On Mon, 2006-07-17 at 16:27 -0700, Rick Stevens wrote:
[snip]
> >
> > The ethtool trick did not work. I went back to kernel 2080 and enabled
> > the ati-fglrx module again. All is well now. Maybe this working setup is
> > useful to others although it does not explain why this was happening.
>
> Hmmm. An interrupt conflict, possibly? A lot of video cards use IRQ9,
> and so do some network cards. Can cause issues...
It seems so except IRQ 19 and IRQ 233 are in play here. Come to think of
it, this is the first time I've seen such a high number being used for
an IRQ. With kernel 2080 and the ati-fglrx module enabled there is no
longer a radeon entry sharing an IRQ with eth0. Actually, looking at the
output below it looks much saner than the one I initially posted when
running kernel 2157. Weird stuff.
[patrick@laptop ~]$ cat /proc/interrupts
CPU0
0: 1674389 local-APIC-edge timer
1: 5621 IO-APIC-edge i8042
8: 2838140 IO-APIC-edge rtc
12: 2437 IO-APIC-edge i8042
14: 121841 IO-APIC-edge ide0
15: 59487 IO-APIC-edge ide1
16: 1 IO-APIC-level yenta
17: 129627 IO-APIC-level ATI IXP, ATI IXP Modem
19: 12221547 IO-APIC-level eth0
20: 31059 IO-APIC-level ohci_hcd:usb1, ohci_hcd:usb2,
ehci_hcd:usb3
21: 5081 IO-APIC-level acpi
NMI: 2740
LOC: 1674596
ERR: 124
MIS: 0
Regards,
Patrick
