Bug 140979

Summary:	[acpi] [panic] Kernel panic (fatal trap 12: page fault when in kernel mode) on FreeBSD 8.0 with ACPI because of "ec" sub-device
Product:	Base System	Reporter:	Tarick <tungan>
Component:	kern	Assignee:	Andriy Gapon <avg>
Status:	Closed FIXED
Severity:	Affects Only Me
Priority:	Normal
Version:	Unspecified
Hardware:	Any
OS:	Any

Description Tarick 2009-11-28 18:30:01 UTC

Booting 8.0-stable and from the 8.0 livefs CD produces kernel panic with trace:
___________________________________________________________________
--skipped--
acpi_lid0: <Control Method Lid Switch> on acpi0
acpi_acad0: <AC Adapter> on acpi0

Fatal trap 12: page fault while in kernel mode
cpuid = 0; apic id = 00
fault viftual address = 0x40
fault code            = supervisor read data, page not present
instruction pointer   = 0x20:0xffffffff001ccfa0
stack pointer         = 0x20:0xffffffff00c6f070
frame pointer         = 0x20:0xffffffff00c6f0a0
code segment          = base rx0, limit 0xffffff, type 0x1b
                      = DPL 0, pres 1, long 1, def32 0, gran 1
processor eflags      = interrupt enabled, resume, IOPL=0
current process       = 0 (swapper)
trap number           = 12
panic:  page fault
cpuid = 0
Uptime: 1s
Automatic reboot <skipped>
______________________________________________________________________

The system is Dell Inspiron 1501, latest BIOS 2.6.3. FreeBSD was upgraded from working previously 7.2 with ACPI.

Disabling ACPI during the boot helps, and later I was able to isolate the problem to "ec" sub-device (debug.acpi.disabled="ec")

Fix: 

Put into /boot/loader.conf:
debug.acpi.disabled="ec"
How-To-Repeat: Boot with FreeBSD 8.0 CD with ACPI enabled on Dell Inspiron 1501. Don't have another one to verify though.

Comment 1 Mark Linimon freebsd_committer

2009-11-28 20:21:14 UTC

Responsible Changed
From-To: freebsd-bugs->freebsd-acpi

looks like this is ACPI-related.

Comment 2 Andriy Gapon 2009-11-30 13:57:23 UTC

Could you please reproduce this with debug enabled?
E.g.:
http://www.freebsd.org/doc/en/books/developers-handbook/kerneldebug.html

Without a backtrace it's hard to see where the problem is.

-- 
Andriy Gapon

Comment 3 Tarick 2009-11-30 19:45:55 UTC

Unfortunately I can't save core dump despite configured debugging  
settings, system just doesn't see it in swap, probably because the crash  
occurs before swap is configured.  
What other options do I have?  

--- Original Message ---  
From: Andriy Gapon <avg@icyb.net.ua>  
To: bug-followup@FreeBSD.org, tungan@ukr.net  
Date: 30 november, 15:57:23  
Subject: Re: kern/140979: [acpi] [panic] Kernel panic (fatal trap 12: page fault when in kernel mode) on FreeBSD 8.0 with ACPI because of "ec" sub-device  

  Could you please reproduce this with debug enabled?  
E.g.:  
http://www.freebsd.org/doc/en/books/developers-handbook/kerneldebug.html  

Without a backtrace it's hard to see where the problem is.  

--  
Andriy Gapon

Comment 4 Andriy Gapon 2009-12-01 13:48:38 UTC

on 30/11/2009 21:38 Tarick said the following:
> Unfortunately I can't save core dump despite configured debugging
> settings, system just doesn't see it in swap, probably because the crash
> occurs before swap is configured.
> What other options do I have?

Could you please boot with ec disabled (via the hint) and do the following?

In shell:
$ kgdb /boot/kernel/kernel /dev/mem
In kgdb:
(kgdb) info line *0xffffffff001ccfa0

And send back the output of the last command.
Thanks.

-- 
Andriy Gapon

Comment 5 Tarick 2009-12-01 19:39:09 UTC

Sure, as I understand this means booting with debug.acpi.disabled="ec"
line in /boot/loader.conf. I use this setting constantly right now.
But I recompiled kernel recently, and the instruction pointer address
changed to 0xffffffff801bfe20. Here is result with this address, I hope
this will help:
_______________________________________________________________
~# kgdb /boot/kernel/kernel /dev/mem
GNU gdb 6.1.1 [FreeBSD]
Copyright 2004 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you
are
welcome to change it and/or distribute copies of it under certain
conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB.  Type "show warranty" for
details.
This GDB was configured as "amd64-marcel-freebsd"...
Reading symbols from /boot/kernel/sem.ko...Reading symbols
from /boot/kernel/sem.ko.symbols...done.
done.
Loaded symbols for /boot/kernel/sem.ko
Reading symbols from /usr/local/lib/oss/modules/osscore.ko...done.
Loaded symbols for /usr/local/lib/oss/modules/osscore.ko
Reading symbols from /usr/local/lib/oss/modules/oss_hdaudio.ko...done.
Loaded symbols for /usr/local/lib/oss/modules/oss_hdaudio.ko
Reading symbols from /usr/local/modules/rtc.ko...done.
Loaded symbols for /usr/local/modules/rtc.ko
Reading symbols from /boot/kernel/ng_pppoe.ko...Reading symbols
from /boot/kernel/ng_pppoe.ko.symbols...done.
done.
Loaded symbols for /boot/kernel/ng_pppoe.ko
#0  sched_switch (td=0xffffffff80962140, newtd=Variable "newtd" is not
available.
) at /usr/src/sys/kern/sched_ule.c:1864
1864			cpuid = PCPU_GET(cpuid);

(kgdb) info line *0xffffffff801bfe20
Line 538 of "/usr/src/sys/contrib/dev/acpica/executer/exmutex.c"
   starts at address 0xffffffff801bfe20 <AcpiExReleaseMutex+544> and
ends at 0xffffffff801bfe25 <AcpiExReleaseMutex+549>.
(kgdb) 
__________________________________________________________

-----Original Message-----
From: Andriy Gapon <avg@icyb.net.ua>
To: Tarick <tungan@ukr.net>
Cc: bug-followup@FreeBSD.org
Subject: Re: kern/140979: [acpi] [panic] Kernel panic (fatal trap 12:
page fault when in kernel mode) on FreeBSD 8.0 with ACPI because of "ec"
sub-device
Date: Tue, 01 Dec 2009 15:48:38 +0200

on 30/11/2009 21:38 Tarick said the following:
> Unfortunately I can't save core dump despite configured debugging
> settings, system just doesn't see it in swap, probably because the crash
> occurs before swap is configured.
> What other options do I have?

Could you please boot with ec disabled (via the hint) and do the following?

In shell:
$ kgdb /boot/kernel/kernel /dev/mem
In kgdb:
(kgdb) info line *0xffffffff001ccfa0

And send back the output of the last command.
Thanks.

Comment 6 Andriy Gapon 2009-12-02 06:39:02 UTC

on 01/12/2009 21:39 Tarick said the following:
> Sure, as I understand this means booting with debug.acpi.disabled="ec"
> line in /boot/loader.conf. I use this setting constantly right now.
> But I recompiled kernel recently, and the instruction pointer address
> changed to 0xffffffff801bfe20. Here is result with this address, I hope
> this will help:
[snip]
> (kgdb) info line *0xffffffff801bfe20
> Line 538 of "/usr/src/sys/contrib/dev/acpica/executer/exmutex.c"
>    starts at address 0xffffffff801bfe20 <AcpiExReleaseMutex+544> and
> ends at 0xffffffff801bfe25 <AcpiExReleaseMutex+549>.

So this points us to AcpiExReleaseMutex.
One possible control flow chain is AcpiExReleaseMutex <- AcpiReleaseGlobalLock
<- EcUnlock.  But it's still not clear what could be wrong.

Would it be possible to recompile your kernel with the following options,
reproduce the panic and report full panic message?

makeoptions	DEBUG="-O -g"
options 	DDB
options 	DDB_NUMSYM
options 	KDB
options 	KDB_TRACE

This should enable printing of stack trace on panic.
Thanks!

-- 
Andriy Gapon

Comment 7 Tarick 2009-12-02 17:57:53 UTC

Done, common panic message, but last line:

[thread pid 0 tid 100000 ]
Stopped at 0xffffffff801c95a8 = AcpiExReleaseMutex+0x218: movzbi 0x40(%
rax), %r14d
db>

If more data are needed from debugger, please tell me so, but I may end
up sending you movies or jpg.

-----Original Message-----
From: Andriy Gapon <avg@icyb.net.ua>
To: Tarick <tungan@ukr.net>
Cc: bug-followup@FreeBSD.org
Subject: Re: kern/140979: [acpi] [panic] Kernel panic (fatal trap 12:
page fault when in kernel mode) on FreeBSD 8.0 with ACPI because of "ec"
sub-device
Date: Wed, 02 Dec 2009 08:39:02 +0200
<cut>
So this points us to AcpiExReleaseMutex.
One possible control flow chain is AcpiExReleaseMutex <- AcpiReleaseGlobalLock
<- EcUnlock.  But it's still not clear what could be wrong.

Would it be possible to recompile your kernel with the following options,
reproduce the panic and report full panic message?

makeoptions	DEBUG="-O -g"
options 	DDB
options 	DDB_NUMSYM
options 	KDB
options 	KDB_TRACE

This should enable printing of stack trace on panic.
Thanks!

Comment 8 Andriy Gapon 2009-12-02 18:07:17 UTC

on 02/12/2009 19:57 Tarick said the following:
> Done, common panic message, but last line:
> 
> [thread pid 0 tid 100000 ]
> Stopped at 0xffffffff801c95a8 = AcpiExReleaseMutex+0x218: movzbi 0x40(%
> rax), %r14d
> db>
> 
> If more data are needed from debugger, please tell me so, but I may end
> up sending you movies or jpg.

Let's start with 'bt' command.

-- 
Andriy Gapon

Comment 9 Andriy Gapon freebsd_committer

2010-02-02 11:55:57 UTC

Responsible Changed
From-To: freebsd-acpi->avg

I will take this as I've been working on thise issue.

Comment 10 Andriy Gapon freebsd_committer

2010-02-02 13:47:15 UTC

State Changed
From-To: open->analyzed

Root cause of this issue has been determined by upstream vendor, 
ACPICA, and a fix for it should be included in the next 
release. 
We will get it after importing the next vendor release. 
Many thanks to originator for providing debugging and testing.

Comment 11 Bruce Cran freebsd_committer

2010-03-08 19:41:16 UTC

State Changed
From-To: analyzed->patched

The 201002 9-CURRENT livecd still had the problem, but it has been fixed in the 
20100308 snapshot from http://pub.allbsd.org/FreeBSD-snapshots/ which contains 
ACPICA 20100304.

Comment 12 Andriy Gapon freebsd_committer

2010-04-27 10:50:26 UTC

ACPICA 20100331 has been merged to stable/8.
Will the original submitter be able to test if the issue is fixed now, please?
Thanks!
-- 
Andriy Gapon

Comment 13 Tarick 2010-04-27 20:10:18 UTC

Updated kernel to the latest stable and booted just fine.

-----Original Message-----
From: Andriy Gapon <avg@freebsd.org>
To: bug-followup@freebsd.org, tungan@ukr.net
Subject: Re: kern/140979: [acpi] [panic] Kernel panic (fatal trap 12:
page fault when in kernel mode) on FreeBSD 8.0 with ACPI because of "ec"
sub-device
Date: Tue, 27 Apr 2010 12:50:26 +0300

ACPICA 20100331 has been merged to stable/8.
Will the original submitter be able to test if the issue is fixed now, please?
Thanks!

Comment 14 Andriy Gapon freebsd_committer

2010-04-27 20:38:20 UTC

on 27/04/2010 22:10 Tarick said the following:
> Updated kernel to the latest stable and booted just fine.

Thank you!

-- 
Andriy Gapon

Comment 15 Andriy Gapon freebsd_committer

2010-04-27 21:06:20 UTC

State Changed
From-To: patched->closed

ACPICA version with fix has been MFCed to stable/8 now. 
ACPI in stable/7 is 'frozen'.