[ltt-dev] Problem in irq handling

Kiran Kulkarni kiran.1267 at gmail.com
Fri Jun 5 11:18:46 EDT 2009


Hello,

I am using LTTng for my device which has arm processor. I have stuck at 
a problem that after each timer interrupt
my cpu spends 10ms time in serving irq (Graph representation also shows 
same). So it actually spends time 10 ms for each timer interrupt,
which I guess not correct. (Note: My timer interrupt update system timer 
by 10ms).

can you help me figure out what causing irq handlers to take 10ms....


Snippet example:

kernel_sched_schedule: 98.730291747 (/home/kiran/Desktop/trace4/cpu_0), 
0, 0, swapper, , 0, 0x0, SYSCALL { prev_pid = 444, next_pid = 0, 
prev_state = 2 }
kernel_irq_entry: 98.730292968 (/home/kiran/Desktop/trace4/cpu_0), 0, 0, 
swapper, , 0, 0x0, IRQ { irq_id = 8, kernel_mode = 1 }
kernel_timer_update_time: 98.740001220 
(/home/kiran/Desktop/trace4/cpu_0), 0, 0, swapper, , 0, 0x0, IRQ { 
jiffies = 4294947170, xtime_sec = 69321, xtime_nsec = 732571667, wallto
monotonic_sec = -69224, walltomonotonic_nsec = 687428333 }
kernel_softirq_raise: 98.740002440 (/home/kiran/Desktop/trace4/cpu_0), 
0, 0, swapper, , 0, 0x0, IRQ { softirq_id = 1 
[run_timer_softirq+0x0/0x208] }
kernel_softirq_raise: 98.740003661 (/home/kiran/Desktop/trace4/cpu_0), 
0, 0, swapper, , 0, 0x0, IRQ { softirq_id = 8 
[rcu_process_callbacks+0x0/0x38] }
kernel_irq_exit: 98.740006102 (/home/kiran/Desktop/trace4/cpu_0), 0, 0, 
swapper, , 0, 0x0, SYSCALL





More information about the lttng-dev mailing list