Firefox 3.5.5 freezes the Athlon X4 620 based box on Asus M4A79 motherboard

[tanix]

I have an Athlon 2, x4 620 quad core processor system on
Asus M4A78 motherboard with 4 Gig DDR2 memory by Team Xtreme.

When I run firefox 3.5.5 or 3.5.6 and refresh several tabs in
rapid succession, it freezes the box unde xp sp3.

What happens is task manager -> performance tab shows one
of cores going upto 100% of CPU useage running ff 3.3.5 and
3 cores from ff 3.5.6, mouse freezes, keyboard is dead and
disk stops flashing every couple of seconds like it normally
does under windows. System is totally dead and needs
to be rebooted. Anyboyd knows anything about it?

I posted a message to ff bb a couple of days ago.
No responses so far.

 

[tanix]

Try starting Firefox in it's safe mode, not Windows Safe mode, and see if
the problem still happens.
The safe mode option can be usually be found in the Start folder for Mozilla
and is called "Mozilla Firefox (Safe Mode)". If for some reason it is not
found on your computer you can change a shortcut to start Firefox to run
using "firefox.exe -safe-mode".

If Firefox does not do lockup things when in the safe mode it's time to
disable all extensions or add-ons and see if the basic program without the
"safe-mode" extension works as well. Then it's a process of elimination
turning back on the add-ons or extensions one-at-a-time and testing until
you identify which one is locking things up for you.

Thanx. I did disable all the addons for now.
Tried to refresh 7 tabs in a rapid succession for about 4 times around.
So far, so good.

Do you happen to know if this is an win xp scheduler bug?
I tried to do the same thing on win 7 on the same box and could not make
it lock up. So I wonder if this is a multicore issue.
The same thing does not happen on a single core processor.

 

[tanix]

tanix wrote:

Try starting Firefox in it's safe mode, not Windows Safe mode, and see if
the problem still happens.
The safe mode option can be usually be found in the Start folder for Mozilla
and is called "Mozilla Firefox (Safe Mode)". If for some reason it is not
found on your computer you can change a shortcut to start Firefox to run
using "firefox.exe -safe-mode".

If Firefox does not do lockup things when in the safe mode it's time to
disable all extensions or add-ons and see if the basic program without the
"safe-mode" extension works as well. Then it's a process of elimination
turning back on the add-ons or extensions one-at-a-time and testing until
you identify which one is locking things up for you.

Thanx. I did disable all the addons and tried refreshing 6 tabs about 5
times around. So far so good.

Do you happen to know if this could be the win xp multicore issue with
xp scheduler? It does not lock up on the same box under win 7,
no matter how hard I try.

Someone said the xp scheduler is not that good in terms of multicore
issues. But what is interesting here is the CPU time going up to 100%.
But under ff 3.5.6 3 threads are locking up. Prolly some dead loop
in some kernel mode driver. I don't know what else to think of.

 

[tanix]

No idea as to what or which add-on is causing the problem but at least you
appear to have ruled out the basic Firefox program. It is back to trial and
error to determine which of your add-ons is causing the problem.

Well, it is too early to say if this is addon problem.
I'll run it for a few days and see if it locks up again.

Btw, what does putting firefox in a safe mode do?
Actually, if it only makes things single threaded, I really don't mind
if THAT helps at all. Better than loosing hours of work.

 

[tanix]

what happens if you run sp2, i had problems with xp sp3 ?

if your running win7 why don't you try virtualbox in win7 and install xp
in virtualbox. and see what happens

That is interesting proposition. Except I don't know what is virtual box
at teh moment. No experience with win 7.
I need to run my own monitoring firewall. Unfortunately, for some strange
reason, it does not run under win 7. Complains about being unable to
connect to NDIS intermediate driver and see not a single reason for it
not to be able to connect. Since is is a clean NDIS driver that works
like a chapm on XP. I just have no time to get into it right now.

If I could install that that driver from virtual box, I'd be willing
to run all my networking from XP in virtual box. But it prolly won't
work I guess. Would be nice if it did with some magic trick.

But do you guys think it may have something to do with XP scheduler?
(i mean firefox locking up the box issue).

I posted to microsoft driver and kernel groups and they claim it is
a firefox issue. But how come it runs fine unded win 7 then?

--
Programmer's Goldmine collections:

http://preciseinfo.org

Code examples and expert discussions on
C++, MFC, VC, ATL, STL, templates, Java, Python, Javascript

 

[Paul]

That is interesting proposition. Except I don't know what is virtual box
at teh moment. No experience with win 7.
I need to run my own monitoring firewall. Unfortunately, for some strange
reason, it does not run under win 7. Complains about being unable to
connect to NDIS intermediate driver and see not a single reason for it
not to be able to connect. Since is is a clean NDIS driver that works
like a chapm on XP. I just have no time to get into it right now.

If I could install that that driver from virtual box, I'd be willing
to run all my networking from XP in virtual box. But it prolly won't
work I guess. Would be nice if it did with some magic trick.

But do you guys think it may have something to do with XP scheduler?
(i mean firefox locking up the box issue).

I posted to microsoft driver and kernel groups and they claim it is
a firefox issue. But how come it runs fine unded win 7 then?

What is missing from this thread, is any evidence the computer has
been tested for correct computing operation. Try a stress test program.

http://www.mersenne.org/freesoft/

This is the download I'd use on my WinXP 32 bit machine.

http://mersenneforum.org/gimps/p95v259.zip

When you run the program the first time, it'll ask you if you
want to "Join GIMPS". You can answer that you're "Just Testing",
as you're going to be using the stress test feature.

When you run a test, the program starts a thread per core. The
program does a math calculation with a known answer. The program
can detect when a math error has occurred.

A math error could be due to bad RAM or mis-adjusted RAM settings.
Or, it could be a problem with the processor (the program is used
for testing overclocked processors, to see if they still work
properly at the overclocked speed).

The program will stop a thread of execution, if an error is detected.
I'd want to see at least a four hour run, error free, before accepting
that my new computer is working well.

Try that test, and report back.

Paul

 

[tanix]

What is missing from this thread, is any evidence the computer has
been tested for correct computing operation. Try a stress test program.

http://www.mersenne.org/freesoft/

This is the download I'd use on my WinXP 32 bit machine.

http://mersenneforum.org/gimps/p95v259.zip

Well, I did run memtest86 for about 20 hrs. and I did run
Prime95 for several hours although I don't think it is sufficient
to test your box.

The problem with Prime95 is that it simply loads up all your cores
with heavy duty math processing. But that does not test other modes
of memory transfers, such as DMA. And if you run it stand alone,
as I did, it does not test the network drivers, or, for that matter,
ANY other drivers in your box.

So, to me, Prime95 is of not much use, at least in the cases like mine.
It does load your cores and your temperature goes up to 53C.
But I just do not feel these tests are sufficient.

The program will stop a thread of execution, if an error is detected.
I'd want to see at least a four hour run, error free, before accepting
that my new computer is working well.

Well, I ran it for about 3 hours.

Try that test, and report back.

Report: no problems found with memory and after running Prime95 for
several hours, which is prolly 10 times more than enough for this
particular problem. Because box is guaranteed to freeze after doing
the first round of page refreshes by switching tabs in a rapid
succession for 6 pages. It pretty much happens from the 1st try.

I did disable all the firefox addons except of imacros.
Since then, i tried several rounds of rapid page refreshes
hitting page refresh as soon as rotating wheel, indicating page
loading, stops rotating and page fully loads.

So far, after about 4 rounds (each refreshing 6 pages), box did not
lock up. Too early to say if it is an addon problem. I'll have to
give it a few days. I just pray it IS the addon issue.

Paul

--
Programmer's Goldmine collections:

http://preciseinfo.org

Tens of thousands of code examples and expert discussions on
C++, MFC, VC, ATL, STL, templates, Java, Python, Javascript,
organized by major topics of language, tools, methods, techniques.

 

[tanix]

go here and read and download http://www.virtualbox.org/wiki/Downloads

Thanx for the link.

Btw, after I disabled all the addons in firefox, the machine has been
stable so far.

Once I run it for several days without a single crash, I'll try to
determine which pluging is doing it.

It was interesting to see the windows kernel and device driver guys
suspecting it is not windows, but firefox.

--
Programmer's Goldmine collections:

http://preciseinfo.org

Tens of thousands of code examples and expert discussions on
C++, MFC, VC, ATL, STL, templates, Java, Python, Javascript,
organized by major topics of language, tools, methods, techniques.

 

[tanix]

I can recall only one lockup in many months on the system
I'm typing on. Coincidentally it was a few days ago with
Firefox 3.5, shortly after I had added an add-on named
Faster Fox and had tried to launch a new window or tab or
something.

Well, so far, it looks like one of the new addons on firefox.
I have only a few addons, the firebug, web developer and a couple
of new ones that I did not use b4.

I don't recall all the variables involved, it
was WinXP /SP3, one of the windows had a Flash game running,
several others had quite a lot of content due to how
Fasterfox pre-loads subsequent pages and appends them onto
the present page making for really huge pages.

I don't have anything fishy running on my box.
It is all basic apps and the main one is the program I use
for years without any problems.

I dn't think it's a scheduler bug though, before that I'd
used quite a lot of tabs and add-ons without this kind of
problem.

Well, try to update > 5 tabs in a rapid succession
if you are running XP SP3. Do a few rounds.
I am not sure if some javascript from the sites that are
displayed in those tabs might have to contribute to this.

I suspect it's just one of the add-ons is a bit
buggy, and something unique about that system or OS
installation is a contributing factor.

Could be. But what I am missing here is how come the MOUSE
is getting frozen. Mouse is about the highest priority there is.

--
Programmer's Goldmine collections:

http://preciseinfo.org

Tens of thousands of code examples and expert discussions on
C++, MFC, VC, ATL, STL, templates, Java, Python, Javascript,
organized by major topics of language, tools, methods, techniques.

 

[Paul]

Could be. But what I am missing here is how come the MOUSE
is getting frozen. Mouse is about the highest priority there is.

You could always file a report on bugzilla.

https://bugzilla.mozilla.org/

Paul

 

[Paul]

Thanx. Appreciated.
The problem with filing a bug report is that I doubt they
would be willing to even look at it without having either
a snapshot, which is impossible to get in this case, or
knowing exactly how to reproduce it.

Considering the fact that XP is a "no priority" issue,
the chances they'll look at it are as slim as it gets,
and the chances of it getting fixed are just the same.

So...

But thanx for your suggestion.

If their add-on or plug-in architecture has some security issues
or holes, even if they aren't interested in the details of your
bug report, it might give them something to think about when
architecting things in the future. If I was a Firefox developer,
I'd be a little curious how your symptoms could happen.

Paul

 

[tanix]

If their add-on or plug-in architecture has some security issues
or holes, even if they aren't interested in the details of your
bug report, it might give them something to think about when
architecting things in the future. If I was a Firefox developer,
I'd be a little curious how your symptoms could happen.

I would be MORE than curious.
Because if it IS a firexox problem, then they have a serious
architectural problem. But the problems is this:
No matter how badly any application misbehaves,
it should not impact the O/S.

So, what are we dealing with here is probably never going to
be known. For one thing, Windows is not trully multitasking O/S,
and you can see it in ALL sorts of cases.
The very fact that some up can totally lock up your O/S
is the problem of O/S more than anything else.

So, try to talk to bill gates.

Happy, happy people.
Happy, happy people.

Ding dong dong.
Ding dong dong.

Paul

--
Programmer's Goldmine collections:

http://preciseinfo.org

Tens of thousands of code examples and expert discussions on
C++, MFC, VC, ATL, STL, templates, Java, Python, Javascript,
organized by major topics of language, tools, methods, techniques.