On Sun, Jun 04, 2006 at 09:26:47PM +0200, Michael Buesch wrote:
> So, here we go. Preemptible Periodic Work for bcm43xx-softmac.
> Please stress-test with all methods which come to your minds.
I tested with the same configuration that I could always crash before
(smp/preempt/debug kernel, then run iperf load tests for an hour or so).
I couldn't get it to crash this time. I saw no problems at all.
This was the non-shared interrupt case - /proc/interrupts looked like
this when I gave up (i lost the beginning):
8: 0 IO-APIC-edge rtc
9: 9185 IO-APIC-level acpi
12: 5305 IO-APIC-edge i8042
14: 62265 IO-APIC-edge ide0
15: 40518 IO-APIC-edge ide1
16: 4 IO-APIC-level yenta, ohci1394
17: 1 IO-APIC-level yenta, rtk
18: 435 IO-APIC-level ohci_hcd:usb1, NVidia nForce3 Modem
19: 36 IO-APIC-level ohci_hcd:usb2, NVidia nForce3
20: 0 IO-APIC-level ehci_hcd:usb3
21: 11233681 IO-APIC-level bcm43xx
NMI: 1419
LOC: 1157643
ERR: 0
MIS: 6251
21
9149
So I decided to test the shared interrupt case. When I boot this machine
with 'noapic', I get a /proc/interrupts that looks like this:
CPU0
0: 209671 XT-PIC timer
1: 1648 XT-PIC i8042
2: 0 XT-PIC cascade
7: 42 XT-PIC parport0
8: 0 XT-PIC rtc
9: 472 XT-PIC acpi
10: 575 XT-PIC yenta, ohci_hcd:usb2, ehci_hcd:usb3, NVidia
nForce3 Modem, rtk
11: 904239 XT-PIC yenta, ohci1394, ohci_hcd:usb1, NVidia
nForce3, bcm43xx
12: 120 XT-PIC i8042
14: 54296 XT-PIC ide0
15: 6822 XT-PIC ide1
NMI: 113
LOC: 209646
ERR: 0
MIS: 0
I found out which of the usb ports was on irq 11, plugged in a usb
mouse, and checked that it generated interrupts. (I had to do
"cat /dev/input/mice > /dev/null"). Then I repeated the network load
tests on the shared bcm43xx irq. Still no problems.
looks good! send it upstream. :)
Jason
_______________________________________________
Bcm43xx-dev mailing list
Bcm43xx-dev@lists.berlios.de
http://lists.berlios.de/mailman/listinfo/bcm43xx-dev
