Re: [patch] Real-Time Preemption, -RT-2.6.10-rc2-mm1-V0.7.27-3

[Florian Schmidt]

On Tue, 16 Nov 2004 22:24:01 +0100
Ingo Molnar <mingo@xxxxxxx> wrote:

> great. The current release is meanwhile at -V0.7.27-10, which includes
> other minor updates:
> 

Ok, this one boots fine again for me (didn't test the ones betwen my last
report and this one).

I have not yet tried to get this kernel to lock up yet, but i made another
interesting observation:

irq 8 at prio 98 (only irq 1 with higher prio 99). running rtc_wakeup in the
console (it runs SCHED_FIFO allright). Switching consoles (different text
consoles - not swithcing to X, though this basically produces similar
results) produces large jitters (around 1 ms) and occasional missed irq's
and piggy messages. This is completely reproducable here. The rtc histogram
doesn't show any large wakeup latencies.

/proc/latency_trace doesn't show that high latencies either on console
switch:

preemption latency trace v1.0.7 on 2.6.10-rc2-mm1-RT-V0.7.27-10
-------------------------------------------------------
 latency: 63 us, entries: 22 (22)   |   [VP:0 KP:1 SP:1 HP:1 #CPUS:1]
    -----------------
    | task: IRQ 8/13, uid:0 nice:-5 policy:1 rt_prio:98
    -----------------
 => started at: try_to_wake_up+0x51/0x170 <c010f3a1>
 => ended at:   finish_task_switch+0x51/0xb0 <c010f911>
=======>
    5 80010004 0.000ms (+0.000ms): trace_start_sched_wakeup (try_to_wake_up)
    5 80010003 0.000ms (+0.000ms): (1) ((98))
    5 80010003 0.000ms (+0.000ms): (13) ((5))
    5 80010003 0.000ms (+0.000ms): try_to_wake_up (wake_up_process)
    5 80010003 0.000ms (+0.000ms): (0) ((1))
    5 80010002 0.000ms (+0.000ms): preempt_schedule (try_to_wake_up)
    5 80010002 0.000ms (+0.000ms): wake_up_process (redirect_hardirq)
    5 80010001 0.000ms (+0.000ms): preempt_schedule (__do_IRQ)
    5 80010001 0.000ms (+0.000ms): irq_exit (do_IRQ)
    5 80000002 0.000ms (+0.000ms): do_softirq (irq_exit)
    5 80000002 0.001ms (+0.061ms): __do_softirq (do_softirq)
    5 00000000 0.062ms (+0.000ms): preempt_schedule (_mmx_memcpy)
    5 90000000 0.062ms (+0.000ms): __schedule (preempt_schedule)
    5 90000000 0.062ms (+0.000ms): profile_hit (__schedule)
    5 90000001 0.062ms (+0.000ms): sched_clock (__schedule)
   13 80000002 0.062ms (+0.000ms): __switch_to (__schedule)
   13 80000002 0.062ms (+0.000ms): (5) ((13))
   13 80000002 0.062ms (+0.000ms): (98) ((1))
   13 80000002 0.062ms (+0.000ms): finish_task_switch (__schedule)
   13 80000001 0.062ms (+0.000ms): trace_stop_sched_switched (finish_task_switch)
   13 80000001 0.063ms (+0.003ms): (13) ((1))
   13 80000001 0.066ms (+0.000ms): trace_stop_sched_switched (finish_task_switch)

I sometimes do get large values in /proc/latency_trace, but they seem to be
unrelated to the console switching.

flo
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/