problems with 7.2, vm_page_insert: page already inserted

[Raphael Becker]

On Sat, Apr 11, 2009 at 08:30:42PM -0400, Ken Smith wrote:
> That said if there is a bug it's likely to be something fairly difficult
> to reproduce (lots of people are running 7.2-PRERELEASE and so far only
> two reports of this issue) so it's not a shock for you to have only had
> it happen once so far.  Odds are it takes a rare sequence of events to
> trigger it.

I also reported this in Mar 19th with some kgdb-output:
http://lists.freebsd.org/pipermail/freebsd-stable/2009-March/048911.html

== uname -a ==
FreeBSD top.uugrn.org 7.1-STABLE FreeBSD 7.1-STABLE #0: Sat Mar 14
20:06:04 CET 2009
root at top.uugrn.org:/usr/obj/usr/src_RELENG_7/sys/TOP  i386


== /usr/src/sys/i386/conf/TOP ==
include GENERIC
ident           TOP
nodevice                plip            # TCP/IP over parallel

== kldstat ==
Id Refs Address    Size     Name
 1    9 0xc0400000 9ec994   kernel
 2    1 0xc0ded000 164e8    geom_mirror.ko
 3    1 0xc0e04000 6a45c    acpi.ko
 4    1 0xc5b9a000 e000     ipfw.ko
 5    1 0xc5c94000 4000     logo_saver.ko
 6    1 0xc5cf4000 4000     nullfs.ko
 7    1 0xc5cfc000 4000     fdescfs.ko



== Hardware == 

The machine ran about 30 months without crashes, at last with 6.4-RELEASE. 
The storage was on a 3ware with two attached SATA150 drives. 

Then we replaced the old 3ware + 2xSATA by two onboard-attached SATA300
drives with geom mirror configuration. The new OS is 7.1-STABLE from Mar
14th, which crashed with "vm_page_insert: page already inserted" under
i/o load.

geom mirror status:
          Name    Status  Components
mirror/TOPRAID  COMPLETE  ad4
                          ad6

atapci1 at pci0:0:31:2:    class=0x010601 card=0x778015d9 chip=0x27c18086 rev=0x01 hdr=0x00
    vendor     = 'Intel Corporation'
    device     = '82801GB I/O Controller Hub SATA cc=AHCI'
    class      = mass storage
    subclass   = SATA


Mar 14 22:45:37 top kernel: atapci1: <Intel AHCI controller> port 0x6898-0x689f,0x687c-0x687f,0x6890-0x6897,0x6878-0x687b,0x6880-0x688f mem 0xed000400-0xed0007ff irq 19 at device 31.2 on pci0
Mar 14 22:45:37 top kernel: atapci1: [ITHREAD]
Mar 14 22:45:37 top kernel: atapci1: AHCI Version 01.10 controller with 4 ports detected
Mar 14 22:45:37 top kernel: ata2: <ATA channel 0> on atapci1
Mar 14 22:45:37 top kernel: ata2: [ITHREAD]
Mar 14 22:45:37 top kernel: ata3: <ATA channel 1> on atapci1
Mar 14 22:45:37 top kernel: ata3: [ITHREAD]

 
Mar 14 22:45:37 top kernel: ad4: 381554MB <Seagate ST3400620NS 3.AEG> at ata2-master SATA300
Mar 14 22:45:37 top kernel: ad6: 381554MB <Seagate ST3400620NS 3.AEG> at ata3-master SATA300


== What happened ==
The crash happened while cvsupping from local cvsup-mirror, hence there
was pretty much read-write i/o on the geom while having some CPU load.

After rebooting the geom-mirror was degraded and rebuilded within some 
couple of hours. I didn't have had any crashes since then but don't 
want to provoke this since this server is kind of "productive".

Maybe those "environmental" information does help someone to spot the
cause of this error.

Regards
Raphael Becker

-- 
Raphael Becker          <rabe at uugrn.org>          http://rabe.uugrn.org/
GnuPG:                E7B2 1D66 3AF2 EDC7 9828  6D7A 9CDA 3E7B 10CA 9F2D
.........|.........|.........|.........|.........|.........|.........|..
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 187 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-stable/attachments/20090412/b679116e/attachment.pgp