9.2-RC4 amd64 panic: vm_page_unwire

[John Marshall]

On Fri, 04 Oct 2013, 23:29 +1000, John Marshall wrote:
> On Mon, 30 Sep 2013, 19:55 +1000, John Marshall wrote:
> > > >> > On Fri, 27 Sep 2013, 11:12 +0300, Konstantin Belousov wrote:
> > > >> >> On Fri, Sep 27, 2013 at 10:07:28AM +1000, John Marshall wrote:
> > > >> >> > I have made the core.txt.[0-2] files available in the following
> > > >> >> > directory.  The directory is not browsable.
> > > >> >> >
> > > >> >> >   http://www.riverwillow.net.au/~john/92rc4/
> > > >> >>
> > > >> >> This might be fixed by r254087-r254090 on stable/9.
> 
> As well as applying those patches to releng/9.2, and still seeing the
> panic, I have upgraded one of the systems to stable/9 and still see the
> panic.
> 
> > Another data point: both systems on which I have seen this panic have
> > the ipmi driver compiled in.  ipmi makes the BMC's watchdog timer
> > available to the system.
> 
> I ruled this out.  I recompiled a kernel with no ipmi driver (so that
> removes the watchdog timer too) but the system would still panic on
> shutdown when stopping ntpd.
> 
> I have also reverted to the STABLE ntpd distribution (4.6.4p5
> 2011-12-24) and still see the panic.
> 
> I took one of the systems back to releng/9.1 (9.1-RELEASE-p7).  Its
> kernel includes the ipmi driver, I am running watchdogd, I am running
> ntpd, I have rebooted several times and have never seen a panic.
> 
> So, I now have one system on 9.1-RELEASE (panic-free); and the second
> system on 9-STABLE (panics).  Previously I had both systems on
> 9.2-RELEASE and panicking.
> 
> Another common factor with all but one of the panics I have seen is that
> the process which appears to trigger the panic is a 'realtime' process.
> 
> rwsrv08# top -b 200 | awk '(/^  PID/ || $5~/^r/)'
>   PID USERNAME    THR PRI NICE   SIZE    RES STATE   C   TIME   WCPU COMMAND
>  1739 root          1 -52   r0 12056K  8040K nanslp  7   0:02  0.00% watchdogd
>  1911 root          1 -52  r0F 32668K 18468K select  7   0:02  0.00% ntpd

I've just squashed this theory too.  I've just tried running ntpd at
normal priority (without the -N switch) and it hasn't made any
difference.  The system will still panic when shutting down ntpd.

> Here is a recent stable/9 tripping over watchdogd on shutdown:
> 
> FreeBSD 9.2-STABLE #0 r256029:
>   .
>   Stopping watchdogd.
>   Waiting for PIDS: 1729
>   panic: vm_page_unwire: page 0xfffffe042bdae2d8's wire count is zero
>   cpuid = 1
>   KDB: stack backtrace:
>   #0 0xffffffff804e8188 at kdb_backtrace+0x68
>   #1 0xffffffff804adcba at panic+0x21a
>   #2 0xffffffff80718dd2 at vm_page_unwire+0x102
>   #3 0xffffffff80705a52 at vm_fault_unwire+0xd2
>   #4 0xffffffff8070db41 at vm_map_delete+0x171
>   #5 0xffffffff8070ddbf at vm_map_remove+0x5f
>   #6 0xffffffff80711199 at vmspace_exit+0xc9
>   #7 0xffffffff804773dd at exit1+0x72d
>   #8 0xffffffff804783de at sys_sys_exit+0xe
>   #9 0xffffffff8073e86f at amd64_syscall+0x3bf
>   #10 0xffffffff80729307 at Xfast_syscall+0xf7

I have just reported this panic in a PR --> kern/182661

-- 
John Marshall
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 196 bytes
Desc: not available
URL: <http://lists.freebsd.org/pipermail/freebsd-stable/attachments/20131005/107b59d3/attachment.sig>