Subject: port-i386/19854: Immediate reboot on HP Omnibook 6000 with ACPI enabled
To: None <gnats-bugs@gnats.netbsd.org>
From: None <sverre@viewmark.com>
List: netbsd-bugs
Date: 01/14/2003 21:05:03
>Number:         19854
>Category:       port-i386
>Synopsis:       Immediate reboot on HP Omnibook 6000 with ACPI enabled
>Confidential:   yes
>Severity:       serious
>Priority:       medium
>Responsible:    port-i386-maintainer
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Tue Jan 14 20:06:01 PST 2003
>Closed-Date:
>Last-Modified:
>Originator:     Sverre Froyen
>Release:        NetBSD 1.6L current 13 Jan 2002
>Organization:
Viewmark
>Environment:
	
	
System: NetBSD abbor.fesk.com 1.6L NetBSD 1.6L (ABBOR) #26: Sun Jan 12 12:05:30 MST 2003 toor@abbor.fesk.com:/usr/src/sys/arch/i386/compile/ABBOR i386
(I can provide the full config file, if required).
Architecture: i386
Machine: i386
>Description:
Added ACPI device to kernel config.  Compiled.  The resulting kernel
causes the computer to reboot during the boot process.  Adding printf
and delay statements to acpi_attach in /usr/src/sys/dev/acpi/acpi.c
I was able to extract the following (commented) boot messages:

acpi0 at mainbus0
acpi0: X/RSDT: OemId <HP-MCD,EA RSDT ,01820000>, AslId < LTP,00000000>

#### The next two lines are from the call to AcpiInstallAddressSpaceHandler.  
Despite the errors, it appears that AcpiInstallAddressSpaceHandler returns 
AE_OK because the acpi_attach code continues.

    ACPI-1286: *** Error: Method execution failed [\_SB_.PCI0.FDS_.FCB1.CSID] 
(Node 0xc090e320), AE_NOT_EXIST
    ACPI-1286: *** Error: Method execution failed [\_SB_.PCI0.FDS_._REG] (Node 
0xc090e6a0), AE_NOT_EXIST

acpi0: SCI interrupting at int 9
#### The next line is from acpi_enable_fixed_events
acpi0: fixed-feature power button present

#### The call to acpi_pci_fixup causes a reboot.

#### If I comment out the call to acpi_pci_fixup, the boot continues until the 
call to acpi_build_tree.  This call causes a reboot as well.  Commenting it 
out allows the computer to complete the boot process.

FYI Similar problem was observed about a year ago for FreeBSD on an
Omnibook 6100.  I have not seen reports that that problem has been fixed.

>How-To-Repeat:
Add the device:

        acpi0  at mainbus0

and the options

        options         ACPI_MADT_DEBUG # ACPI debug flag
        options         ACPI_PROBE_DEBUG        # ACPI debug flag
        options         MPVERBOSE       # ACPI debug flag
        options         ACPIVERBOSE     # verbose ACPI device autoconfig messages
to a working config file.  Compile kernel.  Boot on HP Omnibook 6000.

>Fix:
Do not enable ACPI on HP Omnibook 6000.
>Release-Note:
>Audit-Trail:
>Unformatted: