8-STABLE won't boot with ZFSv28

Jeremy Chadwick freebsd at jdc.parodius.com
Wed Jun 1 08:54:57 UTC 2011

On Wed, Jun 01, 2011 at 08:23:19AM +0000, Holger Kipp wrote:
> I have a very irritating problem with 8-STABLE and ZFSv28
> I upgraded to 8-STABLE as of yesterday (31.05.2011),
> downloaded stable-8-zfsv28-20110521.patch.xz
> and applied the patch using
> cd /usr/src
> patch -E -p0 < /path/to/patchfile
> make buildworld
> make buildkernel KERNCONF=foo
> make installkernel KERNCONF=foo
> make installworld
> mergemaster
> which all went smoothly.
> After reboot, I only got
> unknown: WARNING - ATAPI_IDENTITFY requeued due to channel reset LBA=0
> all the time, and then after an hour or so (wasn't on site),
> system gave
> Fatal trap 12: page fault while in kernel mode
> cupid - 0; apic id = 00
> fault virtual address = 0x8
> fault code                       = supervisor read data, page not present
> instruction pointer    = 0x20:0xffffffff80252301
> stack poiner                   = 0x28:0xffffff80000a7ac0
> frame pointer                = 0x28:0xffffff80000a7b00
> code segment                = base 0x0, limit 0xfffff, type 0x1b
>                                                = DPL 0, pres1, long 1, def32 0, gran 1
> processor eflags         = interrupt enabled, resume, IOPL = 0
> current process           = 0 (thread taskq)trap number               = 12
> panic: page fault
> cpuid = 0
> Uptime: 1h0m13s
> Cannot dump. Device not defined or unavailable.
> Automatic reboot in 15 seconds - press a key on the console to abort
> Needless to say the system did not reboot. Had to powercycle.
> Then always got the
> unknown: WARNING - ATAPI_IDENTITFY requeued due to channel reset LBA=0
> error about once per second.
> Have now used a fixit-disk to change back to the old kernel:
> FreeBSD 8.2-STABLE #12: Mon Apr 18 12:48:56 CEST 2011
> and rebootet.
> Now zfs claims to be v28, current storage pool is at 15.I'd love to
> try ZFSv28, but with the old kernel I don't think
> this is a good idea - but with the new kernel it seems I can't
> even boot properly.
> Any suggestions as to how to proceed?

I think this is much more likely related to an ATA/ATAPI-related change
that was committed on April 17th recently and is not related to ZFSv28.
Please see this thread:

* 2011/05/29 -- ICH9 panic/instability on recent kernel

Holger, can you please provide the following two things?

1) Output from "pciconf -lvcb".

2) Full output from a verbose boot (option "5" at the loader prompt).

I imagine #2 isn't going to work for most users because there's no way
to get pages and pages and pages of data from a panic'd machine without
either serial console (which will require a 2nd machine and possibly a
null-modem cable) or properly setting up a dedicated swap partition and
large-enough /var filesystem, plus their kernel would need DDB support
added to it (so they could properly do "call doadump" then "reboot").

A workaround which one user has confirmed is to enable AHCI for your
SATA controller in your system BIOS (if such is available).  ataahci.ko
will be used (which is AHCI via ATA) and your device names probably
won't change.  Alternatively you could enable AHCI and use ahci.ko
(ahci_load="yes" in /boot/loader.conf) to get AHCI via CAM, which
provides NCQ and other features, but your device names will change.
My familiarity with ATAPI is limited however.

CC'ing mav@ here.

Alexander, Holger's report looks exactly like Michael's report.

Possibly we should consider reverting the April 17th commit until we can
figure out what's going on here.

| Jeremy Chadwick                                   jdc at parodius.com |
| Parodius Networking                       http://www.parodius.com/ |
| UNIX Systems Administrator                  Mountain View, CA, USA |
| Making life hard for others since 1977.               PGP 4BD6C0CB |

More information about the freebsd-stable mailing list