nforce2 one card per irq only

From: Bob
Date: Sat Dec 20 2003 - 06:46:54 EST


Why does dmesg and dmesg -s65536 and
CONFIG_LOG_BUF_SHIFT 14 or 16 have
my dmesg and kern.log truncate here?
I can't see Ross' delay patch printk

In case of odd devfs vc tty thing I
tried "console=vc/0 console=tty0"

dmesg begins--

been registered
PnPBIOS: Scanning system for PnP BIOS support...
PnPBIOS: Found PnP BIOS installation structure at 0xc00fc660
PnPBIOS: PnP BIOS version 1.0, entry 0xf0000:0xc690, dseg 0xf0000
pnp: match found with the PnP device '00:07' and the driver 'system'
pnp: match found with the PnP device '00:08' and the driver 'system'
PnPBIOS: 13 nodes reported by PnP BIOS; 13 recorded by driver
SCSI subsystem initialized
ACPI: PCI Interrupt Link [APCS] enabled at IRQ 23
IOAPIC[0]: Set PCI routing entry (2-23 -> 0xa9 -> IRQ 23 Mode:1 Active:0)

and if I use smp instead of uniprocessor kernel,
I get a few lines more showing so Ross' numbers
show but PIC timer is on so the numbers are
useless

________________________________________________

MSI K7N2 Delta MCP2-T amd xp3000+ 333mhz fsb 1:1

nforce2, Award bios update stopped crashing

kernel 2.6.0-test11

I need Ross' ioapic patch to get ioapic edge timer
working.

cpu disconnect defaults to on, and on or off makes
no diff except 50% more MIS count, ERR 0 either way,
no crashes.

Once I moved a tulip ethernet 10/100 nic off the
irq of a promise ide controller I got ERR 0 and
much less MIS count.

24 hours uptime

CPU0 0: 84527308 IO-APIC-edge timer
1: 37635 IO-APIC-edge i8042
2: 0 XT-PIC cascade
8: 1 IO-APIC-edge rtc
9: 0 IO-APIC-level acpi
12: 532175 IO-APIC-edge i8042
14: 106 IO-APIC-edge ide0
15: 215 IO-APIC-edge ide1
16: 354495 IO-APIC-level ide2, ide3
17: 0 IO-APIC-level yenta, yenta
19: 151532 IO-APIC-level ide4, ide5
20: 309690 IO-APIC-level eth0
21: 1579 IO-APIC-level NVidia nForce2
NMI: 0 LOC: 84527227 ERR: 0
MIS: 188

according to athcool, cpu disconnect on

So one irq per slot is still necessary with pci on
this system, and hdparm unmask off did not help.

This is a uniprocessor but the hd controller and
tulip card on one irq would produce a lot of APIC
errors when cp over nfs while bonnie++, and here
is something from Maciej nominally for smp--

/usr/src/linux-2.6.0-test11/arch/i386/kernel/apic.c
/*
* Some unknown Intel IO/APIC (or APIC) errata is biting us with
* certain networking cards. If high frequency interrupts are
* happening on a particular IOAPIC pin, plus the IOAPIC routing
* entry is masked/unmasked at a high rate as well then sooner or
* later IOAPIC line gets 'stuck', no more interrupts are received
* from the device. If focus CPU is disabled then the hang goes
* away, oh well :-(
*
* [ This bug can be reproduced easily with a level-triggered
* PCI Ne2000 networking cards and PII/PIII processors, dual
* BX chipset. ]
*/
/*
* Actually disabling the focus CPU check just makes the hang less
* frequent as it makes the interrupt distributon model be more
* like LRU than MRU (the short-term load is more even across CPUs).
* See also the comment in end_level_ioapic_irq(). --macro
*/


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/