Re: Panic with areca and dd

On 7/4/07, Jan Schmidle <freebsd-current_at_jschmidle.org> wrote:
> I reduced my Problem from zfs to a general disk, i/o problem.
> But now i don't know if it is areca driver related or a
> general freebsd-current problem.
>
> When I start two dd's like "dd if=/dev/zero of=/dev/da1 bs=1024"
> on different disks, the machine panics with the following backtrace and
> stuff.
> I also appended a complete dmesg before the crash. The machine is
> running a cvsuped GENERIC
> kernel from yesterday.
>
> Does anybody know, how to solve this ?

I'm getting a similar panic with an Areca 1220 handing eight disks in
RAID6, on a 4-way Xeon. It panics both SMP and UP. I tried an Areca
patch that adds locking to arcmsr_interrupt() but that didn't help.

Matt

panic: Bad tailq NEXT(0xffffff0000990c28->tqh_last) != NULL
KDB: enter: panic
[thread pid 23 tid 100021 ]
Stopped at      kdb_enter+0x2f: nop

db> trace
Tracing pid 23 tid 100021 td 0xffffff040fc65b00
kdb_enter() at kdb_enter+0x2f
panic() at panic+0x1d2
xpt_done() at xpt_done+0x14a
arcmsr_interrupt() at arcmsr_interrupt+0x312
ithread_loop() at ithread_loop+0x10c
fork_exit() at fork_exit+0xaa
fork_trampoline() at fork_trampoline+0xe
--- trap 0, rip = 0, rsp = 0xffffffffac0dad30, rbp = 0 ---

db> show alllocks
Process 23 (irq16: arcmsr0) thread 0xffffff040fc65b00 (100021)
exclusive sleep mutex arcmsr mp lock r = 0 (0xffffffff80a45548) locked _at_ /usr/sr
c/sys/dev/arcmsr/arcmsr.c:680
Process 17 (swi2: cambio) thread 0xffffff040fc8a2c0 (100012)
exclusive sleep mutex Giant r = 0 (0xffffffff805fb140) locked _at_ /usr/src/sys/cam
/cam_xpt.c:7107

db> show allpcpu
Current CPU: 0

cpuid        = 0
curthread    = 0xffffff040fc65b00: pid 23 "irq16: arcmsr0"
curpcb       = 0xffffffffac0dad40
fpcurthread  = none
idlethread   = 0xffffff040fc87580: pid 10 "idle"
spin locks held: