Bob Chiodini said the following on 31/08/2005 14:18:
Anything relevant in /var/log/messages? What is the output of route -n?
Bob...
I had a look at /var/log/messages (that I attach, only as e-mail to you...), but I am not very expert. I see a timeout transmit on eth0 repeated many times when using SMP kernel at about 6.35 at the end. I see also many errors conencted to ACPI (on both kernels) that I am not able to decode
Tnx
Antonio,
I culled the messages down into a SMP boot and a UP boot. A quick look indicates one message is missing in the SMP boot log. It's marked with an arrow and indicates that the eth0 device interrupt was detected as IRQ 11 (UP only). Hopefully someone can shed some light on this. Some other information that might help:
cat /proc/interrupts for both the UP and SMP.
Also, do you have the latest BIOS installed for your motherboard? What kind of motherboard is this?
An Asus P5P800, Bios updated in July
Check the IO-APIC options in the BIOS
setup.
You might try moving the realtek NIC to another slot. That's a shot in the dark, but it's pretty early and I'm only into my first cup of coffee!
Bob...
Standard CPU0 0: 20180445 XT-PIC timer 1: 5563 XT-PIC i8042 2: 0 XT-PIC cascade 5: 701131 XT-PIC SysKonnect SK-98xx, Intel ICH5, ehci_hcd:usb1, uhci_hcd:usb3 8: 1 XT-PIC rtc 11: 380784 XT-PIC uhci_hcd:usb2, uhci_hcd:usb4, uhci_hcd:usb5, eth0 12: 801086 XT-PIC i8042 14: 112972 XT-PIC ide0 15: 358109 XT-PIC ide1 NMI: 0 ERR: 0
SMP CPU0 CPU1 0: 137004 48896 IO-APIC-edge timer 1: 29 27 IO-APIC-edge i8042 5: 0 0 IO-APIC-edge SysKonnect SK-98xx, Intel ICH5, ehci _hcd:usb1, uhci_hcd:usb3 8: 1 0 IO-APIC-edge rtc 11: 0 0 IO-APIC-edge uhci_hcd:usb2, uhci_hcd:usb4, uhci_h cd:usb5, eth0 12: 1146 836 IO-APIC-edge i8042 14: 9294 5957 IO-APIC-edge ide0 15: 440 568 IO-APIC-edge ide1 NMI: 0 0 LOC: 181971 182055 ERR: 0 MIS: 0
I am wondering if the two kernels have different management of iptables, in particular IPv4 or IPv6, another shot in the dark, even if is afternoon here
tnx
Antonio,
service iptables stop. I suspect that the IRQ for eth0 is not getting properly registered and thus the timeouts.
Does specifying acpi=ht on the SMP kernel command line make any difference? It probably defeats the purpose of SMP.
Bob...
and furthermore prevents from poweroff .... :-(