1Using RCU's CPU Stall Detector 2 3The CONFIG_RCU_CPU_STALL_DETECTOR kernel config parameter enables 4RCU's CPU stall detector, which detects conditions that unduly delay 5RCU grace periods. The stall detector's idea of what constitutes 6"unduly delayed" is controlled by a set of C preprocessor macros: 7 8RCU_SECONDS_TILL_STALL_CHECK 9 10 This macro defines the period of time that RCU will wait from 11 the beginning of a grace period until it issues an RCU CPU 12 stall warning. This time period is normally ten seconds. 13 14RCU_SECONDS_TILL_STALL_RECHECK 15 16 This macro defines the period of time that RCU will wait after 17 issuing a stall warning until it issues another stall warning 18 for the same stall. This time period is normally set to thirty 19 seconds. 20 21RCU_STALL_RAT_DELAY 22 23 The CPU stall detector tries to make the offending CPU print its 24 own warnings, as this often gives better-quality stack traces. 25 However, if the offending CPU does not detect its own stall in 26 the number of jiffies specified by RCU_STALL_RAT_DELAY, then 27 some other CPU will complain. This delay is normally set to 28 two jiffies. 29 30When a CPU detects that it is stalling, it will print a message similar 31to the following: 32 33INFO: rcu_sched_state detected stall on CPU 5 (t=2500 jiffies) 34 35This message indicates that CPU 5 detected that it was causing a stall, 36and that the stall was affecting RCU-sched. This message will normally be 37followed by a stack dump of the offending CPU. On TREE_RCU kernel builds, 38RCU and RCU-sched are implemented by the same underlying mechanism, 39while on TREE_PREEMPT_RCU kernel builds, RCU is instead implemented 40by rcu_preempt_state. 41 42On the other hand, if the offending CPU fails to print out a stall-warning 43message quickly enough, some other CPU will print a message similar to 44the following: 45 46INFO: rcu_bh_state detected stalls on CPUs/tasks: { 3 5 } (detected by 2, 2502 jiffies) 47 48This message indicates that CPU 2 detected that CPUs 3 and 5 were both 49causing stalls, and that the stall was affecting RCU-bh. This message 50will normally be followed by stack dumps for each CPU. Please note that 51TREE_PREEMPT_RCU builds can be stalled by tasks as well as by CPUs, 52and that the tasks will be indicated by PID, for example, "P3421". 53It is even possible for a rcu_preempt_state stall to be caused by both 54CPUs -and- tasks, in which case the offending CPUs and tasks will all 55be called out in the list. 56 57Finally, if the grace period ends just as the stall warning starts 58printing, there will be a spurious stall-warning message: 59 60INFO: rcu_bh_state detected stalls on CPUs/tasks: { } (detected by 4, 2502 jiffies) 61 62This is rare, but does happen from time to time in real life. 63 64So your kernel printed an RCU CPU stall warning. The next question is 65"What caused it?" The following problems can result in RCU CPU stall 66warnings: 67 68o A CPU looping in an RCU read-side critical section. 69 70o A CPU looping with interrupts disabled. This condition can 71 result in RCU-sched and RCU-bh stalls. 72 73o A CPU looping with preemption disabled. This condition can 74 result in RCU-sched stalls and, if ksoftirqd is in use, RCU-bh 75 stalls. 76 77o A CPU looping with bottom halves disabled. This condition can 78 result in RCU-sched and RCU-bh stalls. 79 80o For !CONFIG_PREEMPT kernels, a CPU looping anywhere in the kernel 81 without invoking schedule(). 82 83o A bug in the RCU implementation. 84 85o A hardware failure. This is quite unlikely, but has occurred 86 at least once in real life. A CPU failed in a running system, 87 becoming unresponsive, but not causing an immediate crash. 88 This resulted in a series of RCU CPU stall warnings, eventually 89 leading the realization that the CPU had failed. 90 91The RCU, RCU-sched, and RCU-bh implementations have CPU stall 92warning. SRCU does not have its own CPU stall warnings, but its 93calls to synchronize_sched() will result in RCU-sched detecting 94RCU-sched-related CPU stalls. Please note that RCU only detects 95CPU stalls when there is a grace period in progress. No grace period, 96no CPU stall warnings. 97 98To diagnose the cause of the stall, inspect the stack traces. 99The offending function will usually be near the top of the stack. 100If you have a series of stall warnings from a single extended stall, 101comparing the stack traces can often help determine where the stall 102is occurring, which will usually be in the function nearest the top of 103that portion of the stack which remains the same from trace to trace. 104If you can reliably trigger the stall, ftrace can be quite helpful. 105 106RCU bugs can often be debugged with the help of CONFIG_RCU_TRACE. 107