On Wednesday 03 November 2004 08:53, Roland Kaeser wrote:
> Hello
>
> I starting the UML Kernel with the following command:
>
> /install/uml/kernel/vmlinux mem=128M devfs=nomount con=xterm
> eth0=tuntap,,,1.1.1.1 eth1=tuntap,,,2.2.2.2 root=/dev/root
> rootflags=/install/uml/instances/hostfc2install rootfstype=hostfs
>
> The system starts up successfully, and opens all the xterm for the
> consoles. I can login and work with the guest system.
> Then I shut down the guest system from inside the guest system using the
> command init 0 or shutdown.
> The guest shuts down normally.
> But in the moment when the uml kernel exits in the xterm where I started
> it, I CAN HEAR the harddisk of the host goes down.
Ok, with this explaination, I can believe that this happens. Sorry for my
first answer, I first took the most likely idea, i.e. the hard disk spins
down because it's no more used, but the host kernel is still alive and
kicking, and the hard drive will spin up when needed. Anyway, this
description, w
You are using a vanilla host kernel (2.6.9), so you have no SKAS patch
running, i.e. this is a host kernel bug, actually, and it is pretty severe.
Attach your .config in next email.
Can you please report that (including the kernel panic message) to LKML,
CC:ing the -devel list? Also, could you try that with different host kernels?
Different setups? Are you running your UML as root or not? If you are running
it as root, then try running it with an unprivileged userid.
> After this, nothing happens more on the host, even the mouse freezes. I
> cannot start an other program or even access the harddisk or login on a
> other tty.
> It freezes the whole guest system.
You mean obiously the "host" system (i.e. the hardware one; "guest" refers to
the UML instance).
> About a minute later I get a
> Kernel Panic from the host kernel
Try to copy and send the text of the panic, (most of the numbers can be
omitted, even because they are often hard to decode; the list of function
calls is more important than anything else).
--
Paolo Giarrusso, aka Blaisorblade
Linux registered user n. 292729
Hello
Thank You for Your answer. But my host kernel is a 2.6.7 with the skas patch. The
guest is a 2.6.9. Just an another info: The behaviour does not depend on the hostfs
configuration. It also happens when I use a ubd device.
And no, the HOST!! freezes after exit of the guest kernel. And i get a Kernel panic
from the HOST!! kernel, this in case the host (ide) harddisk drive spins down (but
not spins up anymore). My idea is that some routines to spin down the harddisk are
been routed outside the uml guest kernel or not been sucessfully removed for the uml
architecture. Is it possible that the /sbin/halt binary can have made something with
the hosts harddisk?
How can i get the kernel panic message from the host?
Roland
--- Blaisorblade <[email protected]> schrieb:
> On Wednesday 03 November 2004 08:53, Roland Kaeser wrote:
> > Hello
> >
> > I starting the UML Kernel with the following command:
> >
> > /install/uml/kernel/vmlinux mem=128M devfs=nomount con=xterm
> > eth0=tuntap,,,1.1.1.1 eth1=tuntap,,,2.2.2.2 root=/dev/root
> > rootflags=/install/uml/instances/hostfc2install rootfstype=hostfs
> >
> > The system starts up successfully, and opens all the xterm for the
> > consoles. I can login and work with the guest system.
> > Then I shut down the guest system from inside the guest system using the
> > command init 0 or shutdown.
>
> > The guest shuts down normally.
> > But in the moment when the uml kernel exits in the xterm where I started
> > it, I CAN HEAR the harddisk of the host goes down.
>
> Ok, with this explaination, I can believe that this happens. Sorry for my
> first answer, I first took the most likely idea, i.e. the hard disk spins
> down because it's no more used, but the host kernel is still alive and
> kicking, and the hard drive will spin up when needed. Anyway, this
> description, w
>
> You are using a vanilla host kernel (2.6.9), so you have no SKAS patch
> running, i.e. this is a host kernel bug, actually, and it is pretty severe.
> Attach your .config in next email.
>
> Can you please report that (including the kernel panic message) to LKML,
> CC:ing the -devel list? Also, could you try that with different host kernels?
> Different setups? Are you running your UML as root or not? If you are running
> it as root, then try running it with an unprivileged userid.
>
> > After this, nothing happens more on the host, even the mouse freezes. I
> > cannot start an other program or even access the harddisk or login on a
> > other tty.
>
> > It freezes the whole guest system.
> You mean obiously the "host" system (i.e. the hardware one; "guest" refers to
> the UML instance).
> > About a minute later I get a
> > Kernel Panic from the host kernel
> Try to copy and send the text of the panic, (most of the numbers can be
> omitted, even because they are often hard to decode; the list of function
> calls is more important than anything else).
> --
> Paolo Giarrusso, aka Blaisorblade
> Linux registered user n. 292729
>
___________________________________________________________
Gesendet von Yahoo! Mail - Jetzt mit 100MB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de
Hello
I've forgotten one point. The guest kernel creates a (how is the correct english
word for this?) a memory error (German Speicherfehler) so it seems that the kernel
crashes. I have to say that I run the guest system with root privileges (on the root
account).
Roland
--- Blaisorblade <[email protected]> schrieb:
> On Wednesday 03 November 2004 08:53, Roland Kaeser wrote:
> > Hello
> >
> > I starting the UML Kernel with the following command:
> >
> > /install/uml/kernel/vmlinux mem=128M devfs=nomount con=xterm
> > eth0=tuntap,,,1.1.1.1 eth1=tuntap,,,2.2.2.2 root=/dev/root
> > rootflags=/install/uml/instances/hostfc2install rootfstype=hostfs
> >
> > The system starts up successfully, and opens all the xterm for the
> > consoles. I can login and work with the guest system.
> > Then I shut down the guest system from inside the guest system using the
> > command init 0 or shutdown.
>
> > The guest shuts down normally.
> > But in the moment when the uml kernel exits in the xterm where I started
> > it, I CAN HEAR the harddisk of the host goes down.
>
> Ok, with this explaination, I can believe that this happens. Sorry for my
> first answer, I first took the most likely idea, i.e. the hard disk spins
> down because it's no more used, but the host kernel is still alive and
> kicking, and the hard drive will spin up when needed. Anyway, this
> description, w
>
> You are using a vanilla host kernel (2.6.9), so you have no SKAS patch
> running, i.e. this is a host kernel bug, actually, and it is pretty severe.
> Attach your .config in next email.
>
> Can you please report that (including the kernel panic message) to LKML,
> CC:ing the -devel list? Also, could you try that with different host kernels?
> Different setups? Are you running your UML as root or not? If you are running
> it as root, then try running it with an unprivileged userid.
>
> > After this, nothing happens more on the host, even the mouse freezes. I
> > cannot start an other program or even access the harddisk or login on a
> > other tty.
>
> > It freezes the whole guest system.
> You mean obiously the "host" system (i.e. the hardware one; "guest" refers to
> the UML instance).
> > About a minute later I get a
> > Kernel Panic from the host kernel
> Try to copy and send the text of the panic, (most of the numbers can be
> omitted, even because they are often hard to decode; the list of function
> calls is more important than anything else).
> --
> Paolo Giarrusso, aka Blaisorblade
> Linux registered user n. 292729
>
___________________________________________________________
Gesendet von Yahoo! Mail - Jetzt mit 100MB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de
On Wednesday 03 November 2004 20:41, Roland Kaeser wrote:
> Hello
>
> I've forgotten one point. The guest kernel creates a (how is the correct
> english word for this?) a memory error (German Speicherfehler) so it seems
> that the kernel crashes. I have to say that I run the guest system with
> root privileges (on the root account).
Could you first try running it on a normal account? If that workarounds the
problem, then the second try goes to trying a different root_fs.
--
Paolo Giarrusso, aka Blaisorblade
Linux registered user n. 292729
On Wednesday 03 November 2004 20:21, Roland Kaeser wrote:
> Hello
>
> Thank You for Your answer. But my host kernel is a 2.6.7 with the skas
> patch. The guest is a 2.6.9. Just an another info: The behaviour does not
> depend on the hostfs configuration. It also happens when I use a ubd
> device.
> And no, the HOST!! freezes after exit of the guest kernel.
> And i get a
> Kernel panic from the HOST!! kernel, this in case the host (ide) harddisk
> drive spins down (but not spins up anymore).
Yes, I already understood this.
> My idea is that some routines
> to spin down the harddisk are been routed outside the uml guest kernel or
> not been sucessfully removed for the uml architecture.
> Is it possible that
> the /sbin/halt binary can have made something with the hosts harddisk?
I don't think that /sbin/halt itself can do anything. I would be suspicious of
hdparm, instead. It would be a UML bug anyway, but it's the more likely thing
it's possible to cause this.
> How can i get the kernel panic message from the host?
If you are on a virtual console, you see them on your screen. Otherwise, how
did you guess you have a kernel panic? Messages on /var/log/messages (I guess
no)? Or what?
> Roland
> --- Blaisorblade <[email protected]> schrieb:
> > On Wednesday 03 November 2004 08:53, Roland Kaeser wrote:
> > > Hello
> > > I starting the UML Kernel with the following command:
> > > /install/uml/kernel/vmlinux mem=128M devfs=nomount con=xterm
> > > eth0=tuntap,,,1.1.1.1 eth1=tuntap,,,2.2.2.2 root=/dev/root
> > > rootflags=/install/uml/instances/hostfc2install rootfstype=hostfs
> > > The system starts up successfully, and opens all the xterm for the
> > > consoles. I can login and work with the guest system.
> > > Then I shut down the guest system from inside the guest system using
> > > the command init 0 or shutdown.
> > > The guest shuts down normally.
> > > But in the moment when the uml kernel exits in the xterm where I
> > > started it, I CAN HEAR the harddisk of the host goes down.
> > Ok, with this explaination, I can believe that this happens. Sorry for my
> > first answer, I first took the most likely idea, i.e. the hard disk spins
> > down because it's no more used, but the host kernel is still alive and
> > kicking, and the hard drive will spin up when needed. Anyway, this
> > description, w
> > You are using a vanilla host kernel (2.6.9), so you have no SKAS patch
> > running, i.e. this is a host kernel bug, actually, and it is pretty
> > severe. Attach your .config in next email.
> > Can you please report that (including the kernel panic message) to LKML,
> > CC:ing the -devel list? Also, could you try that with different host
> > kernels? Different setups? Are you running your UML as root or not? If
> > you are running it as root, then try running it with an unprivileged
> > userid.
> > > After this, nothing happens more on the host, even the mouse freezes. I
> > > cannot start an other program or even access the harddisk or login on a
> > > other tty.
> > > It freezes the whole guest system.
> > You mean obiously the "host" system (i.e. the hardware one; "guest"
> > refers to the UML instance).
> > > About a minute later I get a
> > > Kernel Panic from the host kernel
> > Try to copy and send the text of the panic, (most of the numbers can be
> > omitted, even because they are often hard to decode; the list of function
> > calls is more important than anything else).
--
Paolo Giarrusso, aka Blaisorblade
Linux registered user n. 292729
On Wed, Nov 03, 2004 at 08:21:55PM +0100, Roland Kaeser wrote:
> And no, the HOST!! freezes after exit of the guest kernel.
does this happen w/o the SKAS patch?
> And i get a Kernel panic from the HOST!! kernel, this in case the
> host (ide) harddisk drive spins down (but not spins up anymore).
oops details please
> My idea is that some routines to spin down the harddisk are been
> routed outside the uml guest kernel or not been sucessfully removed
> for the uml architecture.
unlikely, uml catches the ioctls hdparm uses (i have a patch for this
cleaning things up somewhere or maybe it got merged). uml shouldn't
propagate the ioctls out
> Is it possible that the /sbin/halt binary can have made something
> with the hosts harddisk?
not directly, but it might trigger something
> How can i get the kernel panic message from the host?
uml is triggering a host OS bug
Hello
To Your questions:
> does this happen w/o the SKAS patch?
I don't have tried it. I will do it today
> uml is triggering a host OS bug
A host OS bug? How can I get this out from the machine?
As I wrote in a later mail, at the time the harddisk spins down, the uml kernel
causes a memory error. (German: Speicherfehler) core dump.
I will try to setup a test machine which You can access to test it by Yourself, I
think i can't discribe it completely enough to give You a point for a solution.
Roland
--- Chris Wedgwood <[email protected]> schrieb:
> On Wed, Nov 03, 2004 at 08:21:55PM +0100, Roland Kaeser wrote:
>
> > And no, the HOST!! freezes after exit of the guest kernel.
>
>
> > And i get a Kernel panic from the HOST!! kernel, this in case the
> > host (ide) harddisk drive spins down (but not spins up anymore).
>
> oops details please
>
> > My idea is that some routines to spin down the harddisk are been
> > routed outside the uml guest kernel or not been sucessfully removed
> > for the uml architecture.
>
> unlikely, uml catches the ioctls hdparm uses (i have a patch for this
> cleaning things up somewhere or maybe it got merged). uml shouldn't
> propagate the ioctls out
>
> > Is it possible that the /sbin/halt binary can have made something
> > with the hosts harddisk?
>
> not directly, but it might trigger something
>
> > How can i get the kernel panic message from the host?
>
> uml is triggering a host OS bug
>
>
> -------------------------------------------------------
> This SF.Net email is sponsored by:
> Sybase ASE Linux Express Edition - download now for FREE
> LinuxWorld Reader's Choice Award Winner for best database on Linux.
> http://ads.osdn.com/?ad_id=5588&alloc_id=12065&op=click
> _______________________________________________
> User-mode-linux-user mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/user-mode-linux-user
>
___________________________________________________________
Gesendet von Yahoo! Mail - Jetzt mit 100MB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de
Hello
Here comes the Test report for the UML SKAS and User run Tests
If I run the UML Kernel without the HOSTS SKAS Patch, it works normally, no hd
shutdown!
If i run the UML Kernel with the HOSTS SKAS Patch but as "normal" user, it works
also normally, without any harddisk shutdown!
It seems more and more to be a kind of bug in the UML Patch which allows the uml
kernel to call kernel functions on the host kernel.
Roland
--- Blaisorblade <[email protected]> schrieb:
> On Wednesday 03 November 2004 20:21, Roland Kaeser wrote:
> > Hello
> >
> > Thank You for Your answer. But my host kernel is a 2.6.7 with the skas
> > patch. The guest is a 2.6.9. Just an another info: The behaviour does not
> > depend on the hostfs configuration. It also happens when I use a ubd
> > device.
>
> > And no, the HOST!! freezes after exit of the guest kernel.
>
> > And i get a
> > Kernel panic from the HOST!! kernel, this in case the host (ide) harddisk
> > drive spins down (but not spins up anymore).
> Yes, I already understood this.
>
> > My idea is that some routines
> > to spin down the harddisk are been routed outside the uml guest kernel or
> > not been sucessfully removed for the uml architecture.
>
> > Is it possible that
> > the /sbin/halt binary can have made something with the hosts harddisk?
> I don't think that /sbin/halt itself can do anything. I would be suspicious of
> hdparm, instead. It would be a UML bug anyway, but it's the more likely thing
> it's possible to cause this.
>
> > How can i get the kernel panic message from the host?
>
> If you are on a virtual console, you see them on your screen. Otherwise, how
> did you guess you have a kernel panic? Messages on /var/log/messages (I guess
> no)? Or what?
>
> > Roland
>
> > --- Blaisorblade <[email protected]> schrieb:
> > > On Wednesday 03 November 2004 08:53, Roland Kaeser wrote:
> > > > Hello
>
> > > > I starting the UML Kernel with the following command:
>
> > > > /install/uml/kernel/vmlinux mem=128M devfs=nomount con=xterm
> > > > eth0=tuntap,,,1.1.1.1 eth1=tuntap,,,2.2.2.2 root=/dev/root
> > > > rootflags=/install/uml/instances/hostfc2install rootfstype=hostfs
>
> > > > The system starts up successfully, and opens all the xterm for the
> > > > consoles. I can login and work with the guest system.
> > > > Then I shut down the guest system from inside the guest system using
> > > > the command init 0 or shutdown.
>
> > > > The guest shuts down normally.
> > > > But in the moment when the uml kernel exits in the xterm where I
> > > > started it, I CAN HEAR the harddisk of the host goes down.
>
> > > Ok, with this explaination, I can believe that this happens. Sorry for my
> > > first answer, I first took the most likely idea, i.e. the hard disk spins
> > > down because it's no more used, but the host kernel is still alive and
> > > kicking, and the hard drive will spin up when needed. Anyway, this
> > > description, w
>
> > > You are using a vanilla host kernel (2.6.9), so you have no SKAS patch
> > > running, i.e. this is a host kernel bug, actually, and it is pretty
> > > severe. Attach your .config in next email.
>
> > > Can you please report that (including the kernel panic message) to LKML,
> > > CC:ing the -devel list? Also, could you try that with different host
> > > kernels? Different setups? Are you running your UML as root or not? If
> > > you are running it as root, then try running it with an unprivileged
> > > userid.
>
> > > > After this, nothing happens more on the host, even the mouse freezes. I
> > > > cannot start an other program or even access the harddisk or login on a
> > > > other tty.
>
> > > > It freezes the whole guest system.
>
> > > You mean obiously the "host" system (i.e. the hardware one; "guest"
> > > refers to the UML instance).
>
> > > > About a minute later I get a
> > > > Kernel Panic from the host kernel
>
> > > Try to copy and send the text of the panic, (most of the numbers can be
> > > omitted, even because they are often hard to decode; the list of function
> > > calls is more important than anything else).
>
> --
> Paolo Giarrusso, aka Blaisorblade
> Linux registered user n. 292729
>
>
>
> -------------------------------------------------------
> This SF.Net email is sponsored by:
> Sybase ASE Linux Express Edition - download now for FREE
> LinuxWorld Reader's Choice Award Winner for best database on Linux.
> http://ads.osdn.com/?ad_id=5588&alloc_id=12065&op=click
> _______________________________________________
> User-mode-linux-user mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/user-mode-linux-user
>
___________________________________________________________
Gesendet von Yahoo! Mail - Jetzt mit 100MB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de
On Thursday 04 November 2004 10:41, Roland Kaeser wrote:
> Hello
>
> Here comes the Test report for the UML SKAS and User run Tests
>
> If I run the UML Kernel without the HOSTS SKAS Patch, it works normally, no
> hd shutdown!
>
> If i run the UML Kernel with the HOSTS SKAS Patch but as "normal" user, it
> works also normally, without any harddisk shutdown!
> It seems more and more to be a kind of bug in the UML Patch which allows
> the uml kernel to call kernel functions on the host kernel.
Also, there is a host kernel bug which causes the host to crash. Maybe (it is
just a random try, actually, but there should not be a lot of bugs in 2.6
kernels), you could try this: would you please read this thread and tell if
you can reproduce the described panic and if it is anyhow similar to the
other one.
http://groups.google.com/groups?hl=en&lr=&threadm=2WG11-2iJ-1%40gated-at.bofh.it&prev=/groups%3Fnum%3D25%26hl%3Den%26lr%3D%26group%3Dlinux.kernel%26start%3D50
> Roland
--
Paolo Giarrusso, aka Blaisorblade
Linux registered user n. 292729