[gentoo-dev] Serious: XFree halts system with some C++ codes running (in particular, Diffpack) on SMP kernel 2.4.19-gentoo-r7

[gentoo-dev] Serious: XFree halts system with some C++ codes running (in particular, Diffpack) on SMP kernel 2.4.19-gentoo-r7

[Denis Shcherbakov]

Hello,

I have a very freaky problem here.  I recently ran a few major updates,
such as glibc, threading libraries, some other libraries, and XFree86
4.2.1.  Now the C++ codes that I have to run (in a package called
Diffpack) invoke total system halt.  Screen turns off and goes into
power-save mode, and hard-drive light comes up.  All daemons hang (i can
no longer ssh to my PC after such crash).  With the older X, 4.2.0, I
observed the same kind of system halt, but it was not as frequent as it is
now.  Now, the code runs for several minutes, and the system crashes.

When I booted in TEXT mode, without starting X, this problem DID NOT
OCCUR!!!!  I have no idea what to do.  I got Nvidia GeForce3 chipset in my
system, and I use nvidia's GLX and kernel modules provided with Gentoo.  I
can provide my XF86Config file to anyone who'd be willing to look at it.
I disabled DPMS in the extensions module, but the problem persists.  I
wonder if there are other extension modules I should disable.  I also use
the option "NvAGP" "1", but I have crashes without it just the same.

The codes are run remotely on my machine by one of the users via SSH.

This is the first time I've ever seen anything like this.  Ever since I
installed Gentoo in August 2002, I had not a single major problem!

System details:  2 2.2 GHz Intel Xeons with hyperthreading technology,
Tyan motherboard, 2 IDE Western Digital hard-drives, nVidia GeForce3
chipset on ASUS V8200 card.  I have 2.4.19-gentoo-r7 kernel with SMP
enabled (never had any problems with the kernel before).  XFree 4.2.1,
latest glibc, and other libraries relating to C and C++.

ANY help would be greatly appreciated!!  I really have no idea what to do
and I really do not have time to try to solve this on my own!!

Thank you in advance!
Denis


--
gentoo-dev@gentoo.org mailing list

[gentoo-dev] XFree halts system with some C++ codes running (in particular, Diffpack) on SMP kernel 2.4.19-gentoo-r7

[Werner Van Belle]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

This sounds like a hardware problem.

If you have a dell, contact dell-support, and let them replace your 
power-supply and io-motherboard. I had similar things happening over here.

With kind regards,

Werner,-


On Tuesday 17 December 2002 16:35, Denis Shcherbakov wrote:
> When I booted in TEXT mode, without starting X, this problem DID NOT
> OCCUR!!!!  I have no idea what to do.  I got Nvidia GeForce3 chipset in my
> system, and I use nvidia's GLX and kernel modules provided with Gentoo.  I
> can provide my XF86Config file to anyone who'd be willing to look at it.
> I disabled DPMS in the extensions module, but the problem persists.  I
> wonder if there are other extension modules I should disable.  I also use
> the option "NvAGP" "1", but I have crashes without it just the same.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (GNU/Linux)

iD8DBQE9/0WnxiU3uL/NdUYRApzXAJ90PKy15RsdcDTdY4Q0Nc2SVB3ilQCgkgvS
43PhQ5vm3R8i93dGUqal7CE=
=S4y9
-----END PGP SIGNATURE-----


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Serious: XFree halts system with some C++ codes running (in particular, Diffpack) on SMP kernel 2.4.19-gentoo-r7

[Denis Shcherbakov]

Thanks for your reply!

But the thing is - I have NEVER had any crashes like that!  Even when
running graphics-intensive software on Linux.  This new package (Diffpack)
has been installed manually, without an ebuild, so there could be some
broken dependencies or something.  The package has some sort of a
visualization in it, and this could be what's breaking my X.

There is a possibility that this is a hardware issue, but then how come it
never happened before I started using this new C++ package!

I would like to hear what some of the other guru's think before I go
replacing my hardware.

Thanks,
Denis


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Serious: XFree halts system with some C++ codes running (in particular, Diffpack) on SMP kernel 2.4.19-gentoo-r7

[Simon Keimer]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

hello Denis,

I had the same problem some time ago, together with the nvidia guys, I found 
that the issue is a memory leak in the X libs.
Adding

    Option "NoRenderAccel" "true"

to your XF86Config's Device section should fix that for now.

Note: This does not disable the 3D-, only the 2D rendering-acceleration 
(that's what they told me)


> Hello,
>
> I have a very freaky problem here.  I recently ran a few major updates,
> such as glibc, threading libraries, some other libraries, and XFree86
> 4.2.1.  Now the C++ codes that I have to run (in a package called
> Diffpack) invoke total system halt.  Screen turns off and goes into
> power-save mode, and hard-drive light comes up.  All daemons hang (i can
> no longer ssh to my PC after such crash).  With the older X, 4.2.0, I
> observed the same kind of system halt, but it was not as frequent as it is
> now.  Now, the code runs for several minutes, and the system crashes.
>
> When I booted in TEXT mode, without starting X, this problem DID NOT
> OCCUR!!!!  I have no idea what to do.  I got Nvidia GeForce3 chipset in my
> system, and I use nvidia's GLX and kernel modules provided with Gentoo.  I
> can provide my XF86Config file to anyone who'd be willing to look at it.
> I disabled DPMS in the extensions module, but the problem persists.  I
> wonder if there are other extension modules I should disable.  I also use
> the option "NvAGP" "1", but I have crashes without it just the same.
>
> The codes are run remotely on my machine by one of the users via SSH.
>
> This is the first time I've ever seen anything like this.  Ever since I
> installed Gentoo in August 2002, I had not a single major problem!
>
> System details:  2 2.2 GHz Intel Xeons with hyperthreading technology,
> Tyan motherboard, 2 IDE Western Digital hard-drives, nVidia GeForce3
> chipset on ASUS V8200 card.  I have 2.4.19-gentoo-r7 kernel with SMP
> enabled (never had any problems with the kernel before).  XFree 4.2.1,
> latest glibc, and other libraries relating to C and C++.
>
> ANY help would be greatly appreciated!!  I really have no idea what to do
> and I really do not have time to try to solve this on my own!!
>
> Thank you in advance!
> Denis
>
>
> --
> gentoo-dev@gentoo.org mailing list

- -- 

Simon Keimer
skeimer@gmx.net

Help fighting the software patents delusion!
http://petition.eurolinux.org/
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (GNU/Linux)

iD8DBQE9/3WpceCdIlIaYYgRAjvUAKCbI9/x1BURD3qmuGobuYRjFS0AsQCfQqH2
jqJuAG6+IbrpDTa7yxNnAc4=
=ZNKX
-----END PGP SIGNATURE-----


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Serious: XFree halts system with some C++ codes running (in particular, Diffpack) on SMP kernel 2.4.19-gentoo-r7

[Denis Shcherbakov]

Hello,

So I tried the options suggested previously.  Those were to rock the video
card in its slot to make sure it's well-connected and dust-free and to set
"NoRenderAccel" "true" in the device options of XF86Config.  With these
options set, the system was still getting broken from executing the code.
However, it seems to have been resolved temporarily by switching off the
use of X libraries and of X-Motif libraries by the code.  Diffpack is a
comprehensive package for solving differential equations numerically, and
graphical interface can be written to it.  But since our code was simply
text-based, the use of these libraries was not necessary.  Diffpack comes
with outdated versions of Tcl/tk and some of their utilities, and so quite
possibly they are not meant to be used with new versions of XFree86.

I have not had any other problems with Motif or Tcl/Tk on my system except
for this package.

Hardware conflict is probably what happens here.  Not sure why.  Maybe
some broken dependencies within that software are triggering something in
that regard.

Thank you all for your help with this issue!  It's very much appreciated.

Denis





--
gentoo-dev@gentoo.org mailing list

[gentoo-dev] SMP kernel halts system - hyperthreading??

[Denis Shcherbakov]

Dear gentoo-dev,

If some of you recall, I wrote to the list about a system halt resulting
from running C++ codes and it possibly having to do with nVidia graphics
or XFree 4.2.1.

Well, the problem persists in text mode also.  It happens when 3 or more
CPU-intensive processes run simultaneously on the machine.  I have an SMP
kernel (2.4.19-gentoo-r7) with 2 Intel Xeon 2.2-GHz processors equipped
with hyperthreading technology.

I have 4 processors initialized at boot time, and 4 processors show up in
"top".  This is probably because of hyperthreading, as otherwise there
should be only two.

How do I turn off hyperthreading?  Do you think hyperthreading is what's
breaking my system and it could be more stable without hyperthreading?
Can a certain process be confined only to a certain processor?

Thanks for any help!
Denis


--
gentoo-dev@gentoo.org mailing list

RE: [gentoo-dev] SMP kernel halts system - hyperthreading??

[Riyad Kalla]

Denis,
That sounds like the server at work I'm using, nice machine..

Anywya, there was a thread on something like this in the forums, and it
basically turned out to be the SMP Gentoo kernel, the guy went to
vanilla kernel and everything worked just dandy. Can you verify this?
Maybe the kernel team needs to focus on something?

Can you run 4 threads of Seti@Home to see if it kills the machine? That
should be a good test as it will saturate your CPUs

Best,
-Riyad

-----Original Message-----
From: Denis Shcherbakov [mailto:deniss@Princeton.EDU] 
Sent: Wednesday, December 18, 2002 8:47 PM
To: gentoo-dev@gentoo.org
Subject: [gentoo-dev] SMP kernel halts system - hyperthreading??



Dear gentoo-dev,

If some of you recall, I wrote to the list about a system halt resulting
from running C++ codes and it possibly having to do with nVidia graphics
or XFree 4.2.1.

Well, the problem persists in text mode also.  It happens when 3 or more
CPU-intensive processes run simultaneously on the machine.  I have an
SMP
kernel (2.4.19-gentoo-r7) with 2 Intel Xeon 2.2-GHz processors equipped
with hyperthreading technology.

I have 4 processors initialized at boot time, and 4 processors show up
in
"top".  This is probably because of hyperthreading, as otherwise there
should be only two.

How do I turn off hyperthreading?  Do you think hyperthreading is what's
breaking my system and it could be more stable without hyperthreading?
Can a certain process be confined only to a certain processor?

Thanks for any help!
Denis


--
gentoo-dev@gentoo.org mailing list


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] SMP kernel halts system - hyperthreading??

[Marko Mikulicic]

Riyad Kalla wrote:
> Denis,
> That sounds like the server at work I'm using, nice machine..
> 
> Anywya, there was a thread on something like this in the forums, and it
> basically turned out to be the SMP Gentoo kernel, the guy went to
> vanilla kernel and everything worked just dandy. Can you verify this?
> Maybe the kernel team needs to focus on something?
> 
> Can you run 4 threads of Seti@Home to see if it kills the machine? That
> should be a good test as it will saturate your CPUs
> 
> Best,
> -Riyad

I have a 2x2.4Ghz Xeon with Hyperthreading enabled on a 2.4.19-xfs 
kernel and everything works great.
  However hyperthreading can be disabled in the BIOS. (when I bought the 
machine HT was disabled indeed ....). Dennis, do you have a E7500 
motherboard ?

Marko



--
gentoo-dev@gentoo.org mailing list