rcutorture: Avoid RCU CPU stall warning and RT throttling
authorPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Sun, 31 Jan 2016 05:32:09 +0000 (21:32 -0800)
committerPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Thu, 31 Mar 2016 20:39:47 +0000 (13:39 -0700)
Running rcuperf can result in RCU CPU stall warnings and RT throttling.
These occur because on of the real-time writer processes does
ftrace_dump() while still running at real-time priority.  This commit
therefore prevents these problems by setting the writer thread back to
SCHED_NORMAL (AKA SCHED_OTHER) before doing ftrace_dump().

In addition, this commit adds a small fixed delay before dumping ftrace
buffer in order to decrease the probability that this dumping will
interfere with other writers' grace periods.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
kernel/rcu/rcuperf.c

index 2786001..4c05728 100644 (file)
@@ -404,11 +404,15 @@ rcu_perf_writer(void *arg)
                        started = true;
                if (!done && i >= MIN_MEAS) {
                        done = true;
+                       sp.sched_priority = 0;
+                       sched_setscheduler_nocheck(current,
+                                                  SCHED_NORMAL, &sp);
                        pr_alert("%s" PERF_FLAG
                                 "rcu_perf_writer %ld has %d measurements\n",
                                 perf_type, me, MIN_MEAS);
                        if (atomic_inc_return(&n_rcu_perf_writer_finished) >=
                            nrealwriters) {
+                               schedule_timeout_interruptible(10);
                                rcu_ftrace_dump(DUMP_ALL);
                                PERFOUT_STRING("Test complete");
                                t_rcu_perf_writer_finished = t;