[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Shutdown problem on Tyan S2875 opteron SMP



On Sat, Sep 16, 2006 at 07:30:14PM -0400, Don Hayward wrote:
> 
> 
> On Sat, 16 Sep 2006, Martin Jambor wrote:
> 
> >On 9/16/06, garrone <pgarrone@acay.com.au> wrote:
> >>On Sat, Sep 16, 2006 at 11:57:44AM +0200, Martin Jambor wrote:
> >>> Hi,
> >>>
> >>> On 9/16/06, garrone <pgarrone@acay.com.au> wrote:
> >>> >> >The command "shutdown -h" does not completely poweroff the system
> >>> >> >for kernels 2.6.11, 2.6.12, and 2.6.16. However 2.6.8 does
> >>> >> >turn off as it should!
> >>> >>
> >>> >Thanks. I was momentarily excited that 2.6.8 worked but the others
> >>> >didnt, suggesting that the problem was with the linux kernel rather
> >>> >than the board. However the situation where some kernel versions
> >>> >work and others don't now suggests to me that the problem is more 
> >>likely
> >>> >to be with the board than the kernel. The linux code seems to follow
> >>> >the standard pretty closely. I tried 2.6.17 and that does
> >>> >not work either. I am running bios 3.03.
> >>>
> >>> Maybe I am stating the obvious, but have you had a look at and
> >>> compared  the configurations of the kernels that do and don't work?
> >>>
> >>> I run vanilla kernels and whenever such problems arise (on both amd64
> >>> and i386) it is most certainly my configuration error, not a kernel
> >>> version issue (there have been exceptions like burning through ATA
> >>> etc. though).
> >>>
> >>> HTH
> >>>
> >>> Martin
> >>
> >>So what configuration options do you mean? When I debugged a failing
> >>2.6.12, my debug statements were definitely being executed
> >>at the point where it was supposed to stop.
> >>I did briefly look at the configuration, but didnt see anything.
> >>These are standard debian pre-packaged images as well.
> >>The ACPI config from zless /proc/config.gz | grep ACPI is
> >>
> >>CONFIG_X86_64_ACPI_NUMA=y
> >># ACPI (Advanced Configuration and Power Interface) Support
> >>CONFIG_ACPI=y
> >>CONFIG_ACPI_SLEEP=y
> >>CONFIG_ACPI_SLEEP_PROC_FS=y
> >># CONFIG_ACPI_SLEEP_PROC_SLEEP is not set
> >>CONFIG_ACPI_AC=m
> >>CONFIG_ACPI_BATTERY=m
> >>CONFIG_ACPI_BUTTON=m
> >>CONFIG_ACPI_VIDEO=m
> >>CONFIG_ACPI_HOTKEY=m
> >>CONFIG_ACPI_FAN=m
> >>CONFIG_ACPI_PROCESSOR=m
> >>CONFIG_ACPI_HOTPLUG_CPU=y
> >>CONFIG_ACPI_THERMAL=m
> >>CONFIG_ACPI_NUMA=y
> >>CONFIG_ACPI_ASUS=m
> >>CONFIG_ACPI_IBM=m
> >>CONFIG_ACPI_TOSHIBA=m
> >>CONFIG_ACPI_BLACKLIST_YEAR=0
> >># CONFIG_ACPI_DEBUG is not set
> >>CONFIG_ACPI_EC=y
> >>CONFIG_ACPI_POWER=y
> >>CONFIG_ACPI_SYSTEM=y
> >>CONFIG_ACPI_CONTAINER=m
> >>CONFIG_X86_POWERNOW_K8_ACPI=y
> >>CONFIG_X86_SPEEDSTEP_CENTRINO_ACPI=y
> >>CONFIG_X86_ACPI_CPUFREQ=m
> >># CONFIG_X86_ACPI_CPUFREQ_PROC_INTF is not set
> >>CONFIG_HOTPLUG_PCI_ACPI=m
> >>CONFIG_HOTPLUG_PCI_ACPI_IBM=m
> >>CONFIG_PNPACPI=y
> >>CONFIG_SERIAL_8250_ACPI=y
> >>
> >>It all looks pretty set to me.
> >
> >Yeah, it really looks good. Any differences between the 2.6.8 config?
> >
> >I also recall encountering a motherboard that wouldn't shut down with
> >ACPI and needed APM, didn't it go away?
> >
> 
> I also run stock kernels.  Here is a diff from 2.6.15 to 2.6.16 for
> ACPI configs.  Do they look like the changes from 2.6.8?  I don't have
> a 2.6.8 to to look at.
> 
> >
> < # Linux kernel version: 2.6.15-1-amd64-k8-smp
> < # Mon Mar 20 08:27:27 2006
> ---
> ># Linux kernel version: 2.6.16
> ># Fri Aug 18 20:35:54 2006
> 
> >
> >CONFIG_ACPI_SLEEP=y
> >CONFIG_ACPI_SLEEP_PROC_FS=y
> ># CONFIG_ACPI_SLEEP_PROC_SLEEP is not set
> >CONFIG_ACPI_HOTPLUG_CPU=y
> 
> They seem to be included in the 2.6.12 config listed above.
> 
> Don
> 
> 
> --
> Don.Hayward at pomobuli.net

In reply to both Don and Martin.
I had completely expurgitated the 2.6.8 from my system,
but I downloaded it again, reinstalled it, rebooted in recovery mode, 
and saved the /proc/config.gz,
then confirmed that it does indeed halt everything with shutdown -h.
The ACPI options are set as follows:

# ACPI (Advanced Configuration and Power Interface) Support
CONFIG_ACPI=y
CONFIG_ACPI_BOOT=y
CONFIG_ACPI_INTERPRETER=y
# CONFIG_ACPI_SLEEP is not set
CONFIG_ACPI_AC=m
CONFIG_ACPI_BATTERY=m
CONFIG_ACPI_BUTTON=m
CONFIG_ACPI_FAN=m
CONFIG_ACPI_PROCESSOR=m
CONFIG_ACPI_THERMAL=m
CONFIG_ACPI_ASUS=m
CONFIG_ACPI_TOSHIBA=m
# CONFIG_ACPI_DEBUG is not set
CONFIG_ACPI_BUS=y
CONFIG_ACPI_EC=y
CONFIG_ACPI_POWER=y
CONFIG_ACPI_PCI=y
CONFIG_ACPI_SYSTEM=y
# CONFIG_SERIAL_8250_ACPI is not set

The CONFIG_ACPI_SLEEP option is irrelevent, as we are not
trying to send the system to sleep.

ACPI is a standard. The kernel isn't supposed to vary from board to
board. If the hardware doesnt poweroff when its told to, then its the
fault of the hardware. It's up to the manufacturer
to program their bios so their board shuts down properly.
Perhaps we should try to find a tyan s2875 that actually works
properly with multiple versions of linux. If there are none,
then we know its the fault of the board design.

Peter



Reply to: