637278 - Full CPU load as long as o2online.de is displayed with hardware acceleration disabled

Status: RESOLVED FIXED

(Core :: Graphics, defect)

Description

[Michael]

User-Agent:       Mozilla/5.0 (Windows NT 6.1; WOW64; rv:2.0b12) Gecko/20100101 Firefox/4.0b12
Build Identifier: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:2.0b12) Gecko/20100101 Firefox/4.0b12

If I display o2online.de the CPU load increases to MAX. If the tab is in background everything is fine.

Reproducible: Always

Steps to Reproduce:
1. open o2online.de with Firefox 4.0b12
Actual Results:  
CPU load is at 100% for one core

Expected Results:  
CPU load should decrease immediately after load (like it does using Firefox 4.0b11)

Comment 1

[Bebe]

This bug is also reproducible on WinXp 
Mozilla/5.0 (Windows NT 5.1; rv:2.0b13pre) Gecko/20110227 Firefox/4.0b13pre

Comment 2

[Bebe]

Regression range on WinXp: 
Last Working: 
Mozilla/5.0 (Windows NT 5.1; rv:2.0b12pre) Gecko/20110216 Firefox/4.0b12pre
http://hg.mozilla.org/mozilla-central/rev/8812253737ce

First Broken:
Mozilla/5.0 (Windows NT 5.1; rv:2.0b12pre) Gecko/20110217 Firefox/4.0b12pre
http://hg.mozilla.org/mozilla-central/rev/b4aa47ca42c1

http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=b4aa47ca42c1&tochange=8812253737ce

Comment 3

[Matthias Versen [:Matti]]

wfm with SM trunk and FF4.0b12 on win7

Does it work in the Firefox safemode: http://support.mozilla.com/en-US/kb/Safe+Mode ?

Comment 4

[Michael]

CPU load also increases in safe mode as well as with a completely new profile and also with new profile and safe mode. No differences for me.

Comment 5

[Matthias Versen [:Matti]]

Is your graphic acceleration activated ?
Please post the graphic section from about:support.

Comment 6

[Bebe]

Matthias  is your window Maximized because on XP is reproducible only if the ff window is maximized


Graphics
      
      
      
        
Adapter DescriptionIntel(R) 82865G Graphics Controller
Vendor ID8086
Device ID2572
Adapter RAM Unknown
Adapter Drivers ialmrnt5
Driver Version6.13.10.3510 
Driver Date4-15-2003
Direct2D Enabled false
DirectWrite Enabled false (0.0.0.0, font cache n/a)
WebGL Renderer(WebGL unavailable)G
PU Accelerated Windows0/1

Comment 7

[Michael]

Looks like Direct2D has been activated but blocked
and DirectWrite is disabled. Here the information:

Karten-Beschreibung NVIDIA Quadro FX 770M
Vendor-ID 10de
Geräte-ID 065c
Karten-Ram 512
Karten-Treiber nvd3dumx,nvwgf2umx,nvwgf2umx nvd3dum,nvwgf2um,nvwgf2um
Treiber-Version 8.15.11.8621
Treiber-Datum 6-16-2009
Direct2D aktiviert Wurde auf Grund Ihres Grafiktreibers blockiert. Versuchen Sie, Ihren Grafiktreiber auf mindestens Version 257.21 zu aktualisieren.
DirectWrite aktiviert false (6.1.7601.17514, font cache n/a)
WebGL-Renderer Google Inc. -- ANGLE -- OpenGL ES 2.0 (ANGLE 0.0.0.541)
GPU-beschleunigte Fenster 0/1

Comment 8

[Matthias Versen [:Matti]]

I suggested the safemode because I wanted to test Graphic acceleration, JIT and the addons (they are all disabled in the safemode). No difference in the safemode= can't be addons, jit is unlikely because I can't reproduce it -> your hardware acceleration must be disabled (or you have a different flash version).

I can confirm 25% CPU on my 4 Core system with Firefox 4.0b12, maximized window and hardware acceleration disabled.

Comment 9

[Michael]

This error also occurs on my private system (WinXP). On this machine Graphic acceleration is also not active.
Direct2D aktiviert false
DirectWrite aktiviert false (0.0.0.0, font cache n/a)

In difference to the other reports for me it also occurs when the window is not maximized, but only "large enough". So for a small window everything is fine, but as it gets large enough the high CPU load occurs (50% on a dual core system) although the window is not maximized.

Comment 10

[Mike Beltzner [:beltzner, not reading bugmail]]

Speculatively hard blocking as the regression range and description implies that this might be due to the recent changes to plugin compositing.

Comment 11

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

I can also reproduce on win7 with GPU rendering off, on the latest nightly.  Also happens on my linux 64-bit desktop.

Comment 12

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

Confirmed that this is a regression from bug 626602.

To reproduce, the window doesn't need to be full-screen; just the header starting from the O2 image at the top-left of the page extending over to the "Suche" bar needs to be visible.  The CPU is spiking because we're constantly updating the background behind the plugin instance that implements the menu.  I'm not sure why we're doing that yet.

Comment 13

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

(Oops, lost this in the mid-air.)

Comment 14

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

Attachment: INVALIDATE_NO_UPDATE_LAYER_TREE should subsume INVALIDATE_NO_THEBES_LAYERS

We need to be able to test for NO_UPDATE_LAYER_TREE distinct from INVALIDATE_NO_THEBES, and there's only one unique user of NO_UPDATE_LAYER_TREE, so I went with the smaller patch to just fix that user.

The mac-specific plugin/layers code here had essentially duplicated nsIFrame::InvalidateLayer() for reasons I don't understand.  I went ahead and fixed that.

Comment 15

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

Attachment: Fix indentation

This whitespace fix obscured the previous patch, so I split it out.

Comment 16

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Comment on attachment 515836 [details] [diff] [review]
INVALIDATE_NO_UPDATE_LAYER_TREE should subsume INVALIDATE_NO_THEBES_LAYERS

This is a no-brainer.

Comment 17

[[e-mail]]

On Ubuntu 10.10 it also happens.

Comment 18

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

http://hg.mozilla.org/mozilla-central/rev/88074e563627
http://hg.mozilla.org/mozilla-central/rev/52ffeed74f96