> Interesting; lots of ACPI edge-triggered interrupts:
>
> dagda:~# cat /proc/interrupts
> CPU0
> 0: 519365 IO-APIC-edge timer
> 1: 16713 IO-APIC-edge keyboard
> 2: 0 XT-PIC cascade
> 8: 4 IO-APIC-edge rtc
> 9: 0 IO-APIC-level acpi
> 14: 863415 IO-APIC-edge ide0
> 15: 201651 IO-APIC-edge ide1
> 19: 306188 IO-APIC-level nvidia
> 20: 57261 IO-APIC-level usb-ohci, eth0
> 21: 0 IO-APIC-level ehci_hcd, NVidia nForce2
> 22: 3 IO-APIC-level usb-ohci, ohci1394
> NMI: 0
> LOC: 519312
> ERR: 0
> MIS: 0
Your interrupts look fine, this is the way they should be.
> ... but no stability problems since the primary drive has been
> running at UDMA133. Earlier UDMA100 freezes were completely
> repeatable; identical kernel, just without your two patches.
You can try downgrading your drive to udma5 to see if udma6 really does make
it more stable (hdparm -X udma5 /dev/hdX) but I can't think of any reason
why it should.
> I take it that I should boot with noapic in future to be safe.
I've been telling people to disable APIC / ACPI because of the interrupt
problem, but your interrupts are fine, so I'd leave it alone. I'm curious,
what version BIOS do you have?
-Allen
r/[email protected]/2003.09.19/16:49:45
>You can try downgrading your drive to udma5 to see if udma6 really does make
>it more stable (hdparm -X udma5 /dev/hdX) but I can't think of any reason
>why it should.
That's it; downgrading to UDMA100 crashes within a few minutes
of heavy I/O. Running at UDMA133 is rock solid.
>> I take it that I should boot with noapic in future to be safe.
>
>I've been telling people to disable APIC / ACPI because of the interrupt
>problem, but your interrupts are fine, so I'd leave it alone. I'm curious,
>what version BIOS do you have?
Very good, thanks. Shuttle SN45G, FN45 mobo, AwardBIOS v6.00PG.
Is that the info you're looking for?
Thanks, Merlin
On Saturday 20 September 2003 00:49, Allen Martin wrote:
> > Interesting; lots of ACPI edge-triggered interrupts:
> >
> > dagda:~# cat /proc/interrupts
> > CPU0
> > 0: 519365 IO-APIC-edge timer
> > 1: 16713 IO-APIC-edge keyboard
> > 2: 0 XT-PIC cascade
> > 8: 4 IO-APIC-edge rtc
> > 9: 0 IO-APIC-level acpi
> > 14: 863415 IO-APIC-edge ide0
> > 15: 201651 IO-APIC-edge ide1
> > 19: 306188 IO-APIC-level nvidia
> > 20: 57261 IO-APIC-level usb-ohci, eth0
> > 21: 0 IO-APIC-level ehci_hcd, NVidia nForce2
> > 22: 3 IO-APIC-level usb-ohci, ohci1394
> > NMI: 0
> > LOC: 519312
> > ERR: 0
> > MIS: 0
>
> Your interrupts look fine, this is the way they should be.
>
> > ... but no stability problems since the primary drive has been
> > running at UDMA133. Earlier UDMA100 freezes were completely
> > repeatable; identical kernel, just without your two patches.
>
> You can try downgrading your drive to udma5 to see if udma6 really does
> make it more stable (hdparm -X udma5 /dev/hdX) but I can't think of any
> reason why it should.
>
> > I take it that I should boot with noapic in future to be safe.
>
> I've been telling people to disable APIC / ACPI because of the interrupt
> problem, but your interrupts are fine, so I'd leave it alone. I'm curious,
> what version BIOS do you have?
ACPI/APIC should work on most nforce/nforce2 boards in 2.4.22 upwards. If not,
let me know, and I'll look into it.