Re: rcu: INFO: rcu_preempt detected stalls on CPUs/tasks on v3.6

[Avi Kivity <avi@redhat.com>]

On 07/30/2012 12:05 AM, Sasha Levin wrote:
> On 07/29/2012 02:48 PM, Avi Kivity wrote:
> > On 07/27/2012 02:27 PM, Sasha Levin wrote:
> >> On 07/26/2012 01:42 PM, Avi Kivity wrote:
> >>> On 07/24/2012 08:10 PM, Sasha Levin wrote:
> >>>> [  215.026612] NMI backtrace for cpu 1
> >>>> [  215.026612] CPU 1
> >>>> [  215.026612] Pid: 2395, comm: pageattr-test Tainted: G        W    3.5.0-sasha-01644-g824681b #267
> >>>> [  215.026612] RIP: 0010:[<ffffffff8109bd5a>]  [<ffffffff8109bd5a>] native_write_msr_safe+0xa/0x10
> >>>> [  215.026612] RSP: 0018:ffff88001b803f70  EFLAGS: 00000046
> >>>> [  215.026612] RAX: 0000000000000000 RBX: 000000000000ef80 RCX: 000000000000080b
> >>>> [  215.026612] RDX: 0000000000000000 RSI: 0000000000000000 RDI: 000000000000080b
> >>>> [  215.026612] RBP: ffff88001b803f70 R08: 00003ffffffff000 R09: ffff880000000000
> >>>> [  215.026612] R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000
> >>>> [  215.026612] R13: ffff880007a98000 R14: ffff88001fc3a4b8 R15: ffff880000000000
> >>>> [  215.026612] FS:  0000000000000000(0000) GS:ffff88001b800000(0000) knlGS:0000000000000000
> >>>> [  215.026612] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
> >>>> [  215.026612] CR2: 000000000000000c CR3: 0000000019dcf000 CR4: 00000000000406e0
> >>>> [  215.026612] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> >>>> [  215.026612] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
> >>>> [  215.026612] Process pageattr-test (pid: 2395, threadinfo ffff88000ced0000, task ffff88000d773000)
> >>>> [  215.026612] Stack:
> >>>> [  215.026612]  ffff88001b803f88 ffffffff81095d07 000000000000ef80 ffff88001b803fa8
> >>>> [  215.026612]  ffffffff810901c4 ffff88000ced1db0 0000000000007a98 ffff88000ced1d00
> >>>> [  215.026612]  ffffffff8369c4af ffff88000ced1c78 <EOI>  ffff88000ced1d00 ffffffff8369aeb4
> >>>> [  215.026612] Call Trace:
> >>>> [  215.026612]  <IRQ>
> >>>> [  215.026612]  [<ffffffff81095d07>] native_apic_msr_eoi_write+0x27/0x30
> >>>> [  215.026612]  [<ffffffff810901c4>] smp_apic_timer_interrupt+0x34/0xa0
> >>>> [  215.026612]  [<ffffffff8369c4af>] apic_timer_interrupt+0x6f/0x80
> >>>> [  215.026612]  <EOI>
> >>>> [  215.026612]  [<ffffffff8369aeb4>] ? retint_restore_args+0x13/0x13
> >>>> [  215.026612]  [<ffffffff810a2cb2>] ? lookup_address+0x112/0x170
> >>>> [  215.026612]  [<ffffffff810a32ae>] print_split+0x7e/0x200
> >>>> [  215.026612]  [<ffffffff810a4300>] pageattr_test+0x80/0x4a0
> >>>> [  215.026612]  [<ffffffff8114a2f8>] ? trace_hardirqs_on_caller+0x118/0x140
> >>>> [  215.026612]  [<ffffffff83696cbe>] ? schedule_timeout+0x2ce/0x370
> >>>> [  215.026612]  [<ffffffff8369aeb4>] ? retint_restore_args+0x13/0x13
> >>>> [  215.026612]  [<ffffffff810a4720>] ? pageattr_test+0x4a0/0x4a0
> >>>> [  215.026612]  [<ffffffff810a473f>] do_pageattr_test+0x1f/0x50
> >>>> [  215.026612]  [<ffffffff8110607d>] kthread+0xad/0xc0
> >>>> [  215.026612]  [<ffffffff8369cdb4>] kernel_thread_helper+0x4/0x10
> >>>> [  215.026612]  [<ffffffff8369aeb4>] ? retint_restore_args+0x13/0x13
> >>>> [  215.026612]  [<ffffffff81105fd0>] ? kthread_flush_work_fn+0x10/0x10
> >>>> [  215.026612]  [<ffffffff8369cdb0>] ? gs_change+0x13/0x13
> >>>> [  215.026612] Code: 00 55 89 f9 48 89 e5 0f 32 45 31 c0 89 c7 48 89 d0 44 89 06 48 c1 e0 20 89 f9 5d 48 09 c8 c3 66 90 55 89 f0 89 f9 48 89 e5 0f 30 <31> c0 5d c3 66 90 55 89 f9 48 89 e5 0f 33 89 c7 48 89 d0 5d 48
> >>>
> >>> This is immediately after the EOI.  It looks like the timer interrupt is
> >>> stuck somehow.  Can you run kvm_stat and confirm?
> >>
> >> Yup, looks like it. kvm_stats is something like this:
> >>
> >>  kvm_entry                                142104033  939393
> >>  kvm_exit                                 142104004  939390
> >>  kvm_apic                                  84781115  582714
> >>  kvm_msr                                   84682085  582714
> >>  kvm_apic_accept_irq                       65219959  522263
> >>  kvm_inj_virq                              43173709  291385
> >>
> >> With everything else at 0 when it happens.
> > 
> > Please run a trace and verify that the injected vector is indeed the
> > timer interrupt.
>
> Looks like it is.
>
> > Then work backwards to see the last place it is
> > programmed (APIC_TMICT/APIC_TDCR).
>
> This looks like what you're looking for:
>
> kvm_apic: apic_write APIC_TMICT = 0x3e
>
>
> Basically it's spinning on the following 4 lines of trace when it happens:
>
> 	16063.549189: kvm_apic: apic_write APIC_TMICT = 0x3e
> 	16063.549190: kvm_msr: msr_write 838 = 0x3e
> 	16063.549190: kvm_exit: reason EXTERNAL_INTERRUPT rip 0xffffffff8115de02 info 0 0
> 	16063.549190: kvm_entry: vcpu 2
>

What, no kvm_inj_virq?

Are you running with pveoi?  Try disabling it.


-- 
I have a truly marvellous patch that fixes the bug which this
signature is too narrow to contain.