i created a virtual machine with 2 processors in virtual box, and installed centos 6.0 as the os.

after setting up the lamp environment, i wrote a simple php script:
test.php
<?php
for ($loop=0; $loop<10000; $loop++)
echo "<img src='image/default/nopic.gif?$loop' width=5 />";
?>

i visited test.php, and then typed /proc/interrupts in the terminal.
CPU0 CPU1
0: 145 0 IO-APIC-edge timer
1: 1601 0 IO-APIC-edge i8042
8: 0 0 IO-APIC-edge rtc0
9: 0 0 IO-APIC-fasteoi acpi
12: 1420 0 IO-APIC-edge i8042
14: 3595 0 IO-APIC-edge ata_piix
15: 89 0 IO-APIC-edge ata_piix
19: 34125 0 IO-APIC-fasteoi eth0
21: 6 0 IO-APIC-fasteoi Intel 82801AA-ICH
NMI: 0 0 Non-maskable interrupts
LOC: 58185 44542 Local timer interrupts
SPU: 0 0 Spurious interrupts
PMI: 0 0 Performance monitoring interrupts
PND: 0 0 Performance pending work
RES: 6603 10727 Rescheduling interrupts
CAL: 42 123 Function call interrupts
TLB: 195 316 TLB shootdowns
TRM: 0 0 Thermal event interrupts
THR: 0 0 Threshold APIC interrupts
MCE: 0 0 Machine check exceptions
MCP: 7 7 Machine check polls
ERR: 0
MIS: 0

/proc/irq/19/smp_affinity:
3

cpu1 didn't handle irq 19.

anyone can tell me what is the wrong?

thanks!

Hi

Nothing is wrong in your case. That's how Linux kernel works when handling interrupts.

More precisely, Linux kernel tends to make the last serving core to handle the same interrupt again and again. You might ask "weird, doesn't that mean load will be unbalanced?". True. But in this case, Linux kernel tries to maintain cache hotness. As you can guess, interrupt handling involves code (the handler) and data (the packets etc). By making sure interrupt is handled by same core, there is no need to cache the same data on another core. The (wanted) net result is eventually faster execution.

Thanks, bro!

You welcome GreenForest