John Klos | 4 Apr 10:05 2012

Reproducible panic in NetBSD-6

Hi,

On an Amiga 1200 running NetBSD-6 from 26-March-2012 and kernel from 
4-April-2012, I'm seeing this when trying to compile m4 from pkgsrc:

uvm_fault(0x2fc478, 0x0, 0x1) -> 0xe
   type 8, code [mmu,,ssw]: 1050200
trap type 8, code = 1050200, v = 0
pid = 13549, lid = 1, pc = 0019B794, ps = 2600, sfc = 1, dfc = 1
Registers:
              0        1        2        3        4        5        6        7
dreg: 00000000 00000001 00000000 01792000 00000000 009B53BC 001347E6 009B5098
areg: 00000000 00AC0000 0CA84040 00C828A0 00304600 001153AA 0CA87DF8 1DFFE0CC

Kernel stack (0CA87C6C):
A87C6C: 002124D0 0CA87DA0 00000080 002CC33C 00000008 01050200 00000000 00000008
A87C8C: 0CA87D10 00212B8C 00000008 01050200 00000000 0CA87DA0 00000000 00000000
A87CAC: 00000008 009B53BC 001347E6 009B5098 00C828A0 0CA87DA0 00AC0000 0BC24000
A87CCC: 00000000 00000000 00000001 11C30348 000000FC 00000001 00000000 00000000
A87CEC: 00000000 00000001 00000000 00000000 00000008 00000000 00000000 00000000
A87D0C: 00000000 0CA87D88 00212ED2 00000008 01050200 00000000 0CA87DA0 00C828A0
A87D2C: 00000000 00000000 00000000 01792000 00000000 0CA84040 00C828A0 00304600
A87D4C: 001153AA 00000000 00000000 00000001 00000000 00000000 00000000 00000000
A87D6C: 00000000 00000000 00000008 00000000 00000000 00000000 00000000 0CA87DF8
A87D8C: 00002060 0CA87DA0 00000008 01050200 00000000 00000000 00000001 00000000
A87DAC: 01792000 00000000 009B53BC 001347E6 009B5098 00000000 00AC0000 0CA84040
A87DCC: 00C828A0 00304600 001153AA 0CA87DF8 1DFFE0CC 00000000 26000019 B7944008
A87DEC: 00000000 01050200 0CA84040 0CA87E40 00002830 00C828A0 00126A98 00C82320
A87E0C: 00C828A0 00000000 00000000 0179BE00 009B50A4 00C82320 00115240 00C828A0
A87E2C: 009B6FB0 00000000 00000480 A56A167B 6A87D0EE 0CA87E5C 00123F9A 00C82320
(Continue reading)

Frank Wille | 5 Apr 16:09 2012
Picon

Re: Reproducible panic in NetBSD-6

On Wed, 4 Apr 2012 08:05:16 +0000 (UTC)
John Klos <john <at> ziaspace.com> wrote:

> On an Amiga 1200 running NetBSD-6 from 26-March-2012 and kernel from 
> 4-April-2012, I'm seeing this when trying to compile m4 from pkgsrc:

Current pkgsrc or 2011Q4?

> uvm_fault(0x2fc478, 0x0, 0x1) -> 0xe
>    type 8, code [mmu,,ssw]: 1050200
> trap type 8, code = 1050200, v = 0

An MMU fault due to an invalid page descriptor.

> [...] (conftest) at    netbsd:cpu_Debugger+0x6:        unlk    a6

Is it possible to isolate this problem from conftest (this is a script?) ?

> Ideas?

Not really.

I will update my A3k to the recent NetBSD6-version over Easter and have
a look.

--

-- 
Frank Wille

John Klos | 6 Apr 04:25 2012

Re: Reproducible panic in NetBSD-6

>> On an Amiga 1200 running NetBSD-6 from 26-March-2012 and kernel from
>> 4-April-2012, I'm seeing this when trying to compile m4 from pkgsrc:
>
> Current pkgsrc or 2011Q4?

Current (like cvs, if there's no tag, it's current).

>> uvm_fault(0x2fc478, 0x0, 0x1) -> 0xe
>>    type 8, code [mmu,,ssw]: 1050200
>> trap type 8, code = 1050200, v = 0
>
> An MMU fault due to an invalid page descriptor.
>
>
>> [...] (conftest) at    netbsd:cpu_Debugger+0x6:        unlk    a6
>
> Is it possible to isolate this problem from conftest (this is a script?) ?

I just reran it with only 128 megs of memory (it's an A1200 with a 
Blizzard 1260) and paid attention to where it stopped:

checking for GNU libc compatible malloc... yes
checking for mbstate_t... yes
checking for a traditional japanese locale... ja_JP.eucJP
checking for a transitional chinese locale... zh_CN.GB18030
checking for a french Unicode locale... fr_FR.UTF-8
checking for mmap... yes
checking for MAP_ANONYMOUS... yes
checking whether memchr works... yes
checking for special C compiler options needed for large files... no
(Continue reading)

Frank Wille | 6 Apr 17:11 2012
Picon

Re: Reproducible panic in NetBSD-6

Hi John Klos wrote:

>> Current pkgsrc or 2011Q4?
>
> Current (like cvs, if there's no tag, it's current).
>
>>> uvm_fault(0x2fc478, 0x0, 0x1) -> 0xe
>>>    type 8, code [mmu,,ssw]: 1050200
>>> trap type 8, code = 1050200, v = 0
>>[..m4-conftest..]
> checking for promoted mode_t type... mode_t
> checking whether posix_spawn works...

Indeed, I can reproduce your panic, using the current pkgsrc-archive from
ftp.netbsd.org dated 31-mar-2012.

The crash location is in pmap_activate() from m68k/m68k/pmap_motorola.c:

        pmap_t pmap = l->l_proc->p_vmspace->vm_map.pmap;:

vm_map seems to be NULL.

As this function is common to all m68k ports, can you reproduce the crash on
other m68k machines (I have just Amiga hardware here)? Maybe it has
something to do with the last kmem changes.

Also the patch suggested by Michael Hitch in
  http://mail-index.netbsd.org/port-m68k/2012/02/07/msg000319.html
had not yet been committed. Maybe it is related?

(Continue reading)

John Klos | 7 Apr 08:37 2012

Re: Reproducible panic in NetBSD-6

> Indeed, I can reproduce your panic, using the current pkgsrc-archive from
> ftp.netbsd.org dated 31-mar-2012.
>
> The crash location is in pmap_activate() from m68k/m68k/pmap_motorola.c:
>
>        pmap_t pmap = l->l_proc->p_vmspace->vm_map.pmap;:
>
> vm_map seems to be NULL.
>
> As this function is common to all m68k ports, can you reproduce the crash on
> other m68k machines (I have just Amiga hardware here)? Maybe it has
> something to do with the last kmem changes.

The only non-Amiga m68k I have is a Mac running NetBSD 6, and it compiled 
m4 just fine. Hmmm...

John

Michael van Elst | 7 Apr 10:50 2012
Picon

Re: Reproducible panic in NetBSD-6

john <at> ziaspace.com (John Klos) writes:

>> Indeed, I can reproduce your panic, using the current pkgsrc-archive from
>> ftp.netbsd.org dated 31-mar-2012.
>>
>> The crash location is in pmap_activate() from m68k/m68k/pmap_motorola.c:
>>
>>        pmap_t pmap = l->l_proc->p_vmspace->vm_map.pmap;:
>>
>> vm_map seems to be NULL.
>>
>> As this function is common to all m68k ports, can you reproduce the crash on
>> other m68k machines (I have just Amiga hardware here)? Maybe it has
>> something to do with the last kmem changes.

>The only non-Amiga m68k I have is a Mac running NetBSD 6, and it compiled 
>m4 just fine. Hmmm...

That crash seems to be easily reproducable by the posix_spawn test.

-- 
--

-- 
                                Michael van Elst
Internet: mlelstv <at> serpens.de
                                "A potential Snark may lurk in every tree."

Frank Wille | 7 Apr 12:08 2012
Picon

Re: Reproducible panic in NetBSD-6

Michael van Elst wrote:

>>The only non-Amiga m68k I have is a Mac running NetBSD 6, and it
>>compiled m4 just fine. Hmmm...
>
>
> That crash seems to be easily reproducable by the posix_spawn test.

True, but the problem cannot be isolated. I tried to extract the posix_spawn
test source from the configure script and compiled it on my Amiga. But it
worked!

Source is attached.

--

-- 
Frank Wille

#include <errno.h>
#include <fcntl.h>
#include <signal.h>
#include <spawn.h>
#include <stdbool.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include <sys/types.h>
#include <sys/wait.h>
(Continue reading)

John Klos | 8 Apr 18:36 2012

Re: Reproducible panic in NetBSD-6

>> As this function is common to all m68k ports, can you reproduce the 
>> crash on other m68k machines (I have just Amiga hardware here)? Maybe 
>> it has something to do with the last kmem changes.
>
> The only non-Amiga m68k I have is a Mac running NetBSD 6, and it compiled m4 
> just fine. Hmmm...

Not sure if this is related to the kmem changes, but on that m68k Mac 
running NetBSD-6, I just had a reboot last night while compiling PHP 5.3 
from pkgsrc. Maybe I can connect a serial console and see if it happens 
again.

One of the things I thought was strange is that the total kernel memory 
used on this machine hovers around 39 megabytes, which seems very high for 
a machine which has 68 megs total. But maybe this hasn't changed with 
NetBSD-6, since a NetBSD 5 mac68k machine with 132 megs shows the kernel 
using 65 megabytes. Hmmm...

John

Radoslaw Kujawa | 8 Apr 21:59 2012
Picon

Re: Reproducible panic in NetBSD-6

Hi guys,

On 8 kwi 2012, at 18:36, John Klos wrote:

>>> As this function is common to all m68k ports, can you reproduce the crash on other m68k machines (I have
just Amiga hardware here)? Maybe it has something to do with the last kmem changes.

I wonder if kern/46286 is related to these problems?

--

-- 
Best regards,
Radoslaw Kujawa

Al Zick | 27 Apr 19:04 2012

multilink PPP

Hi,

I would like to setup a system to run 4 to 8 modems using multilink  
PPP. I have looked at the man pages for this on one of my systems,  
but it looks like multilink PPP is not supported by NetBSD. Can  
someone verify this?

Also, I need to find a serial card that has at least 4 network ports  
that is supported by NetBSD. Does any one know where I could get one?

Thanks,
Al

        mp     Enables the use of PPP multilink; this is an alias for  
the `mul-
               tilink' option.  This option is currently only   
available  under
               Linux.

        multilink
               Enables the use of the PPP multilink protocol.  If the  
peer also
               supports  multilink,  then this link can become part  
of a bundle
               between the local system and the peer.  If there is  
an  existing
               bundle  to  the  peer,  pppd will join this link to  
that bundle,
               otherwise pppd will create a new bundle.  See the  
MULTILINK sec-
(Continue reading)


Gmane