446265 - everlasting 100% CPU usage upon closing - FX3 only -

Status: RESOLVED WORKSFORME

(Firefox :: General, defect)

Description

[VGR]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.0; fr; rv:1.9.0.1) Gecko/2008070208 Firefox/3.0.1
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.0; fr; rv:1.9.0.1) Gecko/2008070208 Firefox/3.0.1

This is an issue I never had with FX2 latest. It occurs since 3.0.0 (and 3.0.1). Whenever I close FX, I get 100% CPU usage for a lot more then 21s or 30s (see bug https://bugzilla.mozilla.org/show_bug.cgi?id=415201) and it's not related to bookmarks or the like.

Startup time is normal.

The problem doesn't depend on browsed websites (AFAIK) nor on extensions (I have a very few of them, with one disabled because not supporting FX3).

I have no crash so I can't show you an error, but the 100% CPU useage has an impact on my CPU temp and on other processes running, of course. This can last for an entire night. This smells like a MM issue... (MM=memory manager) and I would look at the differences between latest FX2 and FX3.0.1 in this area if I were you ;-)


The process of FX2 unallocating memory was somewhat slow but predictable. I could follow it in Task Manager, seeing virtual memory being freed. The process took sometimes 10s or so.

Now in FX3 it lasts forever and no memory is deallocated. I have to kill the process manually so that the machine is functional again.

After this "kill", FX3 starts nicely, not complaining about a crashed session or the like.

If there is a way to track the problem down without myself compiling FX3 from sources, it would be nice ;-)

HTH

PS time permitting, I will try "FX3 in safe mode" to see if it does make any difference. I will post a followup later on.

Reproducible: Always

Steps to Reproduce:
1.
2.
3.

Comment 1

[VGR]

Hello,

This problem subsists in the latest 3.0 version (3.0.3 if I'm not mistaked)

I gave a try to the latest dev snapshot (20081022 Minefield/3.1b2pre) and I must say that ***apparently*** the problem is not there anymore. It was then probably linked to history/bookmarks (I don't have 4000 bookmarks though).
Now quitting is a small 40% peak and quite robustly quick.

Good job

PLEASE RELEASE THIS ASAP to the next Firefox 3.0 FIX !!!!

regards

(In reply to comment #0)
Gecko/2008070208 Firefox/3.0.1
> This is an issue I never had with FX2 latest. It occurs since 3.0.0 (and
> 3.0.1). Whenever I close FX, I get 100% CPU usage for a lot more then 21s or
> 30s (see bug https://bugzilla.mozilla.org/show_bug.cgi?id=415201) and it's not
> related to bookmarks or the like.

In fact, I had an INFINITE 100% CPU USEAGE. I HAD TO KILL THE PROCESS

Comment 2

[Carlos Alén Silva]

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

Windows XP Pro SP3

Hello.

I have experienced the same problem several times. The last one was a few minutes ago. Fx 3.03 was becoming slow. I closed it and then there was a CPU spike. When I checked the task manager, firefox.exe was still running and hogging CPU. I had to kill the process.

Comment 3

[VGR]

Follow-up on this pest. I reverted to "official" FX 3 (too few extensions supported Minefield) and the bug is still there, even in 3.0.4

Could anybody diff the closing code between minefield 3.1b2pre and FX 3.0.4 so that closing the browser doesn't produce an everlasting 100% CPU useage ? (with no sign of activity on RAM nor VRAM)

Comment 4

[Wayne Mery (:wsmwk)]

You can do a binary search to get a one or two day regression range - see https://wiki.mozilla.org/MozillaQualityAssurance:Home_Page#What_We_Use

Then use the checkin log* with the date range to find a likely patch that caused the problem. *http://hg.mozilla.org/mozilla-central/pushloghtml?startdate=10%2F03%2F2008+11%3A38&enddate=10%2F04%2F2008+11%3A38

Post your results here. This is something someone with average computer skills can accomplish. If need help visit firefox #bugday, runs every tuesday, or #qa on other days. QA can help you shoot this bug. http://quality.mozilla.org/events/bug-days

sampling of possible fixing bugs - no guarantee that any of these resolved your issue - https://bugzilla.mozilla.org/buglist.cgi?query_format=advanced&short_desc_type=anywordssubstr&short_desc=shut+quit&product=Core&product=Firefox&product=Toolkit&long_desc_type=allwordssubstr&long_desc=&bug_file_loc_type=allwordssubstr&bug_file_loc=&status_whiteboard_type=allwordssubstr&status_whiteboard=&keywords_type=allwords&keywords=&resolution=FIXED&bug_severity=blocker&bug_severity=critical&bug_severity=major&emailreporter1=1&emailtype1=substring&email1=&emailassigned_to2=1&emailreporter2=1&emailqa_contact2=1&emailtype2=substring&email2=&bugidtype=include&bug_id=&votes=&chfieldfrom=500d&chfieldto=Now&chfieldvalue=&cmdtype=doit&order=Reuse+same+sort+as+last+time&field0-0-0=short_desc&type0-0-0=nowordssubstr&value0-0-0=crash

Comment 5

[Wayne Mery (:wsmwk)]

VGR, 
Do you get the same results with the latest beta of 3.5?
available from http://www.mozilla.com/en-US/firefox/all-beta.html

If you don't see the problem then this may be bug 453178 comment 1.  
If you do see this, then the blockers to bug 407981 should be checked.

Comment 6

[VGR]

no, since the last time I installed a "beta" (Minefield something) it replaced my FX in stead of installing side-by-side and I had a hard time reinstalling a "valid" FX without losing to much data in the process. (the problem was no extensions ever worked with the "new" version)

I confirm the symptoms with the latest 3.0.10, and on Win XP SP3 as well as on Win 2K SP4.

I will try to find a machine where I can break the browser and install the "latest beta". Thanks for the lead.

Comment 7

[Wayne Mery (:wsmwk)]

do custom install to a directory of your choice and you'll have no problem.
create a new profile if you prefer
 http://support.mozilla.com/en-US/kb/Managing+profiles

Comment 8

[VGR]

so far, so good. I keep testing the latest beta for some days more.

Comment 9

[VGR]

problem not happening with 3.5 beta 4
too bad this plagued me for 10 months since 3.0
all of this only because of the bookmarks/history handling on closure...

thanks for having pointed me to the beta. Halas, I'll have to wait until 3.5 comes out and extensions become compatible ;-)
Any roadmap for the release of 3.5 ?

Comment 10

[Wayne Mery (:wsmwk)]

I believe release candidate 1 is in preparation
=> WFM per reporter