[uml-devel] skas3-related crash of the week

[uml-devel] skas3-related crash of the week

[Nix]

Using unmodified 2.4.22-6. sshed in through the firewall on the UML
instance, with tty logging enabled: typing exit on the last sshed-in
console froze the box yielded this on the console before UML died:

Kernel panic: switch_mm_skas - PTRACE_SWITCH_MM failed, errno = 9

Kernel panic: switch_mm_skas - PTRACE_SWITCH_MM failed, errno = 9

Kernel panic: Kernel mode fault at addr 0x4, ip 0x400e64ce
Kernel panic: Kernel mode fault at addr 0x4, ip 0x400e64ce
Kernel panic: Kernel mode fault at addr 0x4, ip 0x400e64ce
Kernel panic: Kernel mode fault at addr 0x4, ip 0x400e64ce
Kernel panic: switch_mm_skas - PTRACE_SWITCH_MM failed, errno = 9

Kernel panic: Kernel mode fault at addr 0x4, ip 0x400e64ce
Kernel panic: Kernel mode fault at addr 0x4, ip 0x400e64ce
Kernel panic: Kernel mode fault at addr 0x4, ip 0x400e64ce
Kernel panic: Kernel mode fault at addr 0x4, ip 0x400e64ce
Kernel panic: Segfault with no mm
Kernel panic: Segfault with no mm
Kernel panic: Segfault with no mm
Kernel panic: Segfault with no mm
Kernel panic: Segfault with no mm
Kernel panic: Segfault with no mm
Kernel panic: Segfault with no mm
Kernel panic: Segfault with no mm
Kernel panic: Segfault with no mm
Kernel panic: Segfault with no mm
Kernel panic: Segfault with no mm
Kernel panic: Segfault with no mm
In idle task - not syncing
 <0>Kernel panic: Segfault with no mm
In idle task - not syncing

EBADF is a rather odd error to get from ptrace(): I wonder if this has
the same root cause as the EMFILE error I got a couple of weeks ago? It
looks like the error came from proc_mm_get_mm(), so something like
running out of fhs might explain it. Whatever it is, it's pretty
persistent: it looks like it panic-kills every process in turn until
it runs out.

As before, post-crash the box isn't remotely close to running out of
fhs:

1437    208     8192


(Now that the instance has crashed, I have an excuse to rebuild it with
debugging turned on, so the next report like this should have
backtraces.)

-- 
`Me, I want exploding spaceships and pulverized worlds and clashes of
 billion-year-old empires *and* competently written sentences.'
                                                    --- Matt Austern


-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?  SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/
_______________________________________________
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] skas3-related crash of the week

[Nuno Silva]

Hi!!

Nix wrote:
> Using unmodified 2.4.22-6. sshed in through the firewall on the UML
> instance, with tty logging enabled: typing exit on the last sshed-in
> console froze the box yielded this on the console before UML died:
> 
> Kernel panic: switch_mm_skas - PTRACE_SWITCH_MM failed, errno = 9
> 
> Kernel panic: switch_mm_skas - PTRACE_SWITCH_MM failed, errno = 9
> 

I got these some weeks ago. Didn't have time to investigate at the time :(

The odd thing is that this seems to happen only with UMLs running 
redhat. Debian UMLs never showed this...

The host is running debian stable.

I'll try to test UML with a RH filesystem again.

Regards,
Nuno Silva



-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?  SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/
_______________________________________________
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] skas3-related crash of the week

[Nix]

On Wed, 19 Nov 2003, Nuno Silva stipulated:
> The odd thing is that this seems to happen only with UMLs running
> redhat. Debian UMLs never showed this...

My UML is running a self-built-from-source stripped-down system with
handwritten boot scripts and firewall; no RH in sight.

So there might be some commonality between that FS and the RH root
filesystems that isn't shared by the Debian ones, but I'm not sure what
it could be.

-- 
`Me, I want exploding spaceships and pulverized worlds and clashes of
 billion-year-old empires *and* competently written sentences.'
                                                    --- Matt Austern


-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?  SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/
_______________________________________________
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] skas3-related crash of the week

[Nuno Silva]

Nix wrote:
> On Wed, 19 Nov 2003, Nuno Silva stipulated:
> 
>>The odd thing is that this seems to happen only with UMLs running
>>redhat. Debian UMLs never showed this...
> 
> 
> My UML is running a self-built-from-source stripped-down system with
> handwritten boot scripts and firewall; no RH in sight.
> 
> So there might be some commonality between that FS and the RH root
> filesystems that isn't shared by the Debian ones, but I'm not sure what
> it could be.
> 

Maybe it's just a coincidence... I'll try to reproduce this.

Regards,
Nuno Silva



-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?  SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/
_______________________________________________
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] skas3-related crash of the week

[Nuno Silva]

Nuno Silva wrote:
> Hi!!
> 
> Nix wrote:
> 
>> Using unmodified 2.4.22-6. sshed in through the firewall on the UML
>> instance, with tty logging enabled: typing exit on the last sshed-in
>> console froze the box yielded this on the console before UML died:
>>
>> Kernel panic: switch_mm_skas - PTRACE_SWITCH_MM failed, errno = 9
>>
>> Kernel panic: switch_mm_skas - PTRACE_SWITCH_MM failed, errno = 9
>>

I've found a log with the error, it's a bit different from yours:

Kernel panic: save_registers - saving registers failed, errno = 3

Kernel panic: switch_mm_skas - PTRACE_SWITCH_MM failed, errno = 3

In idle task - not syncing

--------------------------------

 From the screen(1) logs this was a few minutes after the boot and this 
UML should be mostly idle. I only have this log and no backtraces... not 
  very helpfull :(

Regards,
Nuno Silva



-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?  SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/
_______________________________________________
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] skas3-related crash of the week

[Jeff Chua]

try uml-patch-2.4.22-5. I couldn't get 2.4.22-6 to work yet.

Linux is 2.4.23-rc2.


Thanks,
Jeff
[ jchua@fedex.com ]

On Thu, 20 Nov 2003, Nuno Silva wrote:

>
>
> Nuno Silva wrote:
> > Hi!!
> >
> > Nix wrote:
> >
> >> Using unmodified 2.4.22-6. sshed in through the firewall on the UML
> >> instance, with tty logging enabled: typing exit on the last sshed-in
> >> console froze the box yielded this on the console before UML died:
> >>
> >> Kernel panic: switch_mm_skas - PTRACE_SWITCH_MM failed, errno = 9
> >>
> >> Kernel panic: switch_mm_skas - PTRACE_SWITCH_MM failed, errno = 9
> >>
>
> I've found a log with the error, it's a bit different from yours:
>
> Kernel panic: save_registers - saving registers failed, errno = 3
>
> Kernel panic: switch_mm_skas - PTRACE_SWITCH_MM failed, errno = 3
>
> In idle task - not syncing
>
> --------------------------------
>
>  From the screen(1) logs this was a few minutes after the boot and this
> UML should be mostly idle. I only have this log and no backtraces... not
>   very helpfull :(
>
> Regards,
> Nuno Silva
>
>
>
> -------------------------------------------------------
> This SF.net email is sponsored by: SF.net Giveback Program.
> Does SourceForge.net help you be more productive?  Does it
> help you create better code?  SHARE THE LOVE, and help us help
> YOU!  Click Here: http://sourceforge.net/donate/
> _______________________________________________
> User-mode-linux-devel mailing list
> User-mode-linux-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel
>




-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?  SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/
_______________________________________________
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] skas3-related crash of the week

[Nix]

On Thu, 20 Nov 2003, Nuno Silva stated:
> I've found a log with the error, it's a bit different from yours:
> 
> Kernel panic: save_registers - saving registers failed, errno = 3

That's ESRCH, `No such process'. Looks like the user process
died :(

> Kernel panic: switch_mm_skas - PTRACE_SWITCH_MM failed, errno = 3

Manipulating the memory mappings of a process that no longer exists is
kind of hard...

-- 
`Me, I want exploding spaceships and pulverized worlds and clashes of
 billion-year-old empires *and* competently written sentences.'
                                                    --- Matt Austern


-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?  SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/
_______________________________________________
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] skas3-related crash of the week

[Jeff Dike]

nix@esperi.org.uk said:
>  That's ESRCH, `No such process'. Looks like the user process died :( 

That also happens when the child isn't stopped.  There used to be races in
UML where it wouldn't make sure that a child had stopped (by returning from
wait with such an indication) and would start ptracing it.  This resulted
in sporadic -ESRCHs before I figured that out.  There may be some more of
these floating around.

				Jeff



-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?  SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel

[uml-devel] Host Oops with 2.4.22-6um on 2.6.0-test9-bk24

[Christopher S. Aker]

Hello,

For what it's worth...

Testing with 2.6.0-test9-bk24 with the skas3 2.6 patch that was posted a while
ago, I'm able to get the following oops on the host with a 2.4.22-6um kernel.
Interestingly, it only happens on a few of the virtual machines on the host, but
it happens every time with the ones that do it.

Reading Oops report from the terminal
Oops: 0002 [#2]
Oops: 0002 [#2]
CPU:    2
EIP:    0060:[<c01112ff>]    Not tainted
EFLAGS: 00010202
EIP is at write_ldt+0x1a4/0x240
eax: 00000000   ebx: 000f0000   ecx: 40cff202   edx: 00000001
CPU:    2
EIP:    0060:[<c01112ff>]    Not tainted
Using defaults from ksymoops -t elf32-i386 -a i386
EFLAGS: 00010202
eax: 00000000   ebx: 000f0000   ecx: 40cff202   edx: 00000001
esi: 70a0ffff   edi: f57e1928   ebp: f57e1800   esp: f580df2c
esi: 70a0ffff   edi: f57e1928   ebp: f57e1800   esp: f580df2c
ds: 007b   es: 007b   ss: 0068
Process 2.4.22-linode12 (pid: 5279, threadinfo=f580c000 task=f584c080)
Stack: f654bba4 00000001 00000001 eda4607e 00000000 00000000 400270a0 000fffff
ds: 007b   es: 007b   ss: 0068
Stack: f654bba4 00000001 00000001 eda4607e 00000000 00000000 400270a0 000fffff
       00000051 a5b801e0 f65626b0 f580dfa0 f65626b0 c01113f3 f65626b0 a5b801e0
       00000051 a5b801e0 f65626b0 f580dfa0 f65626b0 c01113f3 f65626b0 a5b801e0
       00000010 00000001 fffffffb 00000000 c011033f f65626b0 00000001 a5b801e0
Call Trace:
       00000010 00000001 fffffffb 00000000 c011033f f65626b0 00000001 a5b801e0
Call Trace:
 [<c01113f3>] modify_ldt+0x58/0x88
 [<c011033f>] sys_ptrace+0x7d7/0x840
 [<c012e6d7>] sys_rt_sigprocmask+0xb4/0x180
 [<c01113f3>] modify_ldt+0x58/0x88
 [<c011033f>] sys_ptrace+0x7d7/0x840
 [<c012e6d7>] sys_rt_sigprocmask+0xb4/0x180
 [<c010ad9f>] syscall_call+0x7/0xb

 [<c010ad9f>] syscall_call+0x7/0xb
Code: 89 30 89 48 04 31 f6 89 f9 f0 ff 85 28 01 00 00 0f 8e 70 01
Code: 89 30 89 48 04 31 f6 89 f9 f0 ff 85 28 01 00 00 0f 8e 70 01


>>EIP; c01112ff <save_i387_fxsave+3f/140>   <=====

Trace; c01113f3 <save_i387_fxsave+133/140>
Trace; c011033f <sys_ipc+25f/260>
Trace; c012e6d7 <vmtruncate_list+27/70>
Trace; c010ad9f <handle_IRQ_event+f/b0>

Code;  c01112ff <save_i387_fxsave+3f/140>
00000000 <_EIP>:
Code;  c01112ff <save_i387_fxsave+3f/140>   <=====
   0:   89 30                     mov    %esi,(%eax)   <=====
Code;  c0111301 <save_i387_fxsave+41/140>
   2:   89 48 04                  mov    %ecx,0x4(%eax)
Code;  c0111304 <save_i387_fxsave+44/140>
   5:   31 f6                     xor    %esi,%esi
Code;  c0111306 <save_i387_fxsave+46/140>
   7:   89 f9                     mov    %edi,%ecx
Code;  c0111308 <save_i387_fxsave+48/140>
   9:   f0 ff 85 28 01 00 00      lock incl 0x128(%ebp)
Code;  c011130f <save_i387_fxsave+4f/140>
  10:   0f 8e 70 01 00 00         jle    186 <_EIP+0x186>

-Chris




-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?  SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel