On Thu, May 07, 2015 at 12:36:11PM -0700, Paul E. McKenney wrote: > > INFO: rcu_sched detected stalls on CPUs/tasks: > > (detected by 0, t=6502 jiffies, g=75215, c=75214, q=0) > > All QSes seen, last rcu_sched kthread activity 6502 (4295121467-4295114965), jiffies_till_next_fqs=1, root ->qsmask 0x0 > > trinity-c21 R running task 12768 20468 30399 0x0008000c > > 00000000000031e0 00000000f34c833f ffff88043d803da8 ffffffffb30abd15 > > ffffffffb30abc32 ffffffffb3c3b500 0000000000001966 ffff88043d817580 > > ffffffffb3c3b080 0000000000000000 ffff88043d803e28 ffffffffb30ef556 > > Call Trace: > > <IRQ> [<ffffffffb30abd15>] sched_show_task+0x165/0x280 > > [<ffffffffb30abc32>] ? sched_show_task+0x82/0x280 > > [<ffffffffb30ef556>] rcu_check_callbacks+0xb26/0xb30 > > [<ffffffffb30f5389>] update_process_times+0x39/0x60 > > [<ffffffffb3107ef5>] tick_sched_handle.isra.20+0x25/0x60 > > [<ffffffffb3108191>] tick_sched_timer+0x41/0x80 > > [<ffffffffb30f643a>] __run_hrtimer+0xba/0x5f0 > > [<ffffffffb3108150>] ? tick_sched_do_timer+0x40/0x40 > > [<ffffffffb30f6e93>] hrtimer_interrupt+0x103/0x230 > > [<ffffffffb30b10e2>] ? vtime_common_account_irq_enter+0x32/0x50 > > [<ffffffffb3034fec>] local_apic_timer_interrupt+0x3c/0x70 > > [<ffffffffb372bf51>] smp_apic_timer_interrupt+0x41/0x60 > > [<ffffffffb372a980>] apic_timer_interrupt+0x70/0x80 > > <EOI> [<ffffffffb30cde67>] ? lock_acquire+0xd7/0x270 > > [<ffffffffb329e466>] ? ext4_es_find_delayed_extent_range+0x76/0x480 > > [<ffffffffb372949b>] _raw_read_lock+0x3b/0x50 > > [<ffffffffb329e466>] ? ext4_es_find_delayed_extent_range+0x76/0x480 > > [<ffffffffb329e466>] ext4_es_find_delayed_extent_range+0x76/0x480 > > [<ffffffffb324f291>] ext4_llseek+0x291/0x410 > > [<ffffffffb30cb0fd>] ? trace_hardirqs_on+0xd/0x10 > > [<ffffffffb31cd339>] SyS_lseek+0x99/0xc0 > > [<ffffffffb3729b97>] system_call_fastpath+0x12/0x6f > > rcu_sched kthread starved for 6502 jiffies! > > If you don't let RCU's grace-period kthreads run, they aren't going to > be able to help you much. Could you please try building with > CONFIG_RCU_KTHREAD_PRIO=5 (say) and see if that helps? Nope. Though the message is slightly different. INFO: rcu_sched self-detected stall on CPU 1: (6499 ticks this GP) idle=5ed/140000000000001/0 softirq=6071/6071 fqs=6495 (t=6500 jiffies g=17107 c=17106 q=0) Task dump for CPU 1: trinity-c6 R running task 12576 26108 548 0x0008000c 0000000000003120 00000000aa7d3446 ffff88043d843d58 ffffffff8e0acb55 ffffffff8e0aca72 0000000000000001 0000000000000000 0000000000000001 ffffffff8ec3b0c0 0000000000000082 ffff88043d843d78 ffffffff8e0b090d Call Trace: <IRQ> [<ffffffff8e0acb55>] sched_show_task+0x165/0x280 [<ffffffff8e0aca72>] ? sched_show_task+0x82/0x280 [<ffffffff8e0b090d>] dump_cpu_task+0x3d/0x50 [<ffffffff8e0e4971>] rcu_dump_cpu_stacks+0x91/0xd0 [<ffffffff8e0f009c>] rcu_check_callbacks+0x7ec/0xb30 [<ffffffff8e10908e>] ? tick_do_update_jiffies64+0xde/0x140 [<ffffffff8e0f6299>] update_process_times+0x39/0x60 [<ffffffff8e108ed5>] tick_sched_handle.isra.20+0x25/0x60 [<ffffffff8e109171>] tick_sched_timer+0x41/0x80 [<ffffffff8e0f735a>] __run_hrtimer+0xba/0x5f0 [<ffffffff8e109130>] ? tick_sched_do_timer+0x40/0x40 [<ffffffff8e0f7db3>] hrtimer_interrupt+0x103/0x230 [<ffffffff8e0b1f42>] ? vtime_common_account_irq_enter+0x32/0x50 [<ffffffff8e0355dc>] local_apic_timer_interrupt+0x3c/0x70 [<ffffffff8e733251>] smp_apic_timer_interrupt+0x41/0x60 [<ffffffff8e731c80>] apic_timer_interrupt+0x70/0x80 <EOI> [<ffffffff8e0cecc7>] ? lock_acquire+0xd7/0x270 [<ffffffff8e2a206e>] ? ext4_es_lookup_extent+0x5e/0x470 [<ffffffff8e73079b>] _raw_read_lock+0x3b/0x50 [<ffffffff8e2a206e>] ? ext4_es_lookup_extent+0x5e/0x470 [<ffffffff8e2a206e>] ext4_es_lookup_extent+0x5e/0x470 [<ffffffff8e25927b>] ext4_map_blocks+0x5b/0x590 [<ffffffff8e2a1a4c>] ? ext4_es_find_delayed_extent_range+0x42c/0x480 [<ffffffff8e252301>] ext4_llseek+0x271/0x410 [<ffffffff8e0cbf5d>] ? trace_hardirqs_on+0xd/0x10 [<ffffffff8e1cf479>] SyS_lseek+0x99/0xc0 [<ffffffff8e730e97>] system_call_fastpath+0x12/0x6f -- To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
