On Tue, Apr 26, 2011 at 10:12:39AM -0700, Linus Torvalds wrote: > On Tue, Apr 26, 2011 at 9:38 AM, Bruno Prémont > <bonbons@xxxxxxxxxxxxxxxxx> wrote: > > > > Here it comes: > > > > rcu_kthread (when build processes are STOPped): > > [ 836.050003] rcu_kthread R running 7324 6 2 0x00000000 > > [ 836.050003] dd473f28 00000046 5a000240 dd65207c dd407360 dd651d40 0000035c dd473ed8 > > [ 836.050003] c10bf8a2 c14d63d8 dd65207c dd473f28 dd445040 dd445040 dd473eec c10be848 > > [ 836.050003] dd651d40 dd407360 ddfdca00 dd473f14 c10bfde2 00000000 00000001 000007b6 > > [ 836.050003] Call Trace: > > [ 836.050003] [<c10bf8a2>] ? check_object+0x92/0x210 > > [ 836.050003] [<c10be848>] ? init_object+0x38/0x70 > > [ 836.050003] [<c10bfde2>] ? free_debug_processing+0x112/0x1f0 > > [ 836.050003] [<c103d9fd>] ? lock_timer_base+0x2d/0x70 > > [ 836.050003] [<c13c8ec7>] schedule_timeout+0x137/0x280 > > Hmm. > > I'm adding Ingo and Peter to the cc, because this whole "rcu_kthread > is running, but never actually running" is starting to smell like a > scheduler issue. > > Peter/Ingo: RCUTINY seems to be broken for Bruno. During any kind of > heavy workload, at some point it looks like rcu_kthread simply stops > making any progress. It's constantly in runnable state, but it doesn't > actually use any CPU time, and it's not processing the RCU callbacks, > so the RCU memory freeing isn't happening, and slabs just build up > until the machine dies. > > And it really is RCUTINY, because the thing doesn't happen with the > regular tree-RCU. The difference between TINY_RCU and TREE_RCU is that TREE_RCU still uses softirq for the core RCU processing. TINY_RCU switched to a kthread when I implemented RCU priority boosting. There is a similar change in my -rcu tree that makes TREE_RCU use kthreads, and Sedat has been running into a very similar problem with that change in place. Which is why I do not yet push it to the -next tree. > This is without CONFIG_RCU_BOOST_PRIO, so we basically have > > struct sched_param sp; > > rcu_kthread_task = kthread_run(rcu_kthread, NULL, "rcu_kthread"); > sp.sched_priority = RCU_BOOST_PRIO; > sched_setscheduler_nocheck(rcu_kthread_task, SCHED_FIFO, &sp); > > where RCU_BOOST_PRIO is 1 for the non-boost case. Good point! Bruno, Sedat, could you please set CONFIG_RCU_BOOST_PRIO to (say) 50, and see if this still happens? (I bet that you do, but...) > Is that so low that even the idle thread will take priority? It's a UP > config with PREEMPT_VOLUNTARY. So pretty much _all_ the stars are > aligned for odd scheduling behavior. > > Other users of SCHED_FIFO tend to set the priority really high (eg > "MAX_RT_PRIO-1" is clearly the default one - softirq's, watchdog), but > "1" is not unheard of either (touchscreen/ucb1400_ts and > mmc/core/sdio_irq), and there are some other random choises out tere. > > Any ideas? I have found one bug so far in my code, but it only affects TREE_RCU in my -rcu tree, and even then only if HOTPLUG_CPU is enabled. I am testing a fix, but I expect Sedat's tests to still break. I gave Sedat a patch that make rcu_kthread() run at normal (non-realtime) priority, and he did not see the failure. So running non-realtime at least greatly reduces the probability of failure. Thanx, Paul -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
