Home » Mailing list archives » linux-kernel » 2008 » September » 10

Re: OOPS: getting an ACPI interrupt while in an ACPI interrupt hangs my box

Previous thread: [PATCH] firmware: Allow release-specific firmware dir by Jeff Mahoney on Tuesday, September 9, 2008 - 7:15 am. (53 messages)

Next thread: [RFC] [PATCH -mm] cgroup: limit the amount of dirty file pages by Andrea Righi on Tuesday, September 9, 2008 - 8:38 am. (5 messages)
[view original message]
From: Holger Schurig
Subject: OOPS: getting an ACPI interrupt while in an ACPI interrupt hangs my box
Date: Tuesday, September 9, 2008 - 7:05 am

I just got this OOPS:

irq 9: nobody cared (try booting with the "irqpoll" option)
Pid: 0, comm: swapper Tainted: P          2.6.26.3 #3
 [<c012f063>] __report_bad_irq+0x24/0x69
 [<c012f06a>] __report_bad_irq+0x2b/0x69
 [<c012f257>] note_interrupt+0x1af/0x1e4
 [<c01cbc97>] acpi_irq+0xb/0x1c
 [<c012e983>] handle_IRQ_event+0x1a/0x3f
 [<c012f917>] handle_level_irq+0x63/0x84
 [<c01046d0>] do_IRQ+0x4b/0x60
 [<c010320f>] common_interrupt+0x23/0x28
 [<c01d007b>] acpi_ds_create_operands+0x1c/0xf9
 [<c0118f03>] __do_softirq+0x2c/0x75
 [<c0118f6e>] do_softirq+0x22/0x26
 [<c01191ea>] irq_exit+0x25/0x53
 [<c01046d5>] do_IRQ+0x50/0x60
 [<c010320f>] common_interrupt+0x23/0x28
 [<c01ef3c0>] acpi_idle_enter_simple+0x16d/0x1d9
 [<c025bdba>] cpuidle_idle_call+0x49/0x77
 [<c025bd71>] cpuidle_idle_call+0x0/0x77
 [<c010176c>] cpu_idle+0x48/0x61
 =======================
handlers:
[<c01cbc8c>] (acpi_irq+0x0/0x1c)
Disabling IRQ #9
BUG: soft lockup - CPU#0 stuck for 138s! [syslogd:1952]

If I read that OOPS right, then I got an interrupt 
("common_interrupt"), which made the kernel run the ACPI 
function "acpi_ds_create_operands" and while this executed, I've 
got another interupt, which killed my for for 138 seconds ...

Shouldn't there some sort of protection about this or is the DSDT 
of my device just bogus?




Kernel 2.6.26.3

CPU: "Geode(TM) Integrated Processor by AMD PCS"

How to reproduce: press ACPI buttons repeatedly

ACPI related kernel-commandline entries: "pci=noacpi 
acpi_serialize". The first is vital (no boot without), the 
second one is an experiment, I got OOPSes without that, too.
--


[ message continues ]
[view original message]
From: Zhao Yakui
Subject: Re: OOPS: getting an ACPI interrupt while in an ACPI interrupt hangs my box
Date: Tuesday, September 9, 2008 - 6:01 pm

If the system is booted with the boot option of
"processor.max_cstate=1", does the following warning message still
exist?

Of course you can open a new bug in bugzilla and attach the output of
dmesg, acpidump, lspci -vxxx.

http://bugzilla.kernel.org/enter_bug.cgi?product=ACPI


--


[ message continues ]
[view original message]
From: Holger Schurig
Subject: Re: OOPS: getting an ACPI interrupt while in an ACPI interrupt hangs my box
Date: Wednesday, September 10, 2008 - 7:47 am

Yes, if this did help, then only marginally.
--


[ message continues ]
Previous thread: [PATCH] firmware: Allow release-specific firmware dir by Jeff Mahoney on Tuesday, September 9, 2008 - 7:15 am. (53 messages)

Next thread: [RFC] [PATCH -mm] cgroup: limit the amount of dirty file pages by Andrea Righi on Tuesday, September 9, 2008 - 8:38 am. (5 messages)