[gentoo-user] Can't kill Firefox!

[gentoo-user] Can't kill Firefox!

[Mick]

[-- Attachment #1: Type: Text/Plain, Size: 488 bytes --]

Whatever Dale had on his machine must have infected mine!  LOL!

A 32bit x86 box with KDE4.6, running firefox-3.6.17 and xulrunner-1.9.2.17 
after a few hours and loads of tabs (sometimes up to 15 or so) eventually 
hangs X.

I can switch to a console and login as the user running the X session, then 
try to kill the bloody thing with killall, kill -15 <pid> and kill -9 <pid>.  
It won't barge.  Mind you I haven't tried this as root.

How can this be?
-- 
Regards,
Mick

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

Re: [gentoo-user] Can't kill Firefox!

[Dale]

Mick wrote:
> Whatever Dale had on his machine must have infected mine!  LOL!
>
> A 32bit x86 box with KDE4.6, running firefox-3.6.17 and xulrunner-1.9.2.17
> after a few hours and loads of tabs (sometimes up to 15 or so) eventually
> hangs X.
>
> I can switch to a console and login as the user running the X session, then
> try to kill the bloody thing with killall, kill -15<pid>  and kill -9<pid>.
> It won't barge.  Mind you I haven't tried this as root.
>
> How can this be?
>    

I would try to kill it as root.  The -9 option should work.  That hasn't 
failed me yet.  I always run kill commands as root and DOUBLE check the 
PID after typing it in.

I regret to find out that this is not just my problem.  Now we got to 
figure this out.  It even affects x86 huh.  This is interesting.

Dale

:-)  :-)

Re: [gentoo-user] Can't kill Firefox!

[Stroller]

On 21 July 2011, at 19:48, Dale wrote:
> ...
> I would try to kill it as root.  The -9 option should work.  That hasn't failed me yet.  I always run kill commands as root and DOUBLE check the PID after typing it in.

I believe that `kill -9` is bad practice - doesn't it leave memory allocated to the processes as unrecoverable or something?

I believe other signals should be attempted first. See the list in `man kill`. I won't swear to it, but `kill -4` sounds right.

Stroller.

Re: [gentoo-user] Can't kill Firefox!

[Todd Goodman]

* Stroller <stroller@stellar.eclipse.co.uk> [110723 09:21]:
> 
> On 21 July 2011, at 19:48, Dale wrote:
> > ...
> > I would try to kill it as root.  The -9 option should work.  That hasn't failed me yet.  I always run kill commands as root and DOUBLE check the PID after typing it in.
> 
> I believe that `kill -9` is bad practice - doesn't it leave memory allocated to the processes as unrecoverable or something?
> 
> I believe other signals should be attempted first. See the list in `man kill`. I won't swear to it, but `kill -4` sounds right.
> 
> Stroller.
> 
> 

No, a kill -9 shouldn't leave memory allocated.

However, it is best to try other signals first because it gives the app
a chance to clean up before closing (if they handle the signals.)  But
that's also why they don't necessarily work.

SIGHUP (kill -1) is the first thing I generally try.  Depending on the
app I may try SIGQUIT (kill -15) but generally it's straight to kill -9
if the kill -1 doesn't work.

Todd

Re: [gentoo-user] Can't kill Firefox!

[Mick]

[-- Attachment #1: Type: Text/Plain, Size: 1322 bytes --]

On Saturday 23 Jul 2011 14:31:23 Todd Goodman wrote:
> * Stroller <stroller@stellar.eclipse.co.uk> [110723 09:21]:
> > On 21 July 2011, at 19:48, Dale wrote:
> > > ...
> > > I would try to kill it as root.  The -9 option should work.  That
> > > hasn't failed me yet.  I always run kill commands as root and DOUBLE
> > > check the PID after typing it in.
> > 
> > I believe that `kill -9` is bad practice - doesn't it leave memory
> > allocated to the processes as unrecoverable or something?
> > 
> > I believe other signals should be attempted first. See the list in `man
> > kill`. I won't swear to it, but `kill -4` sounds right.
> > 
> > Stroller.
> 
> No, a kill -9 shouldn't leave memory allocated.
> 
> However, it is best to try other signals first because it gives the app
> a chance to clean up before closing (if they handle the signals.)  But
> that's also why they don't necessarily work.
> 
> SIGHUP (kill -1) is the first thing I generally try.  Depending on the
> app I may try SIGQUIT (kill -15) but generally it's straight to kill -9
> if the kill -1 doesn't work.
> 
> Todd

Thanks, I tried kill -15 first and then kill -9.  I also tried pkill, killall, 
but I got no response from firefox and run out of ideas.

What does kill -4 do?  What is ILL?
-- 
Regards,
Mick

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

Re: [gentoo-user] Can't kill Firefox!

[Todd Goodman]

* Mick <michaelkintzios@gmail.com> [110723 10:20]:
> On Saturday 23 Jul 2011 14:31:23 Todd Goodman wrote:
> > * Stroller <stroller@stellar.eclipse.co.uk> [110723 09:21]:
> > > On 21 July 2011, at 19:48, Dale wrote:
> > > > ...
> > > > I would try to kill it as root.  The -9 option should work.  That
> > > > hasn't failed me yet.  I always run kill commands as root and DOUBLE
> > > > check the PID after typing it in.
> > > 
> > > I believe that `kill -9` is bad practice - doesn't it leave memory
> > > allocated to the processes as unrecoverable or something?
> > > 
> > > I believe other signals should be attempted first. See the list in `man
> > > kill`. I won't swear to it, but `kill -4` sounds right.
> > > 
> > > Stroller.
> > 
> > No, a kill -9 shouldn't leave memory allocated.
> > 
> > However, it is best to try other signals first because it gives the app
> > a chance to clean up before closing (if they handle the signals.)  But
> > that's also why they don't necessarily work.
> > 
> > SIGHUP (kill -1) is the first thing I generally try.  Depending on the
> > app I may try SIGQUIT (kill -15) but generally it's straight to kill -9
> > if the kill -1 doesn't work.
> > 
> > Todd
> 
> Thanks, I tried kill -15 first and then kill -9.  I also tried pkill, killall, 
> but I got no response from firefox and run out of ideas.
> 
> What does kill -4 do?  What is ILL?

SIGILL is an illegal instruction.  Not one you'd want to use with kill
in general.

Todd

[gentoo-user] Re: Can't kill Firefox!

[walt]

On 07/21/2011 11:38 AM, Mick wrote:
> Whatever Dale had on his machine must have infected mine!  LOL!
> 
> A 32bit x86 box with KDE4.6, running firefox-3.6.17 and xulrunner-1.9.2.17 
> after a few hours and loads of tabs (sometimes up to 15 or so) eventually 
> hangs X.
> 
> I can switch to a console and login as the user running the X session, then 
> try to kill the bloody thing with killall, kill -15 <pid> and kill -9 <pid>.  
> It won't barge.  Mind you I haven't tried this as root.

I wonder if the machine is busy swapping because RAM is full.  Run top and
see how much swap is used or if some process is using 100% of CPU.

Re: [gentoo-user] Re: Can't kill Firefox!

[Mick]

[-- Attachment #1: Type: Text/Plain, Size: 903 bytes --]

On Thursday 21 Jul 2011 23:31:19 walt wrote:
> On 07/21/2011 11:38 AM, Mick wrote:
> > Whatever Dale had on his machine must have infected mine!  LOL!
> > 
> > A 32bit x86 box with KDE4.6, running firefox-3.6.17 and
> > xulrunner-1.9.2.17 after a few hours and loads of tabs (sometimes up to
> > 15 or so) eventually hangs X.
> > 
> > I can switch to a console and login as the user running the X session,
> > then try to kill the bloody thing with killall, kill -15 <pid> and kill
> > -9 <pid>. It won't barge.  Mind you I haven't tried this as root.
> 
> I wonder if the machine is busy swapping because RAM is full.  Run top and
> see how much swap is used or if some process is using 100% of CPU.

That's the thing, no swap was being used at the time.  Firefox was using 
something like 20% CPU (but can't recall for sure).  I'll keep an eye on it 
next time.
-- 
Regards,
Mick

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

Re: [gentoo-user] Can't kill Firefox!

[Sebastian Beßler]

[-- Attachment #1: Type: text/plain, Size: 719 bytes --]

Am 21.07.2011 20:38, schrieb Mick:
> Whatever Dale had on his machine must have infected mine!  LOL!
> 
> A 32bit x86 box with KDE4.6, running firefox-3.6.17 and xulrunner-1.9.2.17 
> after a few hours and loads of tabs (sometimes up to 15 or so) eventually 
> hangs X.

I had the same problem with firefox and KDE4.6, X just froze and I could
not kill firefox. This happend sometimes twice a day.
I had to kill everything with altgr+print+e, nothing else worked.

After my change to XFCE4 the problem has never occured again for now 2
month. Maybe the people at KDE should use time to fix bugs and make it
stable before putting more and more eyecandy and stuff in it.

Greetings

Sebastian Beßler


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 900 bytes --]

Re: [gentoo-user] Can't kill Firefox!

[Volker Armin Hemmann]

On Friday 22 July 2011 13:54:09 Sebastian Beßler wrote:
> Am 21.07.2011 20:38, schrieb Mick:
> > Whatever Dale had on his machine must have infected mine!  LOL!
> > 
> > A 32bit x86 box with KDE4.6, running firefox-3.6.17 and
> > xulrunner-1.9.2.17 after a few hours and loads of tabs (sometimes up to
> > 15 or so) eventually hangs X.
> 
> I had the same problem with firefox and KDE4.6, X just froze and I could
> not kill firefox. This happend sometimes twice a day.
> I had to kill everything with altgr+print+e, nothing else worked.
> 
> After my change to XFCE4 the problem has never occured again for now 2
> month. Maybe the people at KDE should use time to fix bugs and make it
> stable before putting more and more eyecandy and stuff in it.

maybe it is not KDE's fault when firefox is badly coded?

Using KDE4 with konqueror+chromium for days and weeks of uptime without 
problems.


-- 
#163933

Re: [gentoo-user] Can't kill Firefox!

[Sebastian Beßler]

[-- Attachment #1: Type: text/plain, Size: 458 bytes --]

Am 22.07.2011 15:13, schrieb Volker Armin Hemmann:

> maybe it is not KDE's fault when firefox is badly coded?

Lets think.
KDE4+Firefox = X hangs and firefox can't be killed
XFCE4+Firefox = no problems

Sure, it has to be firefox

Oh and a test 20 minutes ago showed

KDE4+Chromium = X hangs and chromium can't be killed
XFCE4+Chromium = no problems

It looks like there is a pattern but I could be wrong.

Greetings

Sebastian Beßler


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 900 bytes --]

Re: [gentoo-user] Can't kill Firefox!

[Alan McKinnon]

On Friday 22 July 2011 16:39:41 Sebastian Beßler did opine thusly:
> Am 22.07.2011 15:13, schrieb Volker Armin Hemmann:
> > maybe it is not KDE's fault when firefox is badly coded?
> 
> Lets think.
> KDE4+Firefox = X hangs and firefox can't be killed
> XFCE4+Firefox = no problems
> 
> Sure, it has to be firefox
> 
> Oh and a test 20 minutes ago showed
> 
> KDE4+Chromium = X hangs and chromium can't be killed
> XFCE4+Chromium = no problems
> 
> It looks like there is a pattern but I could be wrong.

It may look like KDE is the likely culprit based on just the 
information you provide, but I would be more inclined to look at 
browser plugins first, concentrating on those with both Firefox and 
Chromium versions from the same developer team.

FF, Chromium, KDE, XFCE are all large projects with large userbases. 
The odds of dumbass bugs remaining in the code tends to decrease with 
such projects. Compare that to niche plugins which do not have the 
same eyeball visibility.

There could be a cornercase bug in KDE that only shows up on your 
specific combination, or maybe there is some edge KDE app you use that 
disagrees with violently with FF. Or maybe it's the video driver that 
doesn't actually do what it tells KDE it can do (remember the 
painfully slow nVidia drivers with early KDE4?)

The point I'm making is that your data set and initial conclusions 
appear far too simplistic and have not taken the real world into 
account. 

You have a mere correlation, you cannot conclude causation from that 
data yet.

-- 
alan dot mckinnon at gmail dot com

Re: [gentoo-user] Can't kill Firefox!

[Sebastian Beßler]

[-- Attachment #1: Type: text/plain, Size: 1469 bytes --]

Am 22.07.2011 16:57, schrieb Alan McKinnon:

> It may look like KDE is the likely culprit based on just the 
> information you provide, but I would be more inclined to look at 
> browser plugins first, concentrating on those with both Firefox and 
> Chromium versions from the same developer team.

I have zero browser plugins installed. So that is not the culprit. And
why should plugins lock up X when used under KDE but not under XFCE?
The problem that X freezes with KDE4 is more likely with webbrowsers but
happend when using other programms too. But because of the fact that one
or more browsers are nearly always running it is hard to find a freeze
without a browser running.

> There could be a cornercase bug in KDE that only shows up on your 
> specific combination, or maybe there is some edge KDE app you use that 
> disagrees with violently with FF. Or maybe it's the video driver that 
> doesn't actually do what it tells KDE it can do (remember the 
> painfully slow nVidia drivers with early KDE4?)

I use the opensource drivers for ati-cards. So t is unlikely to be
driver related.

I use no edge KDE apps only amarok, yakuake, kate, konsole, dolphin,
ktorrent from time to time. All of them but dolphin is still in use
under xfce4 so I think I can rule them out too.

I know it could still be a truckload of other things, but as a fan of
Ockham's razor I still say that KDE4 is broken.

Greetings

Sebastian Beßler


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 900 bytes --]

Re: [gentoo-user] Can't kill Firefox!

[Michael Schreckenbauer]

On Friday 22 July 2011 18:41:26 Sebastian Beßler wrote:
> Am 22.07.2011 16:57, schrieb Alan McKinnon:
> > It may look like KDE is the likely culprit based on just the
> > information you provide, but I would be more inclined to look at
> > browser plugins first, concentrating on those with both Firefox and
> > Chromium versions from the same developer team.
> 
> I have zero browser plugins installed. So that is not the culprit. And
> why should plugins lock up X when used under KDE but not under XFCE?

Maybe because you did not enable compositing in xfce4, but use it with kde4?

> The problem that X freezes with KDE4 is more likely with webbrowsers but
> happend when using other programms too. But because of the fact that one
> or more browsers are nearly always running it is hard to find a freeze
> without a browser running.
> 
> > There could be a cornercase bug in KDE that only shows up on your
> > specific combination, or maybe there is some edge KDE app you use that
> > disagrees with violently with FF. Or maybe it's the video driver that
> > doesn't actually do what it tells KDE it can do (remember the
> > painfully slow nVidia drivers with early KDE4?)
> 
> I use the opensource drivers for ati-cards. So t is unlikely to be
> driver related.

Well, have a look at http://www.mail-archive.com/ubuntu-
bugs@lists.ubuntu.com/msg1483058.html for an example how the open source 
atidrivers can hang X with firefox. Note that compositing was enabled here 
(compiz). And here's another one: 
https://bugzilla.redhat.com/show_bug.cgi?id=436632

> I use no edge KDE apps only amarok, yakuake, kate, konsole, dolphin,
> ktorrent from time to time. All of them but dolphin is still in use
> under xfce4 so I think I can rule them out too.
> 
> I know it could still be a truckload of other things, but as a fan of
> Ockham's razor I still say that KDE4 is broken.

kde4 works for me, firefox works for me. I use closed source nvidia drivers. So 
Ockham's razor tells us, the ati-drivers are broken.
Now more seriously: try attaching gdb to the hanging process and get a 
backtrace. You are affected by some bug in some OSS. If you want to be helpful, 
provide the necessary info. Even if you don't understand the backtrace, others 
do.

> Greetings
> Sebastian Beßler

Regards,
Michael

Re: [gentoo-user] Can't kill Firefox!

[Sebastian Beßler]

[-- Attachment #1: Type: text/plain, Size: 194 bytes --]

Am 22.07.2011 21:20, schrieb Michael Schreckenbauer:

> Maybe because you did not enable compositing in xfce4, but use it with kde4?

No, disabling composition is the first thing I do.



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 900 bytes --]

Re: [gentoo-user] Can't kill Firefox!

[Michael Schreckenbauer]

On Saturday 23 July 2011 10:24:17 Sebastian Beßler wrote:
> Am 22.07.2011 21:20, schrieb Michael Schreckenbauer:
> > Maybe because you did not enable compositing in xfce4, but use it with
> > kde4?
> 
> No, disabling composition is the first thing I do.

Ah, thanks.
Without a backtrace it's hard to tell then what's going on here.

Regards,
Michael

Re: [gentoo-user] Can't kill Firefox!

[Mick]

[-- Attachment #1: Type: Text/Plain, Size: 2513 bytes --]

On Friday 22 Jul 2011 20:20:47 Michael Schreckenbauer wrote:
> On Friday 22 July 2011 18:41:26 Sebastian Beßler wrote:
> > Am 22.07.2011 16:57, schrieb Alan McKinnon:
> > > It may look like KDE is the likely culprit based on just the
> > > information you provide, but I would be more inclined to look at
> > > browser plugins first, concentrating on those with both Firefox and
> > > Chromium versions from the same developer team.
> > 
> > I have zero browser plugins installed. So that is not the culprit. And
> > why should plugins lock up X when used under KDE but not under XFCE?
> 
> Maybe because you did not enable compositing in xfce4, but use it with
> kde4?

Well, this is partly correct, although I do not use compositing in KDE - 
because it will crash X.


> > The problem that X freezes with KDE4 is more likely with webbrowsers but
> > happend when using other programms too. But because of the fact that one
> > or more browsers are nearly always running it is hard to find a freeze
> > without a browser running.
> > 
> > > There could be a cornercase bug in KDE that only shows up on your
> > > specific combination, or maybe there is some edge KDE app you use that
> > > disagrees with violently with FF. Or maybe it's the video driver that
> > > doesn't actually do what it tells KDE it can do (remember the
> > > painfully slow nVidia drivers with early KDE4?)
> > 
> > I use the opensource drivers for ati-cards. So t is unlikely to be
> > driver related.

I also use open source drivers and these troubles started recently.  I am 
running:

  x11-base/xorg-server-1.10.2 
  x11-base/xorg-x11-7.4-r1  
  x11-base/xorg-drivers-1.10
  media-libs/mesa-7.10.3

> Well, have a look at http://www.mail-archive.com/ubuntu-
> bugs@lists.ubuntu.com/msg1483058.html for an example how the open source
> atidrivers can hang X with firefox. Note that compositing was enabled here
> (compiz). And here's another one:
> https://bugzilla.redhat.com/show_bug.cgi?id=436632

Thanks for these links.  This machine has an ATI Radeon X600 (RV380) 3E50 
(PCIE).  I have noticed that when I hover over the K menu, or when I select 
Log out/Shut down, I lose the wallpaper and the whole screen is filled with 
horizontal tearing/artifacts.  This is what makes me thing that this is an 
xorg breakage rather than FF or KDE.

KDE is of course heavier than XFCE or Fluxbox in graphic terms and this may be 
pushing things over the edge.

-- 
Regards,
Mick

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

Re: [gentoo-user] Can't kill Firefox!

[Michael Schreckenbauer]

On Saturday 23 July 2011 09:28:43 Mick wrote:
> > Well, have a look at http://www.mail-archive.com/ubuntu-
> > bugs@lists.ubuntu.com/msg1483058.html for an example how the open source
> > atidrivers can hang X with firefox. Note that compositing was enabled
> > here (compiz). And here's another one:
> > https://bugzilla.redhat.com/show_bug.cgi?id=436632
> Thanks for these links.  This machine has an ATI Radeon X600 (RV380) 3E50
> (PCIE).  I have noticed that when I hover over the K menu, or when I select
> Log out/Shut down, I lose the wallpaper and the whole screen is filled with
> horizontal tearing/artifacts.  This is what makes me thing that this is an
> xorg breakage rather than FF or KDE.

Here's a more recent one:
https://bugzilla.kernel.org/show_bug.cgi?id=38792
This is in kernel-bugzilla, but seems to be a bug in the userspace part of the 
driver.

> KDE is of course heavier than XFCE or Fluxbox in graphic terms and this may
> be pushing things over the edge.

That's what I think.
The ati-drivers are in heavy development, KMS and all the other new shiny 
things. There are a few options for xorg.conf you could try. I have read about 
some problems with ColorTiling, so maybe setting ColorTiling to false helps a 
bit. See https://wiki.archlinux.org/index.php/ATI for some details on how to 
do this.

Regards
Michael

Re: [gentoo-user] Can't kill Firefox!

[Volker Armin Hemmann]

On Friday 22 July 2011 16:39:41 Sebastian Beßler wrote:
> Am 22.07.2011 15:13, schrieb Volker Armin Hemmann:
> > maybe it is not KDE's fault when firefox is badly coded?
> 
> Lets think.
> KDE4+Firefox = X hangs and firefox can't be killed
> XFCE4+Firefox = no problems
> 
> Sure, it has to be firefox
> 
> Oh and a test 20 minutes ago showed
> 
> KDE4+Chromium = X hangs and chromium can't be killed
> XFCE4+Chromium = no problems
> 
> It looks like there is a pattern but I could be wrong.
> 
> Greetings
> 
> Sebastian Beßler

and you strace'd firefox, X or chromium to see where this stuff hangs, did 
you?

-- 
#163933

Re: [gentoo-user] Can't kill Firefox!

[Sebastian Beßler]

[-- Attachment #1: Type: text/plain, Size: 703 bytes --]

Am 22.07.2011 17:05, schrieb Volker Armin Hemmann:

> and you strace'd firefox, X or chromium to see where this stuff hangs, did 
> you?

No I did not, a strace is only as good as the person who can read it,
and in my case is that not much, not because I'm too stupid but because
I rather let my drill my teeth before read megabytes of strace output. I
went the easy way, used my time for something useful and fun (analysing
strace output is neither useful nor fun for me, it may be for someone
else) and switched to XFCE4.

Hurray it works like a charm now so everything is good.

I didn't need all that KDE4 cruft, eye candy and semantic crap anyway.

Greetings

Sebastian Beßler


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 900 bytes --]

Re: [gentoo-user] Can't kill Firefox!

[Mick]

[-- Attachment #1: Type: Text/Plain, Size: 982 bytes --]

On Friday 22 Jul 2011 17:18:40 Sebastian Beßler wrote:
> Am 22.07.2011 17:05, schrieb Volker Armin Hemmann:
> > and you strace'd firefox, X or chromium to see where this stuff hangs,
> > did you?
> 
> No I did not, a strace is only as good as the person who can read it,
> and in my case is that not much, not because I'm too stupid but because
> I rather let my drill my teeth before read megabytes of strace output. I
> went the easy way, used my time for something useful and fun (analysing
> strace output is neither useful nor fun for me, it may be for someone
> else) and switched to XFCE4.
> 
> Hurray it works like a charm now so everything is good.
> 
> I didn't need all that KDE4 cruft, eye candy and semantic crap anyway.
> 
> Greetings
> 
> Sebastian Beßler

I haven't straced FF because it takes hours before it freezes X.  I'll see if 
it happens again tomorrow and if I can kill it from a console when I login as 
root.
-- 
Regards,
Mick

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

Re: [gentoo-user] Can't kill Firefox!

[Volker Armin Hemmann]

On Thursday 21 July 2011 19:38:53 Mick wrote:
> Whatever Dale had on his machine must have infected mine!  LOL!
> 
> A 32bit x86 box with KDE4.6, running firefox-3.6.17 and xulrunner-1.9.2.17
> after a few hours and loads of tabs (sometimes up to 15 or so) eventually
> hangs X.
> 
> I can switch to a console and login as the user running the X session, then
> try to kill the bloody thing with killall, kill -15 <pid> and kill -9 <pid>.
> It won't barge.  Mind you I haven't tried this as root.
> 
> How can this be?

don't know, but the next time this happens try to strace firefox.

-- 
#163933