[mythtv-users] Occasional hard freezes

[Ryan Steffes]

On 10/28/05, Michael T. Dean <mtdean at thirdcontact.com> wrote:
>
> Ryan Steffes wrote:
>
> >On 10/28/05, Julian Edwards <myth at julian-edwards.com> wrote:
> >
> >
> >>Ryan Steffes wrote:
> >>
> >>>I thought I had finally gotten this fixed with the kernel option
> >>>nolapic, but it happened to me again. I am experiencing hard lock ups
> >>>that seem to be related to high data throughput, but aren't consistent
> >>>enough for me to track down. When it freezes, only hitting the reset
> >>>or power button fixes it.
> >>>
> >>>I have an MSI K7N2 Delta 2 motherboard with an AMD Barton 2500+ in
> >>>it. It has a nforce northbridge. I am passing the kernel "acpi=no
> >>>noapic nolapic". I'm running 2.6.11-6mdk. I have one PVR 150. I have
> >>>two hard drives, running LVM to make one video share. I'm running
> >>>ivtv 0.3.9
> >>>
> >>>What happens is a hard lock up. It generally happens when I'm
> >>>exercising the ethernet hard.
> >>>
> >>>
> >>This exact same thing was happening to me too with my KT6 Delta.
> >>
> >>[snip]
> >>
> >>>Can anyone give any advice on where to go from here?
> >>>
> >>>
> >>Sure. The problem is almost certainly to do with interrupt sharing. In
> >>my case it was a problem with the ACPI steering assigning too many
> >>devices the same IRQ. I ended up fixing it by disabling unnecessary
> >>onboard hardware (eg I don't need 8 USB controllers or onboard sound),
> >>using acpi=off (is that different to your acpi=no ?) and re-positioning
> >>
> >According to proc/interrupts, there shouldn't be a sharing problem.
> However,
> >I'm unsure if there is a difference between acpi=no and acpi=off. I'll
> try
> >"off" and see if I can pound away on the NIC and the hard drive and see
> what
> >happens.
> >
> vi /usr/src/linux-`uname -r`/Documentation/kernel-parameters.txt
>
> /acpi
>
> acpi= [HW,ACPI] Advanced Configuration and Power Interface
> Format: { force | off | ht | strict }
> force -- enable ACPI if default was off
> off -- disable ACPI if default was on
> noirq -- do not use ACPI for IRQ routing
> ht -- run only enough ACPI to enable Hyper Threading
> strict -- Be less tolerant of platforms that
> are not
> strictly ACPI specification compliant.
>
> I think you were trying for "acpi=noirq noapic nolapic". And, it
> wouldn't hurt to throw in a "pci=noacpi" for good measure. Most of the
> 'net gets these parameters wrong, though, so I realize you saw someone
> else recommend this...
>
> Note, also, that "acpi=off" would give you the same IRQ-related results
> as "acpi=noirq" (and disable the rest of ACPI, also).



Changing to acpi=off did not fix the problem. I'm a little concerned with
whether it's really off though, it seems to turn on local apic before
reading the kernel options:

Oct 29 08:56:54 mythtv kernel: Found and enabled local APIC!
Oct 29 08:56:54 mythtv kernel: Initializing CPU#0
Oct 29 08:56:54 mythtv kernel: Kernel command line: auto BOOT_IMAGE=MythTV
ro root=341 acpi=off noapic nolapic resume=/dev/hdb3 splash=silent
Oct 29 08:56:54 mythtv kernel: bootsplash: silent mode.

Later, it does say:

Oct 29 08:56:54 mythtv kernel: PCI: PCI BIOS revision 2.10 entry at 0xfbe50,
last bus=2
Oct 29 08:56:54 mythtv kernel: PCI: Using configuration type 1
Oct 29 08:56:54 mythtv kernel: mtrr: v2.0 (20020519)
Oct 29 08:56:54 mythtv kernel: ACPI: Subsystem revision 20050211
Oct 29 08:56:54 mythtv kernel: ACPI: Interpreter disabled.
Oct 29 08:56:54 mythtv kernel: Linux Plug and Play Support v0.97 (c) Adam
Belay
Oct 29 08:56:54 mythtv kernel: pnp: PnP ACPI: disabled
Oct 29 08:56:54 mythtv kernel: PnPBIOS: Disabled
Oct 29 08:56:54 mythtv kernel: PCI: Probing PCI hardware
Oct 29 08:56:54 mythtv kernel: PCI: Probing PCI hardware (bus 00)
Oct 29 08:56:54 mythtv kernel: PCI: Using IRQ router default [10de/01e0] at
0000:00:00.0

but it doesn't say anything about local apic again. How can I check that
it's really disabled?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mythtv.org/pipermail/mythtv-users/attachments/20051029/062874ea/attachment.htm