Closed Bug 282392 Opened 20 years ago Closed 19 years ago

firefox hangs on hyperthreading system

Categories

(Firefox :: General, defect)

Product:
Firefox
Component:
General
Platform:
x86
Windows XP
Type:
defect
Priority:
Not set
Severity:
critical

Tracking

()

Status:
RESOLVED INVALID

People

(Reporter: szazharminchet, Assigned: bugzilla)

References

Details

(Keywords: hang) 

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7.5) Gecko/20041107 Firefox/1.0
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7.5) Gecko/20041107 Firefox/1.0

I have been a happy firefox user for years when my company bought me a notebook
computer with a P4 hyperthreading processor. I installed windows xp sp2 on it,
and firefox after that. It freezes randomly. If i set its cpu affinity to just
one processor using the task manager, firefox works fine.
On the same computer firefox works fine under Linux. 

Reproducible: Always

Steps to Reproduce:
1. start browsing.
2.
3.

Actual Results:  
Firefox hangs randomly after a while.

Expected Results:  
Run.
I had this too. Firefox will completely freeze for 10-120 seconds.

To resolve it I did this: Open up task manager, go to the processes tab, right
click over the affected process (firefox) and set the affinity to use just CPU0.
Bingo - it works.

But this needs to be fixed!!!


*** Bug 282639 has been marked as a duplicate of this bug. ***
confirming per duplicate and comment 1.
Status: UNCONFIRMED → NEW
Ever confirmed: true
Keywords: hang
Summary: firefox crashes on hyperthreading system → firefox hangs on hyperthreading system
*** Bug 280995 has been marked as a duplicate of this bug. ***
Hi all,

I've the same problem than the guys above.
When using HT Firefox isn't able to work, it hangs more than works.

I'm testing now by selecting just on cpu for it (I've choosen cpu1, I think it
doesn't matter) and everything seems to be ok.

It's very annoying I can't use the HT function with Firefox.

By the way, I installed Mozilla after changed the option of the CPUs and it used
just one by default. I've changed it to both CPUs and it is working fine.
Mozilla is using HT and is not hanging.



Firefox: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8b2)
Gecko/20050325 Firefox/1.0+


Mozilla: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8b2) Gecko/20050326


Windows XP (I tested yesterday without any SP, a fresh install of XP) SP2
Pentium 4 Prescot HT 3.0GHz 1.0GB RAM  (using a lot less than it)
I didn't have until right now any other problems in my computer regarding HT.
It seems like a bug specific to Firefox on windows.


Important info: in a linux distro based on debian I installed Firefox and used
it with HT features without problems.
sorry, just an aditional comment:
I installed it in a distro based on debian, but it was not the deb package, was
the tar.gz downloaded from ftp.mozilla.org. Version 1.0+, a lot of different
build days :).

another one: I confirmed in 1.0.2 also. Didn't checked the aviary build, just
the nightly.


As more people are starting to use HT, the most won't bother with trying to find
this solution, they will keep with IE.
It should have a high priority.
I've found the following in the Microsoft database and looks related to this bug:
http://support.microsoft.com/default.aspx?scid=kb;en-us;327809
"*Cannot Run Certain Programs on Hyper-Threaded or Dual-Processor Computers with
a CPU Speed of Greater Than 2 GHz*

[snip]
...the result returned by QueryPerformanceFrequency will be the clock speed of
the CPU. On a computer with a processor running faster than 2.1 GHz, this
frequency value requires at least 32 bits of precision. Some programs work with
the result of QueryPerformanceFrequency as a signed integer value, which has
only 31 bits of precision and a sign flag. These programs behave incorrectly on
these faster CPUs. To avoid this problem, programs must use all 64 bits returned
from both
[snip]"
Sorry for the lot of emails, but I think you're receiving it are interested in a
solution:
http://www.robpol86.com/misc_pgs/imagecfg.php
Download imagecfg.exe and then execute "imagecfg.exe -u firefox.exe" (according
to your paths).
It will always start in a single cpu mode, you won't have to bother about it
until the version upgrade.

http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnappcom/html/winntwin95.asp
"The IMAGECFG.EXE tool is useful for verifying version problems in applications.
IMAGECFG adds information to the header of the application's executable file or
DLL responsible for forcing Windows NT to return "incorrect" information from
GetVersion and GetVersionEx calls in place of the "real" operating system
information. Parameters to IMAGECFG give Windows NT the "incorrect" information
to pass to the application. By specifying an earlier version of Windows NT or
the version information associated with Windows 95, applications with version
problems may install or function as expected. IMAGECFG will only write
information to 32-bit headers."

I.e., this tool were made to older versions of win compatibility with new
programs, but the problem it resolves is exactly the same as the one of Firefox
(and Mozilla, it hanges a lot less, but after sometime I verified it also hangs).
This program resolved my problem.
Priority: -- → P2
Just more one comment:

I've installed kcpuload in Debian Sarge and noticed that the way it manages the
cpus is different. It never atributes more than one cpu per process, i.e.,
Firefox runs as in non-hyperthreading systems. This is the reason for the error
be reproducible in WinXP but don't in Linux with SMP active.


Example: when running updatedb I got 100% of cpu load in one cpu and zero in the
other. In windows it try to set one cpu mainly for programs and the other for
the background processess, but it isn't mandatory, both runs in the two cpus.


As I've said before, the fix to this bug is just to tell firefox to run in only
a cpu, it's quite simple.
Bug 67752 has info about better things to do with this bug than force FF to use
just one CPU.
It should have multi-threading enabled, so it would have a better performance in
computers with multiple processors and/or HT enabled.
caiot1@ibest.com.br, please don't touch the priority field, it's reserved for
developers working on this bug. Resetting to default value.

Asking for blocking-aviary1.1 instead: it's a pretty serious issue, which can be
worked around by setting Firefox to work on a single CPU.

Bug 283280 is the Thunderbird bug. As mentioned above Mozilla Suite doesn't seem
to be affected.
Flags: blocking-aviary1.1?
Priority: P2 → --
Version: unspecified → Trunk
Just another comment pertaining to this. Maybe of some interest.

Let me begin by saying that I've been running Firefox since release 0.8 on
multiple HT-enabled systems, even on a ProLiant Server (Windows 2003 / Xeon
2.8Ghz FSB533 / Firefox 1.0PR) with no problems at all.

Trouble showed up for me for the first time when running Firefox 1.0.2 on XP SP2
with a P4 3GHz "Prescott" core. I'm kind of an Intel freak, and if you except
the Xeon processor, I always have been running on Northwoods (FSB400/533/800)
before that, with either XP SP1 or SP2, but no latter version than Firefox 1.0.1.

I think it's safe to assume the bug was either introduced with Firefox 1.0.2 OR
induced by the Prescott architecture.

Oh, by the way - same problem with Thunderbird 1.0.2. An older release
(Thunderbird 1.0) appears to be unaffected on the Northwood systems, though.
Brand New System. P4 3.0 Prescott HT 1mbL2 2GB DC main FoxConn 865A1 (all Intel
chip set). Firefox is first application installed on XP after SP2. Hangs
intermittently. Recovers  W/O intervention in 10-30 secs. Mozilla/5.0 (Windows;
U; Windows NT 5.1; en-US; rv:1.7.7) Gecko/20050414 Firefox/1.0.3 - Tom D.
Hi all,

everyone who can update the bios, I recommend doing it.


The last update to my bios resolved this problem.



Someone after to update the bios still getting problems?
If not, this issue will be closed, as the problem is with the bios, not with
Firefox.
My Bios is not the problem, I just bought the board 2 months ago. I fixed the
problem using imagecfg several days after realizing the problem. This problem
still annoys the **** out of me as everytime I update I have to re-run imagecfg.
Terrible, Mozilla! You guys seem to be paying zero attention to this problem, as
though you just don't care. Create a patch or something! I'll tell you one thing
though, it is chasing multitudes of new users away because not everyone can
figure out how to fix it. I have a prescott core (3.0ghz) and I know that there
is a hell of a lot of Intel customers out there who own them also. This really
has lowered my respect for Mozilla because it is as though the company doesn't
care about losing users. Get on the ball. This is absolutely ridiculous.
Anyone who wants to comment on this bug, please read Bugzilla etiquette page
before commenting. <http://bugzilla.mozilla.org/page.cgi?id=etiquette.html>
xnavydude and all:

my pc also is less than two months old, it also is a prescot as yours. My
mainboard is a intel 865. I updated using the software they provided and it
resolved the problem. Don't you think interesting the most here having the same
configuration? See also comment #12 saying that other computers with HT don't
show this problem.
Using Imagecfg works because it simulates an older enviroment, without the use
of HT, so the bad code don't affect the Mozilla products. As Ryo said, the P4
NorthWood don't show the bug.

The newer computers, with the latest bios, won't show this problem. So we may
call this a recall to the users who were affected by the bad code in the bios.



I have a Prescott P4 and firefox exhibited the same hanging issues. The latest
BIOS update for my motherboard fixed it for me. Had success using imagecfg
before updating BIOS. Here are my system specs and the steps I performed (with
results):

Pentium 4 3.0E Prescott (HT) 800MHz FSB 1MB L2 Cache Socket 478 1.0GB RAM

1. Fresh install Windows XP Pro (no Service Packs): Firefox/1.0.2 runs
correctly, without hanging.

2. Installed SP1: Firefox/1.0.2 runs correctly, without hanging.

3. Installed SP2: Firefox/1.0.2 hangs.

4. Applied imagecfg (see comment #8): Firefox/1.0.2 runs correctly, without hanging.

5. Updated firefox: Firefox/1.0.3 hangs. (had not run imagecfg on updated
firefox.exe)

6. Updated to latest motherboard BIOS which contained changes to the CPU
microcode: Firefox/1.0.3 runs correctly, without hanging.

Thank you for the helpful suggestions, Asrail. 
 


I tried the latest (#21) bios update on my intel 865perl motherboard and all is
well! No more Mozilla hangups. As the guy who started this I thank you all for
your assistance!
Well, as unlikely as it seems, upgrading the bios on my Foxconn 865A1 to 865A01G 
seems to have solved the problem. The funny thing is that there are no listed 
fixes for Hyperthreading problems in Intel's bios fix listings for their 865 
based boards. A mystery, but I am happy that it appears solved. If anyone finds 
out exactly what the underlying problem and fix were, please comment. Tom D.
Bios Update fixed my problem. Thanks
Seems the latest BIOS update also fixed my issue. Currently running FireFox
1.0.3 (just updated it, didn't rerun imagecfg) on previously affected system
with no problems up until now. Kudos going to Asrail for looking up the issue.

Interestingly, the boot splash screen of the mobo (MSI 865PE Neo2-V) also has
changed. Before BIOS update it only showed a generic Intel logo. Now it
correctly identifies an Intel P4 with HT, just like on my other system running a
Northwood-class CPU which didn't need the update. Definitely shows something was
"wrong" somehow with the BIOS... (latest BIOS release just mentions "update for
CPU microcode" and that's all)

When fancy boot logos just tell you more about your issues than you could
possibly believe.  :D

BTW - one user of the mentioned Prescott system also reported intermittent
freezes with MS Word 2002. Freezes that I couldn't relate to damaged documents
or templates or anything. Upon witnessing the issue the only way I could
"unfreeze" Word to avoid loss of document was playing around process priority in
Task Manager (didn't think about modifying CPU affinity right at the time). With
luck this new BIOS will also solve the issue.
So can this be closed as invalid? (As something caused by old BIOS)
Flags: blocking-aviary1.1?
As since comment #14, 2005-04-18, no one said the update don't fixed the bug,
this can be closed.

If someone confirm this bug with newer versions of the bios (after latest
update, be sure about it: the *newest*, not just a new one), fell free to open
it again.
Status: NEW → RESOLVED
Closed: 19 years ago
Resolution: --- → INVALID
*** Bug 286570 has been marked as a duplicate of this bug. ***
Flags: blocking-firefox3.1?
Flags: blocking-firefox3.1?
You need to log in before you can comment on or make changes to this bug.