581459 - cnn.com causes minefield to hit 100% CPU usage when using the web console

Status: RESOLVED INVALID

(DevTools :: General, defect)

Description

[David Dahl :ddahl]

This is when cnn is first loading. Sometimes, when I change urls while it is loading Minefield can crash.

Perhaps we should start looking into profiling some of this code soon?

Comment 1

[David Dahl :ddahl]

This is on a debug build on linux, so we should test this on all platforms with opt builds to see the extent of the problem

Comment 2

[David Dahl :ddahl]

A cursory glance at the performance tab on window's task manager shows that the cnn page loads have shorter and lower CPU usage spikes, this is of course, unscientific.

Comment 3

[Mihai Sucan [:msucan]]

In the builds I compile for myself from mozilla-central I cannot reproduce the high CPU usage. I tried both with debugging enabled and without.

What I thought about was that the cnn.com site might generate too many errors, causing the WebConsole to "work too much". This doesn't seem to be the case. Overall behavior seems normal.

Anything I should do to investigate this issue further?

Comment 4

[David Dahl :ddahl]

(In reply to comment #3)
> In the builds I compile for myself from mozilla-central I cannot reproduce the
> high CPU usage. I tried both with debugging enabled and without.

Thanks for checking on that. What is your processor type/cores and amount of RAM.
I should mention I am seeing this on an AMD MV-40 dual core 1.6 ghz thinkpad (x100e)

Comment 5

[Mihai Sucan [:msucan]]

(In reply to comment #4)
> (In reply to comment #3)
> > In the builds I compile for myself from mozilla-central I cannot reproduce the
> > high CPU usage. I tried both with debugging enabled and without.
> 
> Thanks for checking on that. What is your processor type/cores and amount of
> RAM.
> I should mention I am seeing this on an AMD MV-40 dual core 1.6 ghz thinkpad
> (x100e)

I have an Intel Dual-Core E6500 (2.93 Ghz), 3GB DDR2 RAM, custom desktop PC, using Ubuntu 10.04 (64 bits).

Comment 6

[Kevin Dangoor]

We don't need to spend time tracking this down right now. We should investigate this a bit later to see if we have potential performance issues with the final implementation.

Comment 7

[Kevin Dangoor]

This might be taken care of by bug 583476

Comment 8

[David Dahl :ddahl]

Not spiking like it was (on my amd 1.6 dual core)