368014 - 100% CPU at pokerevaluation.com/

Status: RESOLVED INCOMPLETE

(Core :: General, defect)

Description

[Steve Chapel]

Reproducible: Always

Steps to reproduce:
1. Go to http://www.pokerevaluation.com/

Expected results: CPU usage less than 100%
Actual results: 100% CPU

Firefox 2.0.0.1, SeaMonkey 1.1, and the 20070122 trunk build of Minefield all give 100% CPU on the URL given. Opera 9.01 uses 12-25% CPU and IE uses 0% CPU. This looks like some sort of DHTML performance problem.

Comment 1

[Mats Palmgren (inactive)]

High CPU also with Firefox 2.0.0.1 on Linux

Comment 2

[Ver]

Same here 98% CPU, Firefox 2.0.0.1 on Kubuntu, when I close the tab with that URL the CPU goes back to <1%  Unlike other posters in the forums, it does not seem to slow down any other applications.

Also CPU usage is high in the morning if I leave Firefox running overnight, only way to "fix" it the is to restart Firefox and re-opening the same pages.  I am wondering if this could have some thing to do with the search bar?

So I am guessing that there could be two (or more) different problems. I have removed the search bar and will update if that seems to fix the overnight problem.

Comment 3

[Oleg Romashin (:romaxa)]

There are permanent Restyle Reflow ....

nsCSSFrameConstructor::RestyleEvent::Run  happens always.

................................
#104 0xb61f007f in nsFrameManager::ComputeStyleChangeFor (this=0x83f59d4, aFrame=0x88e2880, aChangeList=0xbfde19f8, aMinChange=0)
    at layout/base/nsFrameManager.cpp:1435
#105 0xb61bbda3 in nsCSSFrameConstructor::RestyleElement (this=0x83f6cd0, aContent=0x850e6e0, aPrimaryFrame=0x88e2880, aMinHint=0)
    at layout/base/nsCSSFrameConstructor.cpp:9927
#106 0xb61bbf5a in nsCSSFrameConstructor::ProcessOneRestyle (this=0x83f6cd0, aContent=0x850e6e0, aRestyleHint=eReStyle_Self, aChangeHint=0)
    at layout/base/nsCSSFrameConstructor.cpp:12796
#107 0xb61bc187 in nsCSSFrameConstructor::ProcessPendingRestyles (this=0x83f6cd0)
    at layout/base/nsCSSFrameConstructor.cpp:12849
#108 0xb61bc292 in nsCSSFrameConstructor::RestyleEvent::Run (this=0x850d280)
    at layout/base/nsCSSFrameConstructor.cpp:12912
#109 0xb6cc06a1 in nsThread::ProcessNextEvent (this=0x80b2950, mayWait=0, result=0xbfde1c48)
    at xpcom/threads/nsThread.cpp:482
#110 0xb6c69216 in NS_ProcessPendingEvents_P (thread=0x80b2950, timeout=20) at nsThreadUtils.cpp:179
#111 0xb6ba7261 in nsBaseAppShell::NativeEventCallback (this=0x82c5d00)
    at widget/src/xpwidgets/nsBaseAppShell.cpp:111
#112 0xb6b8f868 in nsAppShell::EventProcessorCallback (source=0x82c5d40, condition=G_IO_IN, data=0x82c5d00)
    at widget/src/gtk2/nsAppShell.cpp:69
#113 0xb78eec7f in g_io_channel_unix_get_fd () from /usr/lib/libglib-2.0.so.0
#114 0xb78c5731 in g_main_context_dispatch () from /usr/lib/libglib-2.0.so.0
#115 0xb78c87a6 in g_main_context_check () from /usr/lib/libglib-2.0.so.0
#116 0xb78c8b67 in g_main_loop_run () from /usr/lib/libglib-2.0.so.0
#117 0xb7c6c281 in gtk_main () from /usr/lib/libgtk-x11-2.0.so.0
#118 0x0804c563 in main (argc=138263944, argv=0x1)

Comment 4

[Carlos Sanabria]

Tried http://www.pokerevaluation.com/ on Camino Version 2007022813 (1.0.4) running on Mac OS X 10.4.9 with the same results.

Comment 5

[Eli Friedman]

The reason style resolution is really expensive here is because of residual style handling.  That leads to text nested hundreds of content nodes deep, all of which must be re-resolved every time the style on the automatic scroller changes.  Reflow is also pretty expensive; I'd need to profile to see exactly where we're spending that time.

On a side note, it seems like we could save a lot of memory in cases like this by somehow sharing mapped attributes.

Comment 6

[RNicoletto]

Confirmed!

Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.0.1pre) Gecko/2008070106 GranParadiso/3.0.1pre
Shockwave Flash 9.0 r124

Comment 7

[tom]

In using this URL, I get 100% CPU on my IBM T43 laptop running XP Pro SP3... http://www.nasdaq.com/?view=mynasdaq&selected=QCOM  (I have 10 stocks listed)  Safari 3.0 and IE 7.0143 both use less that %20 on the same site on the same computer...  its a real pain...

firefox 3.0.3

Comment 8

[RNicoletto]

(In reply to comment #7)
> In using this URL, I get 100% CPU on my IBM T43 laptop running XP Pro SP3...
> http://www.nasdaq.com/?view=mynasdaq&selected=QCOM  (I have 10 stocks listed) 
> Safari 3.0 and IE 7.0143 both use less that %20 on the same site on the same
> computer...  its a real pain...
> 
> firefox 3.0.3

WFM - DELL LATITUDE D505 running XP PRO SP3 - I have 8 stocks listed

Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.0.3) Gecko/2008092417 Firefox/3.0.3

Comment 9

[RNicoletto]

Confirming high CPU usage on http://www.pokerevaluation.com/ in Mozilla Firefox 3.5 Beta 4 on Windows XP.

Comment 10

[Wayne Mery (:wsmwk)]

Testcase URL is no longer good