Re: [Openipmi-developer] Soft lockup (and reboot): IPMI

[Corey Minyard <minyard@acm.org>]

Philipp Matthias Hahn wrote:
> Matt Domsch wrote:
>
>   
>> On Mon, Aug 14, 2006 at 06:14:46PM +0200, Philipp Matthias Hahn wrote:
>>
>>     
>>> While playing with "openipmigui", the server just rebooted with the
>>> following last message:
>>>
>>> BUG: soft lockup detected on CPU#0!
>>> <c0103416> show_trace+0xd/0xf
>>> <c01034e5> dump_stack+0x15/0x17
>>> <c01382b0> softlockup_tick+0x9d/0xae
>>> <c012190a> run_local_timers+0x12/0x14
>>> <c0121748> update_process_times+0x3c/0x61
>>> <c010e1d9> smp_apic_timer_interrupt+0x57/0x5f
>>> <c010307c> apic_timer_interrupt+0x1c/0x24
>>> <f8aa8783> ipmi_thread+0x43/0x71 [ipmi_si]
>>> <c0129e62> kthread+0x78/0xa0
>>> <c0100e31> kernel_thread_helper+0x5/0xb
>>>
>>> Any idea on what went wrong?
>>>       
>> What kernel please?  If 2.6.17 or higher, this is new.  If < 2.6.17,
>> this is most likely due to the udelay(1) call in ipmi_thread() which
>> in 2.6.17 was replaced with a schedule().
>>     
>
> Sorry for omitting that information, it's a plain 2.6.17.8 kernel.
> model name      : Intel(R) Xeon(TM) CPU 2.66GHz
>
> ipmi message handler version 39.0
> IPMI Watchdog: driver initialized
> ipmi device interface
> IPMI System Interface driver.
> ipmi_si: Trying SMBIOS-specified KCS state machine at I/O address 0xca2,
> slave address 0x24, irq 0
> ipmi: Found new BMC (man_id: 0x002880,  prod_id: 0x0000, dev_id: 0x00)
>  IPMI KCS interface initialized
>   
The only way I can see this would happen is if the BMC is misbehaving
somehow, like leaving the "ready" bit set for a long time (not clearing
it when it is supposed to) and always signalling that there is some
event ready.  I guess if the BMC was so fast that it was alway ready
that could cause this problem, but that doesn't really seem very likely
as userland would have to run to feed it to do that.

This is apparently a Fujitsu device, right?  The fact that the product
id is 0x0000 leads me to believe that this device is not completely
cooked yet.

-Corey
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/