769380 - [ARMv6] Socorro shows up blurry and incomplete (Readability check is slow)

Status: RESOLVED WORKSFORME

(Firefox for Android Graveyard :: Reader View, defect, P4)

Description

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Attachment: screenshot

1. go to http://crash-stats.mozilla.com/report/index/bp-0eae30c3-cc31-4287-b523-c6f5b2120628
2. pan and scroll to the bottom

Expected: website fully loads, the bottom is complete
Actual: see screenshot.  Website is hard to read and looks blurry, bottom is not complete.

Comment 1

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Attachment: logcat

Comment 2

[Aaron Train [:aaronmt]]

Phone, 2.2?

Comment 3

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Samsung GT-I5510 Android 2.2

Comment 4

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

Looks like the screenshot/checkerboard to me

Comment 5

[Kartikaya Gupta (email:[e-mail])]

I was seeing similar behaviour because of the readability heuristics check, see https://bugzilla.mozilla.org/show_bug.cgi?id=765930#c4

Comment 6

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

Naoki, does the burry content eventually get better after some time?

Comment 7

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

I will have to recheck.  Probably won't be able to check until after 3pm PDT today.

Comment 8

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Attachment: portrait

Screenshot of OOM

I guess I might have to log separate bugs.

Comment 9

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Ya.  It looks like the OOM is a separate issue.  The content does get better but only after the whole page loads and everything is calculated for readability.  5 mins at least for the page to load.

Comment 10

[Alex Keybl [:akeybl]]

Can we find other instances of this bug with pages outside of crash-stats? Adding qawanted.

Comment 11

[Kartikaya Gupta (email:[e-mail])]

Adding reader mode decorations to the bug and CC'ing some more people. These was a bug somewhere about optimizing the reader mode check but I can't find it now; this should probably be duped to that and this URL should be added to the list of URLs that need to be re-tested/profiled for improvement.

Comment 12

[Brian Nicholson (:bnicholson)]

Since this bug was filed, the reader mode check has been moved to a web worker. The check shouldn't block the main thread anymore, so this bug might be fixed (or, at least, the symptoms are probably different). Naoki, can you try this out again?

Comment 13

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

I don't have the device, brought it to mnt view.  kbrosnan will check for the device from tchung's desk and see if he can repro.

Comment 14

[Brad Lassey [:blassey] (use needinfo?)]

assumption is that this is resolved by moving the readability stuff to a web worker. Assigning to Kevin to confirm. Also, not ARMv6 specific.

Comment 15

[Kevin Brosnan [Ex-Mozilla]]

This is still an issue on today's nightly.

Comment 16

[Kevin Brosnan [Ex-Mozilla]]

Looking at this some more it seems screenshot related. Drawing https://crash-stats.mozilla.com/img/2.0/page-bg.png on the page seems to take several minuets.

Comment 17

[Brian Nicholson (:bnicholson)]

Here's a build with reader mode parsing disabled: http://dl.dropbox.com/u/35559547/fennec-no-reader.patch

That should help verify whether reader mode is even the culprit here.

Comment 18

[Brian Nicholson (:bnicholson)]

Actually, as of bug 787597, that page isn't even parsed for readability, though I don't know whether that patch was in the nightly from comment 15.

Comment 19

[Alex Keybl [:akeybl]]

We haven't been getting complaints about this from out in the wild. Doesn't seem to affect a significant portion of the web.

Comment 20

[Aaron Train [:aaronmt]]

WFM in re-testing, Beta 20.0

Comment 21

[Lucas Rocha (:lucasr)]

Closing. Feel free to re-open if you can still reproduce it.