)
> How can I check to see if my patch from bug 1139678 fixed this?

Once it lands on m-i we'll automatically test it, if you're eager to see results you can have AWSY run against try pushes as detailed here: https://groups.google.com/forum/#!searchin/mozilla.dev.platform/areweslimyet/mozilla.dev.platform/XOuPqSpgIJw/bo9QIaQTfvUJ

Comment 6

[Nicholas Nethercote [inactive]]

> Once it lands on m-i we'll automatically test it, if you're eager to see
> results you can have AWSY run against try pushes as detailed here:
> https://groups.google.com/forum/#!searchin/mozilla.dev.platform/areweslimyet/
> mozilla.dev.platform/XOuPqSpgIJw/bo9QIaQTfvUJ

Can you update the docs at
https://areweslimyet.com/faq.htm#how-can-i-request-additional-awsy-tests-on-specific-changesets-or-try-pushes to include the new info? Thank you.

Comment 7

[Eric Rahm [:erahm]]

(In reply to Nicholas Nethercote [:njn] from comment #6)
> > Once it lands on m-i we'll automatically test it, if you're eager to see
> > results you can have AWSY run against try pushes as detailed here:
> > https://groups.google.com/forum/#!searchin/mozilla.dev.platform/areweslimyet/
> > mozilla.dev.platform/XOuPqSpgIJw/bo9QIaQTfvUJ
> 
> Can you update the docs at
> https://areweslimyet.com/faq.htm#how-can-i-request-additional-awsy-tests-on-
> specific-changesets-or-try-pushes to include the new info? Thank you.

Yes, I keep forgetting to update the faq!

Comment 8

[:Margaret Leibovic]

(In reply to Eric Rahm [:erahm] from comment #5)
> (In reply to :Margaret Leibovic from comment #4)
> > How can I check to see if my patch from bug 1139678 fixed this?
> 
> Once it lands on m-i we'll automatically test it, if you're eager to see
> results you can have AWSY run against try pushes as detailed here:
> https://groups.google.com/forum/#!searchin/mozilla.dev.platform/areweslimyet/
> mozilla.dev.platform/XOuPqSpgIJw/bo9QIaQTfvUJ

I didn't get around to running this myself, but I'm wondering if you were able to automatically test this when it landed on m-i. Just wondering if more fixes are needed here.

Comment 9

[Eric Rahm [:erahm]]

(In reply to :Margaret Leibovic from comment #8)
> (In reply to Eric Rahm [:erahm] from comment #5)
> > (In reply to :Margaret Leibovic from comment #4)
> > > How can I check to see if my patch from bug 1139678 fixed this?
> > 
> > Once it lands on m-i we'll automatically test it, if you're eager to see
> > results you can have AWSY run against try pushes as detailed here:
> > https://groups.google.com/forum/#!searchin/mozilla.dev.platform/areweslimyet/
> > mozilla.dev.platform/XOuPqSpgIJw/bo9QIaQTfvUJ
> 
> I didn't get around to running this myself, but I'm wondering if you were
> able to automatically test this when it landed on m-i. Just wondering if
> more fixes are needed here.

I see a corresponding decrease around 3/17, I'd say we're good.

Comment 10

[:Margaret Leibovic]

(In reply to Eric Rahm [:erahm] from comment #9)
> (In reply to :Margaret Leibovic from comment #8)
> > (In reply to Eric Rahm [:erahm] from comment #5)
> > > (In reply to :Margaret Leibovic from comment #4)
> > > > How can I check to see if my patch from bug 1139678 fixed this?
> > > 
> > > Once it lands on m-i we'll automatically test it, if you're eager to see
> > > results you can have AWSY run against try pushes as detailed here:
> > > https://groups.google.com/forum/#!searchin/mozilla.dev.platform/areweslimyet/
> > > mozilla.dev.platform/XOuPqSpgIJw/bo9QIaQTfvUJ
> > 
> > I didn't get around to running this myself, but I'm wondering if you were
> > able to automatically test this when it landed on m-i. Just wondering if
> > more fixes are needed here.
> 
> I see a corresponding decrease around 3/17, I'd say we're good.

Hooray! I'll close this out then as fixed by bug 1139678. Thanks for your help!