66271 - Browser uses ~100% CPU after loading page

Status: RESOLVED WORKSFORME

(SeaMonkey :: General, defect)

Description

[Guy Rixon]

From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; SunOS 5.8 sun4u; en-US; m18) Gecko/20010110
BuildID:    2001011021

I loaded a page which included HTMl and JFIF images supplied via calls to CGIs.
The page loaded normally, and the stop button greyed out after the load. But the
process was still using all the CPU!

I killed the browser, restarted, and when back to the same page, but couldn't
reproduce the fault.

This has happened before, randomly, on various pages.



Reproducible: Couldn't Reproduce

Here are the leading entries from top, about five minutes after I noticed the
problem:

load averages:  1.11,  1.19,  1.22
13:38:14
67 processes:  65 sleeping, 1 running, 1 on cpu
CPU states:  0.0% idle, 69.7% user, 30.3% kernel,  0.0% iowait,  0.0% swap
Memory: 256M real, 91M free, 198M swap in use, 996M swap free

  PID USERNAME THR PRI NICE  SIZE   RES STATE   TIME    CPU COMMAND
11387 gtr        7  49    0   48M   38M run   104:10 98.29% mozilla-bin
10845 root       1  59    0   22M   67M sleep   1:15  0.36% Xsun
11696 gtr        1  59    0 6936K 4904K sleep   0:00  0.15% dtterm
12090 gtr        1  59    0 1552K 1352K cpu     0:00  0.05% top
11223 gtr        1  49    0 6840K 4576K sleep   0:00  0.05% dtterm

It looks like it may have run away earlier, but in such a way that the runaway
threads or whatever didn't stop the browser working. That's annoying, because it
could have been any of a dozen pages where it originally went wrong. Or maybe it
really _used_ 104min of CPU today.

On the console, it said:

JavaScript error:
chrome://communicator/content/securityUI.js line 29:
Components.classes['@mozilla.org/secure_browser_ui;1'] has no properties

Comment 1

[Doron Rosenberg (IBM)]

do you have a test url?

Comment 2

[Fabian Guisset]

the javascript error appears because you do not have PSM installed. I guess it
is not relevant in the description of this bug.

Comment 3

[Guy Rixon]

Try
http://cass03.ast.cam.ac.uk/cgi-bin/wfspreview/wfspreview.cgi?153149&ra=10:00:00&dec=00:00:00&radius=30
This doesn't reliably reproduce the mozilla problem, but is very similar to the 
URL where I saw the problem. It hits a bug in my CGI where animage/jpeg header
is returned but no bytes of content; this might be a co-factor.

Try
http://cass03.ast.cam.ac.uk/cgi-bin/wfspreview/wfspreview.cgi?159871&ra=10:00:00&dec=00:00:00&radius=30
if you want a test-case where the CGI works.

Comment 4

[Keyser Sose]

I am not seeing this. Might be a unix only problem.
Platform: PC
OS: Windows 98
Mozilla Build: 2001020104

Comment 5

[Asa Dotzler [:asa]]

neither of the above URLs cause a cpu spike or lockup on my linux mozilla build
020210.

Comment 6

[Fabian Guisset]

Both above-mentionned urls give me a "software error", apparently a syntax error.
Any idea under which component this should go? Is there any item on that page
which, once removed, prevents the bug from triggering? Please reassign if you
find anything relevant. Thank you!

Comment 7

[Guy Rixon]

Sorry, the test URLs were mis-typed.  They should be

http://cass03.ast.cam.ac.uk/cgi-bin/wfspreview/wfspreview.cgi?run=153149&ra=10:00:00&dec=00:00:00&radius=30

http://cass03.ast.cam.ac.uk/cgi-bin/wfspreview/wfspreview.cgi?run=159871&ra=10:00:00&dec=00:00:00&radius=30

Comment 8

[Keyser Sose]

WORKSFORME
Platform: PC
OS: Linux 2.2.17
Mozilla Build: 2001030108

Reporter what version of XFree are you using?

Comment 9

[Guy Rixon]

> Reporter what version of XFree are you using?

I'm not using XFree, I'm on Solaris 8 using CDE. My X server is the one shipped
by Sun and installed in /usr/openwin/bin.

Comment 10

[Keyser Sose]

Reporter is this still a problem in the latest nightlies?

Comment 11

[Keyser Sose]

Marking NEW.

Comment 12

[Matthias Versen [:Matti]]

Reporter: Is this still a problem with a more recent build ?

Comment 13

[Guy Rixon]

As far as I can see without undertaking extensive tests, this particular bug is
no longer present in recent builds. There are problems with downloads of
dynamically generated images, but they don't involve using up all the CPU.

Comment 14

[Matthias Versen [:Matti]]

Thanks for your answer !

I mark this bug worksforme, please reopen if you see this again !
Downloading of dynamically generated images should be another bug...