9.2-BETA2 panics

[Bryan Venteicher]

----- Original Message -----
> Hi,
> this host (used as a zfs server), was working since 8.2, actually was working
> nicely under 9.1, but after upgrading to the latest 9.2, it panics, 2 days
> in a row. Appart of being a newer version, it's now dataless while I run it
> through the loops - which could be the reason for the panics, so while I
> prepare
> it to run off a local disk, and see if that mitigates the problem,
> could someone take a look?
> 
> Fatal trap 12: page fault while in kernel mode
> cpuid = 21; apic id = 25
> fault virtual address   = 0xfffffffffffffcc0
> fault code              = supervisor read data, page not present
> instruction pointer     = 0x20:0xffffffff80d17f66
> stack pointer           = 0x28:0xffffff8d77b415b0
> frame pointer           = 0x28:0xffffff8d77b415f0
> code segment            = base 0x0, limit 0xfffff, type 0x1b
>                         = DPL 0, pres 1, long 1, def32 0, gran 1
> processor eflags        = interrupt enabled, resume, IOPL = 0
> current process         = 89 (txg_thread_enter)
> trap number             = 12
> panic: page fault
> cpuid = 21
> KDB: stack backtrace:
> db_trace_self_wrapper() at db_trace_self_wrapper+0x2a/frame
> 0xffffff8d77b41040
> kdb_backtrace() at kdb_backtrace+0x37/frame 0xffffff8d77b41100
> panic() at panic+0x1ce/frame 0xffffff8d77b41200
> trap_fatal() at trap_fatal+0x290/frame 0xffffff8d77b41260
> trap_pfault() at trap_pfault+0x211/frame 0xffffff8d77b412f0
> trap() at trap+0x344/frame 0xffffff8d77b414f0
> calltrap() at calltrap+0x8/frame 0xffffff8d77b414f0
> --- trap 0xc, rip = 0xffffffff80d17f66, rsp = 0xffffff8d77b415b0, rbp =
> 0xffffff8d77b415f0 ---
> bcopy() at bcopy+0x16/frame 0xffffff8d77b415f0
> kthread_add() at kthread_add+0xe4/frame 0xffffff8d77b41710
> kproc_kthread_add() at kproc_kthread_add+0xe1/frame 0xffffff8d77b418c0
> spa_sync() at spa_sync+0x8d1/frame 0xffffff8d77b41990
> txg_sync_thread() at txg_sync_thread+0x139/frame 0xffffff8d77b41aa0
> fork_exit() at fork_exit+0x11f/frame 0xffffff8d77b41af0
> fork_trampoline() at fork_trampoline+0xe/frame 0xffffff8d77b41af0
> --- trap 0, rip = 0, rsp = 0xffffff8d77b41bb0, rbp = 0 ---
> Uptime: 21h5m46s
> 

The serialization in kthread_add() is wrong. It is possible for the
oldtd it selects to exit and be reaped before we are able duplicate
the copy region. I have a local patch for this, and I talked with
julian@ & jhb@ about it a few weeks ago but haven't sent them a
patch for review. I'll get to that later today.


> more info at:
> 	ftp://ftp.cs.huji.ac.il/users/danny/freebsd/core.txt/1
> 
> thanks,
> 	danny
> 
> 
> _______________________________________________
> freebsd-stable at freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-stable
> To unsubscribe, send any mail to "freebsd-stable-unsubscribe at freebsd.org"
>