port-cobalt/46890: netbsd-6 / 6.0_RC1 very unstable / unusable on (my) cobalt qube 2

[mk%kilbi.de@localhost]

>Number:         46890
>Category:       port-cobalt
>Synopsis:       upcoming NetBSD 6.0 release is very unstable / unusable on 
>cobalt qube 2
>Confidential:   no
>Severity:       critical
>Priority:       high
>Responsible:    port-cobalt-maintainer
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sun Sep 02 16:40:01 +0000 2012
>Originator:     Markus W Kilbinger
>Release:        NetBSD 6.0_RC1 (GENERIC)
>Organization:
and those who don't.
>Environment:
        
        
System: NetBSD q2 6.0_RC1 NetBSD 6.0_RC1 (GENERIC) cobalt
Architecture: mipsel
Machine: cobalt
>Description:
        My cobalt qube 2 happily runs netbsd-5 (kernel and userland)
        many month now.

        Every now and then I've tried to run an actual netbsd-6 kernel
        with / on its netbsd-5 userland what failed with a complete
        unstable system where at the end all binaries dumped core
        (some incompatibilities I thought and waited).

        Now, with 6.0_RC1 I've setup up a fresh / clean cobalt netbsd
        system based on kern-GENERIC.tgz, base.tgz and etc.tgz from
        
http://nyftp.netbsd.org/pub/NetBSD-daily/netbsd-6/201209010830Z/cobalt/binary/sets/
        and booted my replacement qube 2 via NFS:

          [...]
          13.Ethernet Test...........................PASS
          16.RTC Test................................PASS
          
          Cobalt: bfd /boot.gz nfsroot=/usr/u/export/q2/root
          Decompressing done
          Executing bootloader kernel...
          Decompressing -- done.
          
          >> NetBSD/cobalt 6.0_RC1 Bootloader, Revision 1.0 [@0x80f00000]
          >> Model:               Cobalt Qube 2
          >> Memory:              65536 k
          >> PROM boot string:    root=/dev/hda1 ro 
nfsroot=/usr/u/export/q2/root 
          Boot [nfs:netbsd]: 
          Loading: nfs:netbsd
          4086352+130680 [225824+224538]=0x473d94
          Starting at 0x80001000
          
          pmap_steal_memory: seg 0: 0x474 0x474 0x3ffe 0x3ffe
          pmap_steal_memory: seg 0: 0x497 0x497 0x3ffe 0x3ffe
          pmap_steal_memory: seg 0: 0x499 0x499 0x3ffe 0x3ffe
          Copyright (c) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 
2005,
              2006, 2007, 2008, 2009, 2010, 2011, 2012
              The NetBSD Foundation, Inc.  All rights reserved.
          Copyright (c) 1982, 1986, 1989, 1991, 1993
              The Regents of the University of California.  All rights reserved.
          
          NetBSD 6.0_RC1 (GENERIC)
          Cobalt Qube 2
          total memory = 65536 KB
          avail memory = 59596 KB
          mainbus0 (root)
          com0 at mainbus0 addr 0x1c800000 level 3: st16650a, working fifo
          [...]
          wd0 at atabus0 drive 0
          wd0: <IBM-DTLA-307045>
          wd0: 43979 MB, 89355 cyl, 16 head, 63 sec, 512 bytes/sect x 90069840 
sectors
          boot device: tlp0
          root on tlp0
          nfs_boot: trying DHCP/BOOTP
          nfs_boot: DHCP next-server: 172.27.64.183
          nfs_boot: my_name=q2
          nfs_boot: my_domain=kilbi.de
          nfs_boot: my_addr=172.27.64.137
          nfs_boot: my_mask=255.255.0.0
          nfs_boot: gateway=172.27.64.107
          root on 172.27.64.183:/usr/u/export/q2/root
          root file system type: nfs
          pid 1(init): ABI set to O32 (e_flags=0x1007)
          /etc/rc.conf is not configured.  Multiuser boot aborted.
          Enter pathname of shell or RETURN for /bin/sh: 

        ... what looked quite normal.

        Then I've mounted its local harddisk:

          mount -o log /dev/wd0f /mnt

        ... and started a simple test run by unpacking a tar-archive on it:

          # cd /mnt/tmp/
          # pax -zrpe -f perl-5.14.2.tar.gz

        ... which lead to some unexpected messages / errors after some
        time:

          pax: Unable to store mode and times for directory: perl-5.14.2/t/re
          pax: Cannot malloc 29 bytes (Cannot allocate memory)
          pax: Cannot malloc 32 bytes (Cannot allocate memory)
          pax: Cannot malloc 30 bytes (Cannot allocate memory)
          pax: Cannot malloc 32 bytes (Cannot allocate memory)
          pax: Unable to store mode and times for directory: perl-5.14.2/t/uni
          pax: Cannot malloc 31 bytes (Cannot allocate memory)
          pax: Cannot malloc 32 bytes (Cannot allocate memory)
          pax: Cannot malloc 31 bytes (Cannot allocate memory)
          pax: Unable to store mode and times for directory: perl-5.14.2/t/op
          pax: Cannot malloc 32 bytes (Cannot allocate memory)
          pax: Cannot malloc 32 bytes (Cannot allocate memory)
          pax: Cannot malloc 32 bytes (Cannot allocate memory)
          pax: Cannot malloc 29 bytes (Cannot allocate memory)
          pax: Cannot malloc 31 bytes (Cannot allocate memory)
          [...]
          pax: Cannot set permissions on  (No such file or directory)
          pax: Access/modification time set failed on:  (No such file or 
directory)
          pax: Cannot set permissions on  (No such file or directory)
          pax: Access/modification time set failed on:  (No such file or 
directory)
          Memory fault (core dumped) 

        At that point (at least I have a prompt) nearly all binaries
        dump core, e.g.:

          # ls -la
          Memory fault (core dumped)

        All running services (telnet, ftp, sshd et al.) first respond
        to some stage, later become unresponsive. Finally the whole
        machine hangs...

        I can enter ddb at this stage (, but do not know what to
        investigate; maybe, someone tells me ...).

        How do other mips machines behave (cobalt specific?)?
>How-To-Repeat:
        Try to run (at least a) netbsd-6 (kernel) on a cobalt qube
        (2), produce some (i/o) activity (e. g. unpacking a larger
        tar-archive) and see how above mentioned errors appear and the
        whole system becomes more and more unstable / unresponsive.
>Fix:
        Unknown.

>Unformatted: