10% extra seconds in a minute!

[pza]

Hi,

I've just finished setting up my new media box (Core 2 Duo 6300, 2GB, 3x320GB SATA II). The box is certainly well overspeced for what I use it for. When recording digital TV using my WinFast DTV2000 H (cx88 driver) card to the raid 0 drive, I'm finding that the clock in the system is slowing down and losing about 5 seconds per minute. I'm also getting lots of messages (dmesg) saying:

cx88_wakeup: nn buffers handled (should be 1)
(where nnn is a number between 2 and 30)

I've found references to this elsewhere and people have solved it by running irqbalance. I've run irqbalance on my machine and it hasn't made any difference. Anyone got any suggestions?

If anyone cares, this is what my /proc/interrupts says:

CPU0 CPU1
0: 753736657 0 IO-APIC-edge timer
1: 2 0 IO-APIC-edge i8042
6: 6 0 IO-APIC-edge floppy
7: 0 0 IO-APIC-edge parport0
8: 1 0 IO-APIC-edge rtc
9: 0 0 IO-APIC-fasteoi acpi
12: 4 0 IO-APIC-edge i8042
16: 111 10266087 IO-APIC-fasteoi uhci_hcd:usb1, eth0, nvidia
17: 0 0 IO-APIC-fasteoi libata
18: 41539688 12258432 IO-APIC-fasteoi ide0, uhci_hcd:usb5, ehci_hcd:usb6
19: 10142 4410911 IO-APIC-fasteoi uhci_hcd:usb4, libata, EMU10K1
20: 1 4940130 IO-APIC-fasteoi cx88[0], cx88[0], cx88[0]
21: 0 0 IO-APIC-fasteoi uhci_hcd:usb2
23: 1 3 IO-APIC-fasteoi uhci_hcd:usb3, ehci_hcd:usb7
NMI: 0 0
LOC: 753735345 753735334
ERR: 0


Phil

[StefanJ]

I've been seeing the same message, but it in the context of something more egregious.

My MythTV box occasionally locks up; /var/log/messages shows a bunch of messages like these:

Oct 18 13:34:53 localhost kernel: NVRM: Xid (0000:12): 26, Ch 0000001e M 0000030c D 00000000 intr 00400000
Oct 18 13:34:58 localhost kernel: cx88_wakeup: 32 buffers handled (should be 1)
Oct 18 13:35:00 localhost kernel: cx88_wakeup: 32 buffers handled (should be 1)
Oct 18 13:35:06 localhost kernel: NVRM: Xid (0000:12): 26, Ch 0000001f M 0000030c D 00000000 intr 00400000
Oct 18 13:35:06 localhost kernel: cx88_wakeup: 32 buffers handled (should be 1)
Oct 18 13:35:10 localhost kernel: NVRM: Xid (0000:12): 26, Ch 0000001f M 0000030c D 00000000 intr 00400000
Oct 18 13:35:10 localhost kernel: cx88_wakeup: 32 buffers handled (should be 1)
Oct 18 13:35:10 localhost kernel: NVRM: Xid (0000:12): 26, Ch 0000001f M 0000030c D 00000000 intr 00400000
Oct 18 13:35:13 localhost kernel: cx88_wakeup: 32 buffers handled (should be 1)
[BIG SNIP]
Oct 18 17:38:16 localhost kernel: cx88_wakeup: 5 buffers handled (should be 1)
Oct 18 17:38:16 localhost kernel: cx88_wakeup: 3 buffers handled (should be 1)

There were many hundreds of iterations of the cx88 messages between 13:35 and 17:38. After that there were no messages until the initialization messages that were written when I reset the box.

Any ideas? It is a dual proc machine. I looked up references to irqbalance but am not sure what is going on here.

[pza]

From memory, this fixed it's self up after a kernel update. I'm running the latest Fedora 9 updates without any problems