[uml-devel] Unable to handle kernel paging request ...

[uml-devel] Unable to handle kernel paging request ...

[Michael Koehne]

Moin Guru's,

  i have 2.6.x host crashes, and I dont know, if they are related to
  UML. I have a perfectly running 2.4.22 SKAS system on a dual PPro,
  and a laptop who requires a 2.6.x kernel because of hardware. This
  laptop is prepared to run UML's - but it seldom does.

  But the laptop crashes about once or twice a week. This sometimes
  happens on operations like `find . -print | while read x; do diff
  ../foo/$x $x; done` or rsync over large directory trees. And sometimes
  on idle state - e.g. tonight at crontime with a 'postmaster' crash.
  I've talked to people on IRC, and they dont encounter those kind of
  ext3 crashes on 2.6 kernels. I dont had a camera for a screenshot,
  but dump was in ext3 area.

  So it might be something with SKAS host patch - even if the system
  also crashes, if no UML is running.

     33240033 Feb  3 06:41 linux-2.6.1.tar.bz2
   +    15570 Feb  4 02:36 host-2.6.1-skas3.patch
  or 34896138 May 25 20:45 linux-2.6.6.tar.bz2
   +    16182 May 25 20:35 host-skas3-2.6.6-v1.patch

Bye Michael
-- 
  mailto:kraehe@copyleft.de             UNA:+.? 'CED+2+:::Linux:2.4.22'UNZ+1'
  http://www.xml-edifact.org/           CETERUM CENSEO WINDOWS ESSE DELENDAM


-------------------------------------------------------
This SF.Net email sponsored by Black Hat Briefings & Training.
Attend Black Hat Briefings & Training, Las Vegas July 24-29 - 
digital self defense, top technical experts, no vendor pitches, 
unmatched networking opportunities. Visit www.blackhat.com
_______________________________________________
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] Unable to handle kernel paging request ...

[BlaisorBlade]

Alle 11:06, mercoledì 7 luglio 2004, Michael Koehne ha scritto:
> Moin Guru's,
>
>   i have 2.6.x host crashes, and I dont know, if they are related to
>   UML. I have a perfectly running 2.4.22 SKAS system on a dual PPro,
>   and a laptop who requires a 2.6.x kernel because of hardware. This
>   laptop is prepared to run UML's - but it seldom does.

>   But the laptop crashes about once or twice a week. This sometimes
>   happens on operations like `find . -print | while read x; do diff
>   ../foo/$x $x; done` or rsync over large directory trees. And sometimes
>   on idle state - e.g. tonight at crontime with a 'postmaster' crash.
>   I've talked to people on IRC, and they dont encounter those kind of
>   ext3 crashes on 2.6 kernels. I dont had a camera for a screenshot,
>   but dump was in ext3 area.
What you mean with "ext3 crash"? "The dump was in ext3 area"? I don't 
understand you, sorry.

>   So it might be something with SKAS host patch - even if the system
>   also crashes, if no UML is running.
Well, the UML patch should have no influence, if you don't run UML on it (if 
it happens after you have run UML at least once, there is still a chance
>      33240033 Feb  3 06:41 linux-2.6.1.tar.bz2
>    +    15570 Feb  4 02:36 host-2.6.1-skas3.patch
>   or 34896138 May 25 20:45 linux-2.6.6.tar.bz2
>    +    16182 May 25 20:35 host-skas3-2.6.6-v1.patch

Well, since you experience them so often, try compiling a vanilla kernel, 
without the SKAS patch. Also, I'm highly suspicious of the 2.6.1 patch: it's 
likely it is an old, buggy version. Is that from Stephen D. Williams, 
distributed at this URL: http://sdw.st/src/uml/linux-2.6.1-skas3.patch? That 
is a buggy one, known to cause crashes... the second one is safe enough 
(though somebody experience memory leaks... as reported on 
http://www.user-mode-linux.org/~blaisorblade; however, they seldom happen).

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



-------------------------------------------------------
This SF.Net email sponsored by Black Hat Briefings & Training.
Attend Black Hat Briefings & Training, Las Vegas July 24-29 - 
digital self defense, top technical experts, no vendor pitches, 
unmatched networking opportunities. Visit www.blackhat.com
_______________________________________________
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] Unable to handle kernel paging request ...

[BlaisorBlade]

Alle 11:06, mercoledì 7 luglio 2004, Michael Koehne ha scritto:
> Moin Guru's,
>
>   i have 2.6.x host crashes, and I dont know, if they are related to
>   UML. I have a perfectly running 2.4.22 SKAS system on a dual PPro,
>   and a laptop who requires a 2.6.x kernel because of hardware. This
>   laptop is prepared to run UML's - but it seldom does.
>
>   But the laptop crashes about once or twice a week. This sometimes
>   happens on operations like `find . -print | while read x; do diff
>   ../foo/$x $x; done` or rsync over large directory trees. And sometimes
>   on idle state - e.g. tonight at crontime with a 'postmaster' crash.
>   I've talked to people on IRC, and they dont encounter those kind of
>   ext3 crashes on 2.6 kernels. I dont had a camera for a screenshot,
>   but dump was in ext3 area.

You say in the title "Unable to handle kernel paging request". It means you 
got a panic... Have you saved somewhere the complete message? Have you 
checked if it's inside /var/log/messages? Does "dump was in ext3 area" means 
that the call trace mentioned ext3_ <something>? If it speaks about 
__init_new_context, proc_mm or such things then the SKAS patch is likely to 
be at fault, if not it's unlikely...
-- 
Paolo Giarrusso, aka Blaisorblade
Linux registered user n. 292729



-------------------------------------------------------
This SF.Net email sponsored by Black Hat Briefings & Training.
Attend Black Hat Briefings & Training, Las Vegas July 24-29 - 
digital self defense, top technical experts, no vendor pitches, 
unmatched networking opportunities. Visit www.blackhat.com
_______________________________________________
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] Unable to handle kernel paging request ...

[Michael Koehne]

Moin BlaisorBlade,

> >   I've talked to people on IRC, and they dont encounter those kind of
> >   ext3 crashes on 2.6 kernels. I dont had a camera for a screenshot,
> >   but dump was in ext3 area.
> What you mean with "ext3 crash"? "The dump was in ext3 area"? I don't 
> understand you, sorry.

  the crash 2.6.1+SKAS happened on file system operations on the host,
  while UML systems had been up and running - I first seen them once
  on find, and twice on rsync directory file walks on the host. I
  upgraded to 2.6.6+SKAS and they occured seldom. My server moved
  housing also, and is now hosting the UMLs. So I only know that its
  a seldom condition that is ext3_ related, but might be related to
  the UML patch - as others dont tell about so many panics - but does
  not require an UML to touch /proc/mm.

  Yesterday I noticed some points at the kernel panic, that started at
  postmaster at crontab time, and had a lot of ext3 pointers - but I
  did'nt had the patiens to write 3 pages of kernel panic ;(

> Well, the UML patch should have no influence, if you don't run UML on it (if 
> it happens after you have run UML at least once, there is still a chance

  the patch host-skas3-2.6.6-v1.patch is applied and running - but 
  /proc/mm was virgin - as there was no UML started on this machine
  for some days.

> >   or 34896138 May 25 20:45 linux-2.6.6.tar.bz2
> >    +    16182 May 25 20:35 host-skas3-2.6.6-v1.patch

  I'll stick to that combo for some more days (because of laziness ;)
  and hope to be more patient writing down the kernel panic next time.

> the second one is safe enough 
> (though somebody experience memory leaks... as reported on 
> http://www.user-mode-linux.org/~blaisorblade; however, they seldom happen).
 
  *hm* whats the current 'best practice patch' for 2.6.x ? 

  an rsync copy of my server's UML to my laptop's UML is a nice thing I
  wont want to miss anymore, even if the Double-NAT between them was a
  bit confusing at first - and 2.4.22 SKAS server and 2.4.22 UML are
  stable like a rock - at least in my ubd over ext3 over md-raid1
  on a Dual processor environment.

> You say in the title "Unable to handle kernel paging request". It means you 
> got a panic... Have you saved somewhere the complete message? Have you 
> checked if it's inside /var/log/messages?
  
  its not in any logs - as it was ext3 related on the host file system,
  if I managed to understand the dump. The process was postmaster running
  from cron on the host - there was no UML running since boot and the
  trace mentioned a lot of ext3_ <something>.

> Does "dump was in ext3 area" means 
> that the call trace mentioned ext3_ <something>? If it speaks about 
> __init_new_context, proc_mm or such things then the SKAS patch is likely to 
> be at fault, if not it's unlikely...

  I'll look at those more closely if it happens again, any other
  pointers so I could sort out UML related areas here ?

  I'll also might find the time to try a more actual kernel+skas combo
  before - as the crash is seldom, on a machine, I'm using regulary for
  about 2-4 full load hours from compiler per day with normal background
  processes in the other time.

Bye Michael
-- 
  mailto:kraehe@copyleft.de             UNA:+.? 'CED+2+:::Linux:2.4.22'UNZ+1'
  http://www.xml-edifact.org/           CETERUM CENSEO WINDOWS ESSE DELENDAM


-------------------------------------------------------
This SF.Net email sponsored by Black Hat Briefings & Training.
Attend Black Hat Briefings & Training, Las Vegas July 24-29 - 
digital self defense, top technical experts, no vendor pitches, 
unmatched networking opportunities. Visit www.blackhat.com
_______________________________________________
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] Unable to handle kernel paging request ...

[BlaisorBlade]

Alle 05:31, giovedì 8 luglio 2004, Michael Koehne ha scritto:
> Moin BlaisorBlade,

>   I'll stick to that combo for some more days (because of laziness ;)
>   and hope to be more patient writing down the kernel panic next time.
If you can, try also using, instead, a vanilla kernel. I'm highly suspicious 
your crash is not SKAS-related.

If you get that panic with a kernel without any SKAS patch, then you should 
report it onto the LKML (Linux Kernel Mailing List), hopefully with a stack 
trace (yes, it's painful to write it down... - at least try providing the 
names of the funcs involved - for instance, in this panic output (I've 
received it a few minutes ago) There are only some interesting part (I marked 
them with >):

> Jun 27 12:11:25 nimbus kernel: Unable to handle kernel paging request at
virtual address 00100104

Jun 27 12:11:25 nimbus kernel:  printing eip:
Jun 27 12:11:25 nimbus kernel: c011b31f
Jun 27 12:11:25 nimbus kernel: *pde = 00000000
Jun 27 12:11:25 nimbus kernel: Oops: 0002 [#1]
Jun 27 12:11:25 nimbus kernel: SMP 
Jun 27 12:11:25 nimbus kernel: CPU:    0
Jun 27 12:11:25 nimbus kernel: EIP:    0060:[<c011b31f>]    Not tainted
Jun 27 12:11:25 nimbus kernel: EFLAGS: 00010202   (2.6.5040422) 

> Jun 27 12:11:25 nimbus kernel: EIP is at mmput+0x23/0x68

Jun 27 12:11:25 nimbus kernel: eax: 00100100   ebx: ee2d3e00   ecx: ee2d3e34
edx: 00200200
Jun 27 12:11:25 nimbus kernel: esi: f5804c60   edi: 00000037   ebp: 00000000
esp: f085df80
Jun 27 12:11:25 nimbus kernel: ds: 007b   es: 007b   ss: 0068

> Jun 27 12:11:25 nimbus kernel: Process linux-2.6.5-a1- (pid: 8864,
threadinfo=f085c000 task=c1b5f290)

Jun 27 12:11:25 nimbus kernel: Stack: ee2d3e00 c010d0d4 ee2d3e00 00000037
00000044 00000037 f085c000 f085c000 
Jun 27 12:11:25 nimbus kernel:        f5804c60 f636f080 a696b5e8 0000001c
c0149aa4 00000044 bfc00000 00000001 
Jun 27 12:11:25 nimbus kernel:        c0108573 00000037 000022a2 00000000
00000044 00000037 a696b65c 0000001a 

> And finally, these lines are VERY important (I don't care the <c010...> 
part, just the "sys_ptrace"-"sys_write"-"syscall_call" part.

Jun 27 12:11:25 nimbus kernel: Call Trace:
Jun 27 12:11:25 nimbus kernel:  [<c010d0d4>] sys_ptrace+0x754/0x7d0
Jun 27 12:11:25 nimbus kernel:  [<c0149aa4>] sys_write+0x44/0x4c
Jun 27 12:11:25 nimbus kernel:  [<c0108573>] syscall_call+0x7/0xb

Jun 27 12:11:25 nimbus kernel: 
Jun 27 12:11:25 nimbus kernel: Code: 89 50 04 89 02 c7 43 34 00 01 10 00 c7
41 04 00 02 20 00 ff

So you could write down, for that panic:
----
Unable to handle kernel paging request at
virtual address???

EIP is at mmput

Call Trace:

sys_ptrace
sys_write
syscall_call
----
The other part are harder to decode and much less useful, in this situation 
(they can be used, but it's difficult).
-- 
Paolo Giarrusso, aka Blaisorblade
Linux registered user n. 292729



-------------------------------------------------------
This SF.Net email sponsored by Black Hat Briefings & Training.
Attend Black Hat Briefings & Training, Las Vegas July 24-29 - 
digital self defense, top technical experts, no vendor pitches, 
unmatched networking opportunities. Visit www.blackhat.com
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel