[uml-devel] 2.6.3-rc2-uml1 Kernel mode panic

[uml-devel] 2.6.3-rc2-uml1 Kernel mode panic

[William Stearns]

Good evening, Jeff,
	I got a Kernel panic from the above kernel; sorry to say my 
compiler won't let me include all the debugging information that would 
have made this report useful.  :-(

Kernel panic: Kernel mode fault at addr 0x0, ip 0x40101f94

	Sorry I don't have anything better.
	Cheers,
	- Bill

---------------------------------------------------------------------------
	Opinions expressed are my own, but they should be everybody's. 
(Courtesy of Doug Ledford <dledford@dialnet.net>) 
--------------------------------------------------------------------------
William Stearns (wstearns@pobox.com).  Mason, Buildkernel, freedups, p0f,
rsync-backup, ssh-keyinstall, dns-check, more at:   http://www.stearns.org
--------------------------------------------------------------------------



-------------------------------------------------------
SF.Net is sponsored by: Speed Start Your Linux Apps Now.
Build and deploy apps & Web services for Linux with
a free DVD software kit from IBM. Click Now!
http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel

Re: [uml-devel] 2.6.3-rc2-uml1 Kernel mode panic

[BlaisorBlade]

Alle 04:06, domenica 15 febbraio 2004, William Stearns ha scritto:
> Good evening, Jeff,
> 	I got a Kernel panic from the above kernel; sorry to say my
> compiler won't let me include all the debugging information that would
> have made this report useful.  :-(
>
> Kernel panic: Kernel mode fault at addr 0x0, ip 0x40101f94

Could you run with that kernel this loop on a shell:

while cat /dev/null; do cat /dev/null; done

This loop deterministically panics 2.6.2 and 2.6.3-rc2 UML kernels I have 
here(and if the configuration stays the same, even the corrupted values stay 
the same!). However nobody has yet seen it elsewhere.

Also this one "works":
while /bin/true; do /bin/true; done
(please avoid replacing /bin/true with true, since the second one is a shell 
builtin while I need a lot of forks to crash UML).

-- 
Paolo Giarrusso, aka Blaisorblade
Linux registered user n. 292729



-------------------------------------------------------
SF.Net is sponsored by: Speed Start Your Linux Apps Now.
Build and deploy apps & Web services for Linux with
a free DVD software kit from IBM. Click Now!
http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel

Re: [uml-devel] 2.6.3-rc2-uml1 Kernel mode panic

[William Stearns]

Good afternoon, all,

On Tue, 17 Feb 2004, BlaisorBlade wrote:

> Alle 04:06, domenica 15 febbraio 2004, William Stearns ha scritto:
> > 	I got a Kernel panic from the above kernel; sorry to say my
> > compiler won't let me include all the debugging information that would
> > have made this report useful.  :-(
> >
> > Kernel panic: Kernel mode fault at addr 0x0, ip 0x40101f94
> 
> Could you run with that kernel this loop on a shell:
> 
> while cat /dev/null; do cat /dev/null; done
> 
> This loop deterministically panics 2.6.2 and 2.6.3-rc2 UML kernels I have 
> here(and if the configuration stays the same, even the corrupted values stay 
> the same!). However nobody has yet seen it elsewhere.
> 
> Also this one "works":
> while /bin/true; do /bin/true; done
> (please avoid replacing /bin/true with true, since the second one is a shell 
> builtin while I need a lot of forks to crash UML).

	On a kernel using Jeff's patch from 2/16, which looks absolutely 
identical to yours,  20 of these running in the background * produces:
Kernel panic: Kernel BUG at kernel/exit.c:793
	or
Kernel panic: Kernel BUG at kernel/exit.c:793
Kernel panic: Segfault with no mm
Kernel panic: Segfault with no mm
Kernel panic: Segfault with no mm
	or
Kernel mode fault at addr 0x0, ip 0x401e643e
Kernel mode fault at addr 0x0, ip 0x401e643e
Kernel mode fault at addr 0x0, ip 0x401e643e
Kernel mode fault at addr 0x0, ip 0x401e643e

	For reference, when I revert to the kernel without that patch and
run 20 of those *, I got 8 "Kernel panic: segfault with no mm" within a
few minutes too.
	Both kernels are available from http://www.stearns.org/uml/ ; 
they're the ones eding in -gcc333.
	Cheers,
	- Bill

* Along with a continuous loop from the host of:
while : ; do ssh vmname echo vmname ; done

---------------------------------------------------------------------------
	"I would give up my right arm before I'd give up my netfilter
firewall."
	-- Chris Brenton
--------------------------------------------------------------------------
William Stearns (wstearns@pobox.com).  Mason, Buildkernel, freedups, p0f,
rsync-backup, ssh-keyinstall, dns-check, more at:   http://www.stearns.org
--------------------------------------------------------------------------



-------------------------------------------------------
SF.Net is sponsored by: Speed Start Your Linux Apps Now.
Build and deploy apps & Web services for Linux with
a free DVD software kit from IBM. Click Now!
http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel

Re: [uml-devel] 2.6.3-rc2-uml1 Kernel mode panic

[BlaisorBlade]

Alle 22:56, mercoledì 18 febbraio 2004, William Stearns ha scritto:
> Good afternoon, all,
>

>
> 	On a kernel using Jeff's patch from 2/16, which looks absolutely
> identical to yours, 

Yes, I just reposted it to roland, but it was the same.

> 20 of these running in the background * produces:
> Kernel panic: Kernel BUG at kernel/exit.c:793
> 	or
> Kernel panic: Kernel BUG at kernel/exit.c:793

If you get this with the patch applied, then the problem is very serious. I 
thought the patch would have been able to fix this second part (when I got 
this, I received two panics, first the "Kernel mode fault" and then this 
"Kernel BUG" (for more info look where it says: comments explains what 
happens).

> Kernel panic: Segfault with no mm
> Kernel panic: Segfault with no mm
> Kernel panic: Segfault with no mm
> 	or
> Kernel mode fault at addr 0x0, ip 0x401e643e
> Kernel mode fault at addr 0x0, ip 0x401e643e
> Kernel mode fault at addr 0x0, ip 0x401e643e
> Kernel mode fault at addr 0x0, ip 0x401e643e

You said you provide the kernel at:
http://www.stearns.org/uml/

Since I'm not able to upload a kernel while you seem to have broad band, could 
you please (if it is not already so) provide a kernel with debugging symbols 
enabled and showing this bug to Jeff Dike, since it seems he cannot reproduce 
the bug?
-- 
Paolo Giarrusso, aka Blaisorblade
Linux registered user n. 292729



-------------------------------------------------------
SF.Net is sponsored by: Speed Start Your Linux Apps Now.
Build and deploy apps & Web services for Linux with
a free DVD software kit from IBM. Click Now!
http://ads.osdn.com/?ad_id\x1356&alloc_id438&opÌk
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel

Re: [uml-devel] 2.6.3-rc2-uml1 Kernel mode panic

[William Stearns]

Good evening, BlaisorBlade,

On Thu, 19 Feb 2004, BlaisorBlade wrote:

> Alle 22:56, mercoledì 18 febbraio 2004, William Stearns ha scritto:
> > 	On a kernel using Jeff's patch from 2/16, which looks absolutely
> > identical to yours, 
> 
> Yes, I just reposted it to roland, but it was the same.

	Ah, that makes more sense.

> You said you provide the kernel at:
> http://www.stearns.org/uml/

	I do.

> Since I'm not able to upload a kernel while you seem to have broad band, could 

	Nope, IDSL, 144kbps up and down.  I'm just really efficient at 
uploading a 665M website to 8 mirrors.  :-)

> you please (if it is not already so) provide a kernel with debugging symbols 
> enabled and showing this bug to Jeff Dike, since it seems he cannot reproduce 
> the bug?

	The kernel already has debugging symbols.  Jeff also has access to 
the source tree used to build it (Jeff, zaphod:/usr/src/262/ ).
	Cheers,
	- Bill

---------------------------------------------------------------------------
	"As a computer I find your faith in technology amusing."
(Courtesy of Gerhard Mack <gmack@imag.net>)
--------------------------------------------------------------------------
William Stearns (wstearns@pobox.com).  Mason, Buildkernel, freedups, p0f,
rsync-backup, ssh-keyinstall, dns-check, more at:   http://www.stearns.org
--------------------------------------------------------------------------



-------------------------------------------------------
SF.Net is sponsored by: Speed Start Your Linux Apps Now.
Build and deploy apps & Web services for Linux with
a free DVD software kit from IBM. Click Now!
http://ads.osdn.com/?ad_id\x1356&alloc_id438&opÌk
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel

Re: [uml-devel] 2.6.3-rc2-uml1 Kernel mode panic

[BlaisorBlade]

Alle 20:35, giovedì 19 febbraio 2004, Jeff Dike ha scritto:

> I can reproduce it fine.  The recipe that Bill provided works like a charm
> - hit it from the outside with 'ssh uml some command'
> 	run 'while true; do cat /dev/null; done' inside
>
> and it will panic in 5-10 minutes in one of several ways.  The exit.c BUG
> is the process returning from the schedule at the end of exit().

Exactly the same here.

> The
> kernel mode fault is the scheduler seeing that the runqueue bitmap says
> there's a process to be scheduled waiting on a particular queue, but that
> queue is empty.

On my system the queue is not simply empty (i.e. queue[index] == 
queue[index]->next == queue[index]->prev) but has wrong next and prev values 
(one is 0x78 and one 0x0 last time, IIRC; however values <0x100). For the 
rest, it's the same as I get.

Now you can reproduce it, we'll soon see it fixed!

Bye
-- 
Paolo Giarrusso, aka Blaisorblade
Linux registered user n. 292729



-------------------------------------------------------
SF.Net is sponsored by: Speed Start Your Linux Apps Now.
Build and deploy apps & Web services for Linux with
a free DVD software kit from IBM. Click Now!
http://ads.osdn.com/?ad_id\x1356&alloc_id438&opÌk
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel

Re: [uml-devel] 2.6.3-rc2-uml1 Kernel mode panic

[Jeff Dike]

On Thu, Feb 19, 2004 at 07:14:01PM +0100, BlaisorBlade wrote:
> Since I'm not able to upload a kernel while you seem to have broad band, could 
> you please (if it is not already so) provide a kernel with debugging symbols 
> enabled and showing this bug to Jeff Dike, since it seems he cannot reproduce 
> the bug?

I can reproduce it fine.  The recipe that Bill provided works like a charm -
	hit it from the outside with 'ssh uml some command'
	run 'while true; do cat /dev/null; done' inside

and it will panic in 5-10 minutes in one of several ways.  The exit.c BUG
is the process returning from the schedule at the end of exit().  The 
kernel mode fault is the scheduler seeing that the runqueue bitmap says
there's a process to be scheduled waiting on a particular queue, but that
queue is empty.

Since the scheduler is involved in both of these, it's non-deterministic,
and seems to need a supply of interrupts to happen, it smells like the
scheduler is being interrupted when it shouldn't be.

				Jeff



-------------------------------------------------------
SF.Net is sponsored by: Speed Start Your Linux Apps Now.
Build and deploy apps & Web services for Linux with
a free DVD software kit from IBM. Click Now!
http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel