On Thu, Aug 20, 2009 at 9:16 PM, Jason
Harmening<jason.harmening@gmail.com> wrote:> Sorry for the partial dump--I don't (yet) have a swap partition, so I
> can't do postmortem:
>
> Machine is nehalem (4 cores + HT):
>
> FreeBSD riviera.austin.rr.com 8.0-BETA2 FreeBSD 8.0-BETA2 #1: Thu Aug
> 20 09:06:25 CDT 2009
> jason@riviera.austin.rr.com:/usr/obj/usr/src/sys/CUSTOM ?amd64
>
> current process = 11 (idle: cpu1)
> [Thread pid 11 tid 100009 ]
> Stopped at acpi_cpu_c1+0x6: popq %rbp
> Tracing pid 11 tid 100009 td 0xffffff000169dab0
> acpi_cpu_c1() at acpi_cpu_c1+0x6
> acpi_cpu_idle() at acpi_cpu_idle+0x13b
> cpu_idle_acpi() at cpu_idle_acpi+0x1f
> cpu_idle() at cpu_idle+0x1c
> sched_idletd() at sched_idletd+0x258
> fork_exit() at fork_exit+0x147
> fork_trampoline() at fork_trampoline+0xe
> --- trap 0, rip = 0, rsp = 0xffffff8000076d30, rbp = 0
>
So after blindly posting to the list, I took some time to figure out
what FreeBSD uses trap code 30 for--turns out some misbehaving PCIe
hardware was firing an MSI to an unassigned vector. Switching it to
legacy interrupts fixed the problem. Sorry for the spam.